{ "best_metric": 0.6866666674613953, "best_model_checkpoint": "/scratch-ssd/lucelo/max-ent-llms/data/Qwen-2.5-7B-RL-LACPO-NoBaselineAlpha1/checkpoint-19", "epoch": 0.9873417721518988, "eval_steps": 1, "global_step": 39, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "accuracy_reward": 0.640625, "accuracy_reward/correct": 0.9999999403953552, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.23142997920513153, "adam_stats/lr_effective_max": 7.905724487500265e-06, "adam_stats/lr_effective_mean": -3.657434077819488e-11, "adam_stats/lr_effective_min": -7.905713573563844e-06, "adam_stats/m_t_max": 0.011035159230232239, "adam_stats/m_t_mean": -7.945365199102383e-11, "adam_stats/m_t_min": -0.007421876769512892, "adam_stats/v_t_max": 1.2177310964034405e-05, "adam_stats/v_t_mean": 2.893036287962558e-13, "adam_stats/v_t_min": 0.0, "advantages": 0.640625, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.23142997920513153, "all_logprobs": -0.16401661932468414, "all_logprobs/max": 0.0, "all_logprobs/median": -2.276897430419922e-05, "all_logprobs/min": -10.125, "all_logprobs/p25": -0.0179443359375, "all_logprobs/p75": -4.76837158203125e-07, "all_logprobs/var": 0.2920554578304291, "answer_logprob_reward": 2.7831039428710938, "answer_logprob_reward/correct": 3.034297466278076, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.10153310000896454, "answer_logprob_reward/incorrect": 2.335324287414551, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 2.4849066734313965, "answer_logprob_reward/incorrect/min": 1.2321436405181885, "answer_logprob_reward/incorrect/p25": 1.3862943649291992, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.7438019514083862, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.2321436405181885, "answer_logprob_reward/p25": 2.4849066734313965, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.44273096323013306, "clip_ratio": 0.0, "completion_length": 583.7760620117188, "completion_length/correct": 521.78857421875, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 452.0, "completion_length/correct/min": 12.0, "completion_length/correct/p25": 338.5, "completion_length/correct/p75": 686.0, "completion_length/correct/var": 68040.7734375, "completion_length/incorrect": 694.275390625, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 808.0, "completion_length/incorrect/min": 24.0, "completion_length/incorrect/p25": 340.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 129540.015625, "completion_length/max": 1024.0, "completion_length/median": 525.0, "completion_length/min": 12.0, "completion_length/p25": 339.25, "completion_length/p75": 873.5, "completion_length/var": 96464.9765625, "epoch": 0.02531645569620253, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 112085.0, "grad_norm": 0.7914038300514221, "kl": 0.0, "learning_rate": 2.5e-06, "len_reward": 0.01521260291337967, "len_reward/correct": 0.11566465348005295, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.14369310438632965, "len_reward/correct/min": -0.5, "len_reward/correct/p25": -0.023282986134290695, "len_reward/correct/p75": 0.2884446978569031, "len_reward/correct/var": 0.05871717259287834, "len_reward/incorrect": -0.16385412216186523, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.12603966891765594, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.29398593306541443, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.029935725033283234, "len_reward/max": 0.5, "len_reward/median": 0.0, "len_reward/min": -0.5, "len_reward/p25": -0.16415677964687347, "len_reward/p75": 0.21801316738128662, "len_reward/var": 0.06624475121498108, "loss": -0.6406, "mean_logprobs": -0.197265625, "mean_logprobs/var": 0.0306396484375, "missing_response_penalty": -0.1822916716337204, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.5072463750839233, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": -1.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.25362321734428406, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.14984185993671417, "non_smoothed_logprobs/min": -16.75, "num_completions/total": 192, "policy_loss": -0.640625, "policy_loss/max": 0.0, "policy_loss/median": -1.0, "policy_loss/min": -1.0, "policy_loss/p25": -1.0, "policy_loss/p75": 0.0, "policy_loss/var": 0.23142997920513153, "pred_value_log_probs": -0.6782176494598389, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.7578125, "pred_value_log_probs/min": -1.4609375, "pred_value_log_probs/p25": -0.7578125, "pred_value_log_probs/p75": -0.7578125, "pred_value_log_probs/var": 0.046743351966142654, "reward": 0.640625, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.23142997920513153, "rewards/accuracy_reward": 0.640625, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.23142997920513153, "rewards/answer_logprob_reward": 2.7831039428710938, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.2321436405181885, "rewards/answer_logprob_reward/p25": 2.4849066734313965, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.44273096323013306, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.01521260291337967, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.0, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.16415677964687347, "rewards/len_reward/p75": 0.21801316738128662, "rewards/len_reward/var": 0.06624475121498108, "rewards/missing_response_penalty": -0.1822916716337204, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.14984185993671417, "smoothed_logprobs/min": -1.0, "step": 1, "value_loss": 2.7890625, "values": -0.7917071580886841, "values/max": 0.73681640625, "values/median": -0.6884765625, "values/min": -2.0000112056732178, "values/p25": -0.6884765625, "values/p75": -0.6884765625, "values/var": 0.12838509678840637 }, { "adam_stats/lr_effective_max": 7.905724487500265e-06, "adam_stats/lr_effective_mean": -3.657434077819488e-11, "adam_stats/lr_effective_min": -7.905713573563844e-06, "adam_stats/m_t_max": 0.011035159230232239, "adam_stats/m_t_mean": -7.945365199102383e-11, "adam_stats/m_t_min": -0.007421876769512892, "adam_stats/v_t_max": 1.2177310964034405e-05, "adam_stats/v_t_mean": 2.893036287962558e-13, "adam_stats/v_t_min": 0.0, "epoch": 0.02531645569620253, "eval_accuracy_reward": 0.4399999976158142, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 0.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.24722407758235931, "eval_advantages": 0.4399999976158142, "eval_advantages/max": 1.0, "eval_advantages/median": 0.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.24722407758235931, "eval_all_logprobs": -0.17103739082813263, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": -4.601478576660156e-05, "eval_all_logprobs/min": -12.125, "eval_all_logprobs/p25": -0.0211181640625, "eval_all_logprobs/p75": -7.152557373046875e-07, "eval_all_logprobs/var": 0.31482362747192383, "eval_answer_logprob_reward": 2.664313316345215, "eval_answer_logprob_reward/correct": 3.161113739013672, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.525728702545166, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.03698143735527992, "eval_answer_logprob_reward/incorrect": 2.273969888687134, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 2.525728702545166, "eval_answer_logprob_reward/incorrect/min": 0.7339691519737244, "eval_answer_logprob_reward/incorrect/p25": 1.272965669631958, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.9783580303192139, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 0.7339691519737244, "eval_answer_logprob_reward/p25": 1.8325815200805664, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.7572145462036133, "eval_completion_length": 576.9166870117188, "eval_completion_length/correct": 561.4166870117188, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 522.0, "eval_completion_length/correct/min": 66.0, "eval_completion_length/correct/p25": 344.0, "eval_completion_length/correct/p75": 720.0, "eval_completion_length/correct/var": 73380.4921875, "eval_completion_length/incorrect": 589.0952758789062, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 556.0, "eval_completion_length/incorrect/min": 8.0, "eval_completion_length/incorrect/p25": 212.25, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 139438.125, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 531.0, "eval_completion_length/min": 8.0, "eval_completion_length/p25": 301.75, "eval_completion_length/p75": 974.5, "eval_completion_length/var": 110219.5390625, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 173075.0, "eval_len_reward": -0.042816244065761566, "eval_len_reward/correct": 0.06277946382761002, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.09141184389591217, "eval_len_reward/correct/min": -0.4624646008014679, "eval_len_reward/correct/p25": -0.08297038823366165, "eval_len_reward/correct/p75": 0.23608985543251038, "eval_len_reward/correct/var": 0.054971843957901, "eval_len_reward/incorrect": -0.12578429281711578, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": 0.0, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.24150700867176056, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03059995174407959, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.18369412422180176, "eval_len_reward/p75": 0.03296111524105072, "eval_len_reward/var": 0.04996596649289131, "eval_mean_logprobs": -0.22265625, "eval_mean_logprobs/var": 0.0419921875, "eval_missing_response_penalty": -0.26333335041999817, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.4702380895614624, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.25060591101646423, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": -1.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.19463768601417542, "eval_num_completions/total": 300, "eval_pred_value_log_probs": -0.7142305970191956, "eval_pred_value_log_probs/max": -2.2649765014648438e-06, "eval_pred_value_log_probs/median": -0.7578125, "eval_pred_value_log_probs/min": -1.234375, "eval_pred_value_log_probs/p25": -0.7578125, "eval_pred_value_log_probs/p75": -0.7578125, "eval_pred_value_log_probs/var": 0.022694004699587822, "eval_reward": 0.4399999976158142, "eval_reward/max": 1.0, "eval_reward/median": 0.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.24722407758235931, "eval_rewards/accuracy_reward": 0.4399999976158142, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 0.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.24722407758235931, "eval_rewards/answer_logprob_reward": 2.664313316345215, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 0.7339691519737244, "eval_rewards/answer_logprob_reward/p25": 1.8325815200805664, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.7572145462036133, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": -0.042816244065761566, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.18369412422180176, "eval_rewards/len_reward/p75": 0.03296111524105072, "eval_rewards/len_reward/var": 0.04996596649289131, "eval_rewards/missing_response_penalty": -0.26333335041999817, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": -1.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.19463768601417542, "eval_runtime": 171.0734, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.7239292860031128, "eval_values/max": 1.9339599609375, "eval_values/median": -0.6884765625, "eval_values/min": -1.9770984649658203, "eval_values/p25": -0.6884765625, "eval_values/p75": -0.6884765625, "eval_values/var": 0.09213504940271378, "step": 1 }, { "accuracy_reward": 0.6197916865348816, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.2368837296962738, "adam_stats/lr_effective_max": 2.1276326151564717e-05, "adam_stats/lr_effective_mean": 7.12119209086981e-12, "adam_stats/lr_effective_min": -2.1276304323691875e-05, "adam_stats/m_t_max": 0.0102612329646945, "adam_stats/m_t_mean": -4.9798998258410165e-11, "adam_stats/m_t_min": -0.007576906122267246, "adam_stats/v_t_max": 1.2175996744190343e-05, "adam_stats/v_t_mean": 3.826528937461715e-13, "adam_stats/v_t_min": 0.0, "advantages": 0.6197916865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.2368837296962738, "all_logprobs": -0.15524685382843018, "all_logprobs/max": 0.0, "all_logprobs/median": -2.2172927856445312e-05, "all_logprobs/min": -9.25, "all_logprobs/p25": -0.0135498046875, "all_logprobs/p75": -4.76837158203125e-07, "all_logprobs/var": 0.2705397605895996, "answer_logprob_reward": 2.6923842430114746, "answer_logprob_reward/correct": 3.0770535469055176, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.07541394978761673, "answer_logprob_reward/incorrect": 2.0653202533721924, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 1.7917594909667969, "answer_logprob_reward/incorrect/min": 1.0986123085021973, "answer_logprob_reward/incorrect/p25": 1.0986123085021973, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.8078325986862183, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.0986123085021973, "answer_logprob_reward/p25": 2.079441547393799, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.593589186668396, "clip_ratio": 0.0, "completion_length": 641.9896240234375, "completion_length/correct": 566.2269287109375, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 513.0, "completion_length/correct/min": 117.0, "completion_length/correct/p25": 372.5, "completion_length/correct/p75": 710.5, "completion_length/correct/var": 61870.33203125, "completion_length/incorrect": 765.4931640625, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 1024.0, "completion_length/incorrect/min": 17.0, "completion_length/incorrect/p25": 471.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 116871.15625, "completion_length/max": 1024.0, "completion_length/median": 570.0, "completion_length/min": 17.0, "completion_length/p25": 390.25, "completion_length/p75": 1024.0, "completion_length/var": 91685.640625, "epoch": 0.05063291139240506, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 235347.0, "grad_norm": 0.48119205236434937, "kl": 0.0, "learning_rate": 5e-06, "len_reward": -0.0022739358246326447, "len_reward/correct": 0.09396910667419434, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.12620492279529572, "len_reward/correct/min": -0.5, "len_reward/correct/p25": -0.021420922130346298, "len_reward/correct/p75": 0.26176321506500244, "len_reward/correct/var": 0.05913642793893814, "len_reward/incorrect": -0.15916328132152557, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.10764012485742569, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.28188976645469666, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03061472624540329, "len_reward/max": 0.5, "len_reward/median": 0.0, "len_reward/min": -0.5, "len_reward/p25": -0.1725209653377533, "len_reward/p75": 0.17705444991588593, "len_reward/var": 0.0632537305355072, "loss": -0.6198, "mean_logprobs": -0.16796875, "mean_logprobs/var": 0.01263427734375, "missing_response_penalty": -0.2291666716337204, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.602739691734314, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": -1.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.24277016520500183, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.17757420241832733, "non_smoothed_logprobs/min": -16.875, "num_completions/total": 384, "policy_loss": -0.6197916865348816, "policy_loss/max": 0.0, "policy_loss/median": -1.0, "policy_loss/min": -1.0, "policy_loss/p25": -1.0, "policy_loss/p75": 0.0, "policy_loss/var": 0.2368837296962738, "pred_value_log_probs": -0.705960750579834, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.7578125, "pred_value_log_probs/min": -1.46875, "pred_value_log_probs/p25": -0.7578125, "pred_value_log_probs/p75": -0.7578125, "pred_value_log_probs/var": 0.0509880967438221, "reward": 0.6197916865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.2368837296962738, "rewards/accuracy_reward": 0.6197916865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.2368837296962738, "rewards/answer_logprob_reward": 2.6923842430114746, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.0986123085021973, "rewards/answer_logprob_reward/p25": 2.079441547393799, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.593589186668396, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": -0.0022739358246326447, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.0, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.1725209653377533, "rewards/len_reward/p75": 0.17705444991588593, "rewards/len_reward/var": 0.0632537305355072, "rewards/missing_response_penalty": -0.2291666716337204, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.17757420241832733, "smoothed_logprobs/min": -1.0, "step": 2, "value_loss": 2.7421875, "values": -0.7263092994689941, "values/max": 0.6708984375, "values/median": -0.6884765625, "values/min": -2.0, "values/p25": -0.6884765625, "values/p75": -0.6884765625, "values/var": 0.1137716993689537 }, { "adam_stats/lr_effective_max": 2.1276326151564717e-05, "adam_stats/lr_effective_mean": 7.12119209086981e-12, "adam_stats/lr_effective_min": -2.1276304323691875e-05, "adam_stats/m_t_max": 0.0102612329646945, "adam_stats/m_t_mean": -4.9798998258410165e-11, "adam_stats/m_t_min": -0.007576906122267246, "adam_stats/v_t_max": 1.2175996744190343e-05, "adam_stats/v_t_mean": 3.826528937461715e-13, "adam_stats/v_t_min": 0.0, "epoch": 0.05063291139240506, "eval_accuracy_reward": 0.5266667008399963, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.25012266635894775, "eval_advantages": 0.5266667008399963, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.25012266635894775, "eval_all_logprobs": -0.1727336347103119, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": -4.0531158447265625e-05, "eval_all_logprobs/min": -12.5, "eval_all_logprobs/p25": -0.0234375, "eval_all_logprobs/p75": -7.152557373046875e-07, "eval_all_logprobs/var": 0.31189486384391785, "eval_answer_logprob_reward": 2.7104697227478027, "eval_answer_logprob_reward/correct": 3.166231870651245, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.525728702545166, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.03393346816301346, "eval_answer_logprob_reward/incorrect": 2.2033538818359375, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 1.8325815200805664, "eval_answer_logprob_reward/incorrect/min": 0.9162907600402832, "eval_answer_logprob_reward/incorrect/p25": 1.4271163940429688, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.9209528565406799, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 0.9162907600402832, "eval_answer_logprob_reward/p25": 2.4243624210357666, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.6840105056762695, "eval_completion_length": 598.6400146484375, "eval_completion_length/correct": 550.6582641601562, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 492.0, "eval_completion_length/correct/min": 103.0, "eval_completion_length/correct/p25": 349.0, "eval_completion_length/correct/p75": 736.75, "eval_completion_length/correct/var": 68220.328125, "eval_completion_length/incorrect": 652.0281372070312, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 661.0, "eval_completion_length/incorrect/min": 14.0, "eval_completion_length/incorrect/p25": 342.5, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 132255.953125, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 542.0, "eval_completion_length/min": 14.0, "eval_completion_length/p25": 347.0, "eval_completion_length/p75": 990.25, "eval_completion_length/var": 100759.7890625, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 352667.0, "eval_len_reward": -0.0232350155711174, "eval_len_reward/correct": 0.08221171796321869, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.09401709586381912, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": -0.04879438504576683, "eval_len_reward/correct/p75": 0.2662348449230194, "eval_len_reward/correct/var": 0.0583944208920002, "eval_len_reward/incorrect": -0.14056305587291718, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.04492584988474846, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.2636667490005493, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.028694583103060722, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.19380633533000946, "eval_len_reward/p75": 0.10665474832057953, "eval_len_reward/var": 0.05660673975944519, "eval_mean_logprobs": -0.208984375, "eval_mean_logprobs/var": 0.03466796875, "eval_missing_response_penalty": -0.25, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.5281689763069153, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": -1.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.2509739398956299, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": -0.25, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.1881270706653595, "eval_num_completions/total": 600, "eval_pred_value_log_probs": -0.7176648378372192, "eval_pred_value_log_probs/max": -2.2649765014648438e-06, "eval_pred_value_log_probs/median": -0.76171875, "eval_pred_value_log_probs/min": -1.171875, "eval_pred_value_log_probs/p25": -0.76171875, "eval_pred_value_log_probs/p75": -0.76171875, "eval_pred_value_log_probs/var": 0.02239096909761429, "eval_reward": 0.5266667008399963, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.25012266635894775, "eval_rewards/accuracy_reward": 0.5266667008399963, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.25012266635894775, "eval_rewards/answer_logprob_reward": 2.7104697227478027, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 0.9162907600402832, "eval_rewards/answer_logprob_reward/p25": 2.4243624210357666, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.6840105056762695, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": -0.0232350155711174, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.19380633533000946, "eval_rewards/len_reward/p75": 0.10665474832057953, "eval_rewards/len_reward/var": 0.05660673975944519, "eval_rewards/missing_response_penalty": -0.25, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": -0.25, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.1881270706653595, "eval_runtime": 171.3262, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.7187559008598328, "eval_values/max": 1.9427032470703125, "eval_values/median": -0.68359375, "eval_values/min": -1.9771080017089844, "eval_values/p25": -0.68359375, "eval_values/p75": -0.68359375, "eval_values/var": 0.09229511767625809, "step": 2 }, { "accuracy_reward": 0.5885416865348816, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.24342823028564453, "adam_stats/lr_effective_max": 3.7253634218359366e-05, "adam_stats/lr_effective_mean": -6.315423184410918e-11, "adam_stats/lr_effective_min": -3.725680653587915e-05, "adam_stats/m_t_max": 0.012103762477636337, "adam_stats/m_t_mean": -6.582498435214745e-11, "adam_stats/m_t_min": -0.009285036474466324, "adam_stats/v_t_max": 1.2986727597308345e-05, "adam_stats/v_t_mean": 4.846660182564189e-13, "adam_stats/v_t_min": 0.0, "advantages": 0.5885416865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.24342823028564453, "all_logprobs": -0.1633242964744568, "all_logprobs/max": 0.0, "all_logprobs/median": -2.7060508728027344e-05, "all_logprobs/min": -10.5625, "all_logprobs/p25": -0.0196533203125, "all_logprobs/p75": -4.76837158203125e-07, "all_logprobs/var": 0.2853495478630066, "answer_logprob_reward": 2.680971622467041, "answer_logprob_reward/correct": 3.067641019821167, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.4849066734313965, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.06491584330797195, "answer_logprob_reward/incorrect": 2.1278867721557617, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 2.079441547393799, "answer_logprob_reward/incorrect/min": 0.9808292388916016, "answer_logprob_reward/incorrect/p25": 1.0986123085021973, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.8952284455299377, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 0.9808292388916016, "answer_logprob_reward/p25": 2.4849066734313965, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.618637204170227, "clip_ratio": 0.0, "completion_length": 593.7916870117188, "completion_length/correct": 532.1327514648438, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 474.0, "completion_length/correct/min": 75.0, "completion_length/correct/p25": 332.0, "completion_length/correct/p75": 682.0, "completion_length/correct/var": 69216.390625, "completion_length/incorrect": 681.9873657226562, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 1006.0, "completion_length/incorrect/min": 11.0, "completion_length/incorrect/p25": 315.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 152385.125, "completion_length/max": 1024.0, "completion_length/median": 521.0, "completion_length/min": 11.0, "completion_length/p25": 324.75, "completion_length/p75": 1024.0, "completion_length/var": 108284.703125, "epoch": 0.0759493670886076, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 349355.0, "grad_norm": 0.49451473355293274, "kl": -2.2590160369873047e-05, "learning_rate": 7.500000000000001e-06, "len_reward": -0.018583837896585464, "len_reward/correct": 0.08518543094396591, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.13620124757289886, "len_reward/correct/min": -0.5, "len_reward/correct/p25": -0.09035760164260864, "len_reward/correct/p75": 0.24853037297725677, "len_reward/correct/var": 0.05470568314194679, "len_reward/incorrect": -0.1670132875442505, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.11991434544324875, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.2974925935268402, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03210582956671715, "len_reward/max": 0.5, "len_reward/median": 0.0, "len_reward/min": -0.5, "len_reward/p25": -0.17814569175243378, "len_reward/p75": 0.17193058133125305, "len_reward/var": 0.0606730654835701, "loss": -0.5885, "mean_logprobs": -0.1953125, "mean_logprobs/var": 0.0274658203125, "missing_response_penalty": -0.2291666716337204, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.5569620132446289, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": -1.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.2499188333749771, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.17757417261600494, "non_smoothed_logprobs/min": -16.6875, "num_completions/total": 576, "policy_loss": -0.5885425806045532, "policy_loss/max": 4.304658432374708e-06, "policy_loss/median": -1.0, "policy_loss/min": -1.0000050067901611, "policy_loss/p25": -1.0000008344650269, "policy_loss/p75": -1.2218952178955078e-06, "policy_loss/var": 0.24342793226242065, "pred_value_log_probs": -0.7097456455230713, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.76171875, "pred_value_log_probs/min": -1.1796875, "pred_value_log_probs/p25": -0.76171875, "pred_value_log_probs/p75": -0.76171875, "pred_value_log_probs/var": 0.03592011332511902, "reward": 0.5885416865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.24342823028564453, "rewards/accuracy_reward": 0.5885416865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.24342823028564453, "rewards/answer_logprob_reward": 2.680971622467041, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 0.9808292388916016, "rewards/answer_logprob_reward/p25": 2.4849066734313965, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.618637204170227, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": -0.018583837896585464, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.0, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.17814569175243378, "rewards/len_reward/p75": 0.17193058133125305, "rewards/len_reward/var": 0.0606730654835701, "rewards/missing_response_penalty": -0.2291666716337204, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.17757417261600494, "smoothed_logprobs/min": -1.0, "step": 3, "value_loss": 2.5546875, "values": -0.6995077133178711, "values/max": 2.000021457672119, "values/median": -0.68359375, "values/min": -2.0, "values/p25": -0.68359375, "values/p75": -0.68359375, "values/var": 0.1517854928970337 }, { "adam_stats/lr_effective_max": 3.7253634218359366e-05, "adam_stats/lr_effective_mean": -6.315423184410918e-11, "adam_stats/lr_effective_min": -3.725680653587915e-05, "adam_stats/m_t_max": 0.012103762477636337, "adam_stats/m_t_mean": -6.582498435214745e-11, "adam_stats/m_t_min": -0.009285036474466324, "adam_stats/v_t_max": 1.2986727597308345e-05, "adam_stats/v_t_mean": 4.846660182564189e-13, "adam_stats/v_t_min": 0.0, "epoch": 0.0759493670886076, "eval_accuracy_reward": 0.5133333206176758, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.2506577670574188, "eval_advantages": 0.5133333206176758, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.2506577670574188, "eval_all_logprobs": -0.15390874445438385, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": -2.1457672119140625e-05, "eval_all_logprobs/min": -13.25, "eval_all_logprobs/p25": -0.01416015625, "eval_all_logprobs/p75": -4.76837158203125e-07, "eval_all_logprobs/var": 0.2686237096786499, "eval_answer_logprob_reward": 2.7141823768615723, "eval_answer_logprob_reward/correct": 3.101086139678955, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.07403698563575745, "eval_answer_logprob_reward/incorrect": 2.3060784339904785, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 2.525728702545166, "eval_answer_logprob_reward/incorrect/min": 0.9162907600402832, "eval_answer_logprob_reward/incorrect/p25": 1.4271163940429688, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.8032733798027039, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 0.9162907600402832, "eval_answer_logprob_reward/p25": 2.525728702545166, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.5858576893806458, "eval_completion_length": 588.5166625976562, "eval_completion_length/correct": 532.7012939453125, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 453.0, "eval_completion_length/correct/min": 20.0, "eval_completion_length/correct/p25": 345.5, "eval_completion_length/correct/p75": 690.25, "eval_completion_length/correct/var": 65450.05859375, "eval_completion_length/incorrect": 647.390380859375, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 698.0, "eval_completion_length/incorrect/min": 16.0, "eval_completion_length/incorrect/p25": 330.5, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 135004.890625, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 509.0, "eval_completion_length/min": 16.0, "eval_completion_length/p25": 336.75, "eval_completion_length/p75": 960.75, "eval_completion_length/var": 102258.8125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 529222.0, "eval_len_reward": -0.02000892348587513, "eval_len_reward/correct": 0.09573789685964584, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.14518091082572937, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": -0.049733877182006836, "eval_len_reward/correct/p75": 0.2577993869781494, "eval_len_reward/correct/var": 0.0548153854906559, "eval_len_reward/incorrect": -0.14209803938865662, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.040425531566143036, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.29070523381233215, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03147710859775543, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.17550882697105408, "eval_len_reward/p75": 0.1517728716135025, "eval_len_reward/var": 0.05749284476041794, "eval_mean_logprobs": -0.185546875, "eval_mean_logprobs/var": 0.0242919921875, "eval_missing_response_penalty": -0.23000000417232513, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.4726027250289917, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.2509683668613434, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.17769230902194977, "eval_num_completions/total": 900, "eval_pred_value_log_probs": -0.7149859666824341, "eval_pred_value_log_probs/max": -2.1457672119140625e-06, "eval_pred_value_log_probs/median": -0.7578125, "eval_pred_value_log_probs/min": -1.1953125, "eval_pred_value_log_probs/p25": -0.7578125, "eval_pred_value_log_probs/p75": -0.7578125, "eval_pred_value_log_probs/var": 0.022274916991591454, "eval_reward": 0.5133333206176758, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.2506577670574188, "eval_rewards/accuracy_reward": 0.5133333206176758, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.2506577670574188, "eval_rewards/answer_logprob_reward": 2.7141823768615723, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 0.9162907600402832, "eval_rewards/answer_logprob_reward/p25": 2.525728702545166, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.5858576893806458, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": -0.02000892348587513, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.17550882697105408, "eval_rewards/len_reward/p75": 0.1517728716135025, "eval_rewards/len_reward/var": 0.05749284476041794, "eval_rewards/missing_response_penalty": -0.23000000417232513, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.17769230902194977, "eval_runtime": 174.2124, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.720794141292572, "eval_values/max": 1.951904296875, "eval_values/median": -0.6875, "eval_values/min": -1.9825439453125, "eval_values/p25": -0.6875, "eval_values/p75": -0.6875, "eval_values/var": 0.09353592246770859, "step": 3 }, { "accuracy_reward": 0.6354166865348816, "accuracy_reward/correct": 0.9999999403953552, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.23287522792816162, "adam_stats/lr_effective_max": 5.4755189921706915e-05, "adam_stats/lr_effective_mean": -2.1790967641432957e-10, "adam_stats/lr_effective_min": -5.476594014908187e-05, "adam_stats/m_t_max": 0.018315263092517853, "adam_stats/m_t_mean": -1.1892935247725944e-10, "adam_stats/m_t_min": -0.014240323565900326, "adam_stats/v_t_max": 1.8482092855265364e-05, "adam_stats/v_t_mean": 6.644819785552036e-13, "adam_stats/v_t_min": 0.0, "advantages": 0.6354166865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.23287522792816162, "all_logprobs": -0.1348327398300171, "all_logprobs/max": 0.0, "all_logprobs/median": -8.106231689453125e-06, "all_logprobs/min": -9.8125, "all_logprobs/p25": -0.0068359375, "all_logprobs/p75": -1.1920928955078125e-07, "all_logprobs/var": 0.22426800429821014, "answer_logprob_reward": 2.72711181640625, "answer_logprob_reward/correct": 3.032691717147827, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 1.7917594909667969, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.11573203653097153, "answer_logprob_reward/incorrect": 2.194528341293335, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 1.5686159133911133, "answer_logprob_reward/incorrect/min": 1.2321436405181885, "answer_logprob_reward/incorrect/p25": 1.3862943649291992, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.7170899510383606, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.2321436405181885, "answer_logprob_reward/p25": 2.4849066734313965, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.49596965312957764, "clip_ratio": 0.0, "completion_length": 606.4427490234375, "completion_length/correct": 553.5901489257812, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 483.0, "completion_length/correct/min": 100.0, "completion_length/correct/p25": 344.5, "completion_length/correct/p75": 748.75, "completion_length/correct/var": 73280.8515625, "completion_length/incorrect": 698.55712890625, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 883.0, "completion_length/incorrect/min": 11.0, "completion_length/incorrect/p25": 339.75, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 143602.21875, "completion_length/max": 1024.0, "completion_length/median": 563.0, "completion_length/min": 11.0, "completion_length/p25": 342.0, "completion_length/p75": 1024.0, "completion_length/var": 103195.2109375, "epoch": 0.10126582278481013, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 465792.0, "grad_norm": 0.6282712817192078, "kl": 3.647804260253906e-05, "learning_rate": 1e-05, "len_reward": 0.0003473448450677097, "len_reward/correct": 0.08404155820608139, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.15213315188884735, "len_reward/correct/min": -0.5, "len_reward/correct/p25": -0.06679219007492065, "len_reward/correct/p75": 0.257729709148407, "len_reward/correct/var": 0.06340916454792023, "len_reward/incorrect": -0.14551971852779388, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.1081407368183136, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.23544758558273315, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.025769773870706558, "len_reward/max": 0.5, "len_reward/median": 0.0, "len_reward/min": -0.5, "len_reward/p25": -0.16495808959007263, "len_reward/p75": 0.20012810826301575, "len_reward/var": 0.061751849949359894, "loss": -0.6354, "mean_logprobs": -0.1513671875, "mean_logprobs/var": 0.01495361328125, "missing_response_penalty": -0.203125, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.5571428537368774, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": -1.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.2503105401992798, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.1627127081155777, "non_smoothed_logprobs/min": -16.8125, "num_completions/total": 768, "policy_loss": -0.6354151964187622, "policy_loss/max": 2.843683432729449e-05, "policy_loss/median": -0.9999978542327881, "policy_loss/min": -1.0000046491622925, "policy_loss/p25": -0.9999995231628418, "policy_loss/p75": 0.0, "policy_loss/var": 0.23287516832351685, "pred_value_log_probs": -0.7148666381835938, "pred_value_log_probs/max": -0.00012493133544921875, "pred_value_log_probs/median": -0.7578125, "pred_value_log_probs/min": -1.203125, "pred_value_log_probs/p25": -0.7578125, "pred_value_log_probs/p75": -0.7578125, "pred_value_log_probs/var": 0.024815544486045837, "reward": 0.6354166865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.23287522792816162, "rewards/accuracy_reward": 0.6354166865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.23287522792816162, "rewards/answer_logprob_reward": 2.72711181640625, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.2321436405181885, "rewards/answer_logprob_reward/p25": 2.4849066734313965, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.49596965312957764, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.0003473448450677097, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.0, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.16495808959007263, "rewards/len_reward/p75": 0.20012810826301575, "rewards/len_reward/var": 0.061751849949359894, "rewards/missing_response_penalty": -0.203125, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.1627127081155777, "smoothed_logprobs/min": -1.0, "step": 4, "value_loss": 2.390625, "values": -0.6947605013847351, "values/max": 1.9998908042907715, "values/median": -0.6875, "values/min": -1.89459228515625, "values/p25": -0.6875, "values/p75": -0.6875, "values/var": 0.1420281082391739 }, { "adam_stats/lr_effective_max": 5.4755189921706915e-05, "adam_stats/lr_effective_mean": -2.1790967641432957e-10, "adam_stats/lr_effective_min": -5.476594014908187e-05, "adam_stats/m_t_max": 0.018315263092517853, "adam_stats/m_t_mean": -1.1892935247725944e-10, "adam_stats/m_t_min": -0.014240323565900326, "adam_stats/v_t_max": 1.8482092855265364e-05, "adam_stats/v_t_mean": 6.644819785552036e-13, "adam_stats/v_t_min": 0.0, "epoch": 0.10126582278481013, "eval_accuracy_reward": 0.5633333325386047, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.24681159853935242, "eval_advantages": 0.5633333325386047, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.24681159853935242, "eval_all_logprobs": -0.10831712186336517, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": -2.384185791015625e-06, "eval_all_logprobs/min": -11.125, "eval_all_logprobs/p25": -0.002197265625, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.17934179306030273, "eval_answer_logprob_reward": 2.797687530517578, "eval_answer_logprob_reward/correct": 3.1648600101470947, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.040031738579273224, "eval_answer_logprob_reward/incorrect": 2.3240067958831787, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 2.525728702545166, "eval_answer_logprob_reward/incorrect/min": 1.021651268005371, "eval_answer_logprob_reward/incorrect/p25": 1.4271163940429688, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.8312591910362244, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.021651268005371, "eval_answer_logprob_reward/p25": 2.525728702545166, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.5584139823913574, "eval_completion_length": 631.703369140625, "eval_completion_length/correct": 575.7633056640625, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 518.0, "eval_completion_length/correct/min": 126.0, "eval_completion_length/correct/p25": 363.0, "eval_completion_length/correct/p75": 784.0, "eval_completion_length/correct/var": 71524.703125, "eval_completion_length/incorrect": 703.8702392578125, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 803.0, "eval_completion_length/incorrect/min": 24.0, "eval_completion_length/incorrect/p25": 410.0, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 115899.25, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 586.0, "eval_completion_length/min": 24.0, "eval_completion_length/p25": 377.5, "eval_completion_length/p75": 1024.0, "eval_completion_length/var": 94629.28125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 718733.0, "eval_len_reward": -0.01611892506480217, "eval_len_reward/correct": 0.07804737985134125, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.1132001057267189, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": -0.07658227533102036, "eval_len_reward/correct/p75": 0.26623716950416565, "eval_len_reward/correct/var": 0.06207133084535599, "eval_len_reward/incorrect": -0.13760064542293549, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.06359031796455383, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.24769610166549683, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.02720070257782936, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.1695822775363922, "eval_len_reward/p75": 0.14719371497631073, "eval_len_reward/var": 0.05818033963441849, "eval_mean_logprobs": -0.130859375, "eval_mean_logprobs/var": 0.0185546875, "eval_missing_response_penalty": -0.21666666865348816, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.49618321657180786, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.25190839171409607, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.17028985917568207, "eval_num_completions/total": 1200, "eval_pred_value_log_probs": -0.7214562892913818, "eval_pred_value_log_probs/max": -1.7881393432617188e-06, "eval_pred_value_log_probs/median": -0.765625, "eval_pred_value_log_probs/min": -1.078125, "eval_pred_value_log_probs/p25": -0.765625, "eval_pred_value_log_probs/p75": -0.765625, "eval_pred_value_log_probs/var": 0.023659951984882355, "eval_reward": 0.5633333325386047, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.24681159853935242, "eval_rewards/accuracy_reward": 0.5633333325386047, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.24681159853935242, "eval_rewards/answer_logprob_reward": 2.797687530517578, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.021651268005371, "eval_rewards/answer_logprob_reward/p25": 2.525728702545166, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.5584139823913574, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": -0.01611892506480217, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.1695822775363922, "eval_rewards/len_reward/p75": 0.14719371497631073, "eval_rewards/len_reward/var": 0.05818033963441849, "eval_rewards/missing_response_penalty": -0.21666666865348816, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.17028985917568207, "eval_runtime": 177.3334, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.7129337787628174, "eval_values/max": 1.9513092041015625, "eval_values/median": -0.6796875, "eval_values/min": -1.9838800430297852, "eval_values/p25": -0.6796875, "eval_values/p75": -0.6796875, "eval_values/var": 0.09706096351146698, "step": 4 }, { "accuracy_reward": 0.6927083730697632, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.21397797763347626, "adam_stats/lr_effective_max": 5.835445335833356e-05, "adam_stats/lr_effective_mean": 2.061619069904097e-10, "adam_stats/lr_effective_min": -5.8340159739600495e-05, "adam_stats/m_t_max": 0.011445377953350544, "adam_stats/m_t_mean": -6.51839693333045e-11, "adam_stats/m_t_min": -0.007816392928361893, "adam_stats/v_t_max": 2.100208257616032e-05, "adam_stats/v_t_mean": 1.1845901863594133e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.6927083730697632, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.21397797763347626, "all_logprobs": -0.11215102672576904, "all_logprobs/max": 0.0, "all_logprobs/median": -2.0265579223632812e-06, "all_logprobs/min": -12.375, "all_logprobs/p25": -0.0023345947265625, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.189283087849617, "answer_logprob_reward": 2.941647529602051, "answer_logprob_reward/correct": 3.1207261085510254, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.4849066734313965, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.03672637790441513, "answer_logprob_reward/incorrect": 2.5379624366760254, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.5686159133911133, "answer_logprob_reward/incorrect/p25": 1.7917594909667969, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.4904671311378479, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.5686159133911133, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.2469889521598816, "clip_ratio": 0.0, "completion_length": 601.8698120117188, "completion_length/correct": 492.4436340332031, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 409.0, "completion_length/correct/min": 157.0, "completion_length/correct/p25": 309.0, "completion_length/correct/p75": 635.0, "completion_length/correct/var": 57041.28515625, "completion_length/incorrect": 848.5423583984375, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 1024.0, "completion_length/incorrect/min": 10.0, "completion_length/incorrect/p25": 730.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 80482.3515625, "completion_length/max": 1024.0, "completion_length/median": 549.0, "completion_length/min": 10.0, "completion_length/p25": 345.0, "completion_length/p75": 947.0, "completion_length/var": 90994.625, "epoch": 0.12658227848101267, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 581351.0, "grad_norm": 2.539113759994507, "kl": 0.0005397796630859375, "learning_rate": 9.979871469976197e-06, "len_reward": 0.011436531320214272, "len_reward/correct": 0.11529415845870972, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.16104553639888763, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.0, "len_reward/correct/p75": 0.2899543344974518, "len_reward/correct/var": 0.05602065846323967, "len_reward/incorrect": -0.22268320620059967, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.18915121257305145, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.40228500962257385, "len_reward/incorrect/p75": -0.030559808015823364, "len_reward/incorrect/var": 0.034158919006586075, "len_reward/max": 0.5, "len_reward/median": 0.019392916932702065, "len_reward/min": -0.5, "len_reward/p25": -0.15810225903987885, "len_reward/p75": 0.21998034417629242, "len_reward/var": 0.0735311433672905, "loss": -0.6927, "mean_logprobs": -0.126953125, "mean_logprobs/var": 0.0135498046875, "missing_response_penalty": -0.1458333432674408, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.47457626461982727, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.2536528408527374, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.12521816790103912, "non_smoothed_logprobs/min": -16.4375, "num_completions/total": 960, "policy_loss": -0.6926867365837097, "policy_loss/max": 0.0002189636288676411, "policy_loss/median": -0.999973714351654, "policy_loss/min": -0.9999991059303284, "policy_loss/p25": -0.9999872446060181, "policy_loss/p75": 4.314817488193512e-06, "policy_loss/var": 0.213976189494133, "pred_value_log_probs": -0.6869064569473267, "pred_value_log_probs/max": -0.0019989013671875, "pred_value_log_probs/median": -0.765625, "pred_value_log_probs/min": -1.4453125, "pred_value_log_probs/p25": -0.765625, "pred_value_log_probs/p75": -0.765625, "pred_value_log_probs/var": 0.052704740315675735, "reward": 0.6927083730697632, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.21397797763347626, "rewards/accuracy_reward": 0.6927083730697632, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.21397797763347626, "rewards/answer_logprob_reward": 2.941647529602051, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.5686159133911133, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.2469889521598816, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.011436531320214272, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.019392916932702065, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.15810225903987885, "rewards/len_reward/p75": 0.21998034417629242, "rewards/len_reward/var": 0.0735311433672905, "rewards/missing_response_penalty": -0.1458333432674408, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.12521816790103912, "smoothed_logprobs/min": -1.0, "step": 5, "value_loss": 2.6484375, "values": -0.7231578826904297, "values/max": 1.9903125762939453, "values/median": -0.6796875, "values/min": -1.866668701171875, "values/p25": -0.6796875, "values/p75": -0.6796875, "values/var": 0.25189828872680664 }, { "adam_stats/lr_effective_max": 5.835445335833356e-05, "adam_stats/lr_effective_mean": 2.061619069904097e-10, "adam_stats/lr_effective_min": -5.8340159739600495e-05, "adam_stats/m_t_max": 0.011445377953350544, "adam_stats/m_t_mean": -6.51839693333045e-11, "adam_stats/m_t_min": -0.007816392928361893, "adam_stats/v_t_max": 2.100208257616032e-05, "adam_stats/v_t_mean": 1.1845901863594133e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.12658227848101267, "eval_accuracy_reward": 0.596666693687439, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.24146044254302979, "eval_advantages": 0.596666693687439, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.24146044254302979, "eval_all_logprobs": -0.0702953189611435, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": -1.1920928955078125e-07, "eval_all_logprobs/min": -16.25, "eval_all_logprobs/p25": -0.00011110305786132812, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.11423061788082123, "eval_answer_logprob_reward": 2.964020013809204, "eval_answer_logprob_reward/correct": 3.1385061740875244, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.057033028453588486, "eval_answer_logprob_reward/incorrect": 2.7058961391448975, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.021651268005371, "eval_answer_logprob_reward/incorrect/p25": 2.1202635765075684, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.5137178301811218, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.021651268005371, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.2853167951107025, "eval_completion_length": 695.0266723632812, "eval_completion_length/correct": 613.7988891601562, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 555.0, "eval_completion_length/correct/min": 12.0, "eval_completion_length/correct/p25": 363.0, "eval_completion_length/correct/p75": 905.0, "eval_completion_length/correct/var": 84074.828125, "eval_completion_length/incorrect": 815.1900634765625, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 1024.0, "eval_completion_length/incorrect/min": 19.0, "eval_completion_length/incorrect/p25": 611.0, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 70829.8515625, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 689.0, "eval_completion_length/min": 12.0, "eval_completion_length/p25": 419.5, "eval_completion_length/p75": 1024.0, "eval_completion_length/var": 88271.1875, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 927241.0, "eval_len_reward": -0.01350005529820919, "eval_len_reward/correct": 0.08121874183416367, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.15369127690792084, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": -0.13007400929927826, "eval_len_reward/correct/p75": 0.3086293339729309, "eval_len_reward/correct/var": 0.08572941273450851, "eval_len_reward/incorrect": -0.15362125635147095, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.08638916164636612, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.275522381067276, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.02721996046602726, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.21404384076595306, "eval_len_reward/p75": 0.21088477969169617, "eval_len_reward/var": 0.07527714222669601, "eval_mean_logprobs": -0.0732421875, "eval_mean_logprobs/var": 0.00927734375, "eval_missing_response_penalty": -0.13333334028720856, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.3305785059928894, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.22314050793647766, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.11594202369451523, "eval_num_completions/total": 1500, "eval_pred_value_log_probs": -0.7040350437164307, "eval_pred_value_log_probs/max": -2.384185791015625e-06, "eval_pred_value_log_probs/median": -0.74609375, "eval_pred_value_log_probs/min": -1.1015625, "eval_pred_value_log_probs/p25": -0.74609375, "eval_pred_value_log_probs/p75": -0.74609375, "eval_pred_value_log_probs/var": 0.023306304588913918, "eval_reward": 0.596666693687439, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.24146044254302979, "eval_rewards/accuracy_reward": 0.596666693687439, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.24146044254302979, "eval_rewards/answer_logprob_reward": 2.964020013809204, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.021651268005371, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.2853167951107025, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": -0.01350005529820919, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.21404384076595306, "eval_rewards/len_reward/p75": 0.21088477969169617, "eval_rewards/len_reward/var": 0.07527714222669601, "eval_rewards/missing_response_penalty": -0.13333334028720856, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.11594202369451523, "eval_runtime": 169.8057, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.7374175786972046, "eval_values/max": 1.9597244262695312, "eval_values/median": -0.70751953125, "eval_values/min": -1.983614444732666, "eval_values/p25": -0.70751953125, "eval_values/p75": -0.70751953125, "eval_values/var": 0.10163413733243942, "step": 5 }, { "accuracy_reward": 0.6822916865348816, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.21790467202663422, "adam_stats/lr_effective_max": 6.074368502595462e-05, "adam_stats/lr_effective_mean": 3.5626293581891844e-10, "adam_stats/lr_effective_min": -6.089979433454573e-05, "adam_stats/m_t_max": 0.011484922841191292, "adam_stats/m_t_mean": -7.447095717871832e-11, "adam_stats/m_t_min": -0.007474206853657961, "adam_stats/v_t_max": 2.1121284589753486e-05, "adam_stats/v_t_mean": 1.2551746751340853e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.6822916865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.21790467202663422, "all_logprobs": -0.06589704751968384, "all_logprobs/max": 0.0, "all_logprobs/median": -1.1920928955078125e-07, "all_logprobs/min": -9.9375, "all_logprobs/p25": -5.27501106262207e-05, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.10793641209602356, "answer_logprob_reward": 2.9347877502441406, "answer_logprob_reward/correct": 3.069532632827759, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.08449158817529678, "answer_logprob_reward/incorrect": 2.6454176902770996, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.5686159133911133, "answer_logprob_reward/incorrect/p25": 2.079441547393799, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.40474987030029297, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.5686159133911133, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.2238491177558899, "clip_ratio": 0.0, "completion_length": 678.4166870117188, "completion_length/correct": 585.5496215820312, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 522.0, "completion_length/correct/min": 184.0, "completion_length/correct/p25": 366.5, "completion_length/correct/p75": 813.5, "completion_length/correct/var": 72189.390625, "completion_length/incorrect": 877.8524169921875, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 1024.0, "completion_length/incorrect/min": 35.0, "completion_length/incorrect/p25": 828.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 69250.4296875, "completion_length/max": 1024.0, "completion_length/median": 646.0, "completion_length/min": 35.0, "completion_length/p25": 407.75, "completion_length/p75": 1024.0, "completion_length/var": 89506.1796875, "epoch": 0.1518987341772152, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 711607.0, "grad_norm": 0.4030705690383911, "kl": 0.002655029296875, "learning_rate": 9.91964794299315e-06, "len_reward": 0.02160990796983242, "len_reward/correct": 0.12393714487552643, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.15581050515174866, "len_reward/correct/min": -0.5, "len_reward/correct/p25": -0.026404786854982376, "len_reward/correct/p75": 0.3236168324947357, "len_reward/correct/var": 0.07225022464990616, "len_reward/incorrect": -0.19814200699329376, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.18134714663028717, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.34914764761924744, "len_reward/incorrect/p75": -0.04173106700181961, "len_reward/incorrect/var": 0.028696652501821518, "len_reward/max": 0.5, "len_reward/median": 0.0, "len_reward/min": -0.5, "len_reward/p25": -0.18261007964611053, "len_reward/p75": 0.26758527755737305, "len_reward/var": 0.08079454302787781, "loss": -0.6822, "mean_logprobs": -0.064453125, "mean_logprobs/var": 0.00128173828125, "missing_response_penalty": -0.1302083432674408, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.4098360538482666, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.24590164422988892, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.11384708434343338, "non_smoothed_logprobs/min": -19.1875, "num_completions/total": 1152, "policy_loss": -0.6821851134300232, "policy_loss/max": 0.000812639482319355, "policy_loss/median": -0.9998696446418762, "policy_loss/min": -0.9999760389328003, "policy_loss/p25": -0.9999115467071533, "policy_loss/p75": 5.626119673252106e-05, "policy_loss/var": 0.21789757907390594, "pred_value_log_probs": -0.6675887107849121, "pred_value_log_probs/max": -0.0004329681396484375, "pred_value_log_probs/median": -0.74609375, "pred_value_log_probs/min": -1.203125, "pred_value_log_probs/p25": -0.74609375, "pred_value_log_probs/p75": -0.744140625, "pred_value_log_probs/var": 0.04631832242012024, "reward": 0.6822916865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.21790467202663422, "rewards/accuracy_reward": 0.6822916865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.21790467202663422, "rewards/answer_logprob_reward": 2.9347877502441406, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.5686159133911133, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.2238491177558899, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.02160990796983242, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.0, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.18261007964611053, "rewards/len_reward/p75": 0.26758527755737305, "rewards/len_reward/var": 0.08079454302787781, "rewards/missing_response_penalty": -0.1302083432674408, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.11384708434343338, "smoothed_logprobs/min": -1.0, "step": 6, "value_loss": 2.6328125, "values": -0.7369192838668823, "values/max": 1.5349578857421875, "values/median": -0.70751953125, "values/min": -1.932159423828125, "values/p25": -0.70751953125, "values/p75": -0.70751953125, "values/var": 0.23455187678337097 }, { "adam_stats/lr_effective_max": 6.074368502595462e-05, "adam_stats/lr_effective_mean": 3.5626293581891844e-10, "adam_stats/lr_effective_min": -6.089979433454573e-05, "adam_stats/m_t_max": 0.011484922841191292, "adam_stats/m_t_mean": -7.447095717871832e-11, "adam_stats/m_t_min": -0.007474206853657961, "adam_stats/v_t_max": 2.1121284589753486e-05, "adam_stats/v_t_mean": 1.2551746751340853e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.1518987341772152, "eval_accuracy_reward": 0.6533333659172058, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.22724635899066925, "eval_advantages": 0.6533333659172058, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.22724635899066925, "eval_all_logprobs": -0.04569603130221367, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -8.0625, "eval_all_logprobs/p25": -4.291534423828125e-06, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.0705338642001152, "eval_answer_logprob_reward": 2.985471248626709, "eval_answer_logprob_reward/correct": 3.1265928745269775, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.07172445952892303, "eval_answer_logprob_reward/incorrect": 2.7195112705230713, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.1394343376159668, "eval_answer_logprob_reward/incorrect/p25": 2.1202635765075684, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.4924987852573395, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.1394343376159668, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.2540917694568634, "eval_completion_length": 595.2000122070312, "eval_completion_length/correct": 518.8673095703125, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 476.0, "eval_completion_length/correct/min": 98.0, "eval_completion_length/correct/p25": 327.5, "eval_completion_length/correct/p75": 678.0, "eval_completion_length/correct/var": 52477.60546875, "eval_completion_length/incorrect": 739.0577392578125, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 752.0, "eval_completion_length/incorrect/min": 131.0, "eval_completion_length/incorrect/p25": 487.75, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 78506.140625, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 549.0, "eval_completion_length/min": 98.0, "eval_completion_length/p25": 360.0, "eval_completion_length/p75": 808.75, "eval_completion_length/var": 72286.203125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 1105801.0, "eval_len_reward": 0.0503496453166008, "eval_len_reward/correct": 0.15384456515312195, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.18005843460559845, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": -0.01752505823969841, "eval_len_reward/correct/p75": 0.3630443811416626, "eval_len_reward/correct/var": 0.06298115104436874, "eval_len_reward/incorrect": -0.14469848573207855, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.06391526758670807, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.27572810649871826, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.030527573078870773, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.12186340987682343, "eval_len_reward/p75": 0.2680339813232422, "eval_len_reward/var": 0.07184485346078873, "eval_mean_logprobs": -0.044921875, "eval_mean_logprobs/var": 0.00113677978515625, "eval_missing_response_penalty": -0.11666666716337204, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.3365384638309479, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.2254481017589569, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.10340023040771484, "eval_num_completions/total": 1800, "eval_pred_value_log_probs": -0.6845210790634155, "eval_pred_value_log_probs/max": -1.1920928955078125e-07, "eval_pred_value_log_probs/median": -0.7265625, "eval_pred_value_log_probs/min": -1.0546875, "eval_pred_value_log_probs/p25": -0.7265625, "eval_pred_value_log_probs/p75": -0.7265625, "eval_pred_value_log_probs/var": 0.024145593866705894, "eval_reward": 0.6533333659172058, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.22724635899066925, "eval_rewards/accuracy_reward": 0.6533333659172058, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.22724635899066925, "eval_rewards/answer_logprob_reward": 2.985471248626709, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.1394343376159668, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.2540917694568634, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.0503496453166008, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.12186340987682343, "eval_rewards/len_reward/p75": 0.2680339813232422, "eval_rewards/len_reward/var": 0.07184485346078873, "eval_rewards/missing_response_penalty": -0.11666666716337204, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.10340023040771484, "eval_runtime": 166.4312, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.764191210269928, "eval_values/max": 1.9675979614257812, "eval_values/median": -0.73388671875, "eval_values/min": -1.9830894470214844, "eval_values/p25": -0.73388671875, "eval_values/p75": -0.73388671875, "eval_values/var": 0.1079685240983963, "step": 6 }, { "accuracy_reward": 0.7239583730697632, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.2008889615535736, "adam_stats/lr_effective_max": 6.182432116474956e-05, "adam_stats/lr_effective_mean": 4.6742158754664587e-11, "adam_stats/lr_effective_min": -6.246730481507257e-05, "adam_stats/m_t_max": 0.009906132705509663, "adam_stats/m_t_mean": -5.941885872218222e-11, "adam_stats/m_t_min": -0.006426454987376928, "adam_stats/v_t_max": 2.111867979692761e-05, "adam_stats/v_t_mean": 1.2852747295274969e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7239583730697632, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.2008889615535736, "all_logprobs": -0.04034246504306793, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -7.375, "all_logprobs/p25": -1.6689300537109375e-06, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.0620364248752594, "answer_logprob_reward": 2.985330104827881, "answer_logprob_reward/correct": 3.119436264038086, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.04714789614081383, "answer_logprob_reward/incorrect": 2.6336169242858887, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.3862943649291992, "answer_logprob_reward/incorrect/p25": 2.079441547393799, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.5028464794158936, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.3862943649291992, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.2183794379234314, "clip_ratio": 0.0, "completion_length": 587.6927490234375, "completion_length/correct": 519.0935668945312, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 472.0, "completion_length/correct/min": 163.0, "completion_length/correct/p25": 351.5, "completion_length/correct/p75": 648.0, "completion_length/correct/var": 45573.63671875, "completion_length/incorrect": 767.603759765625, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 976.0, "completion_length/incorrect/min": 12.0, "completion_length/incorrect/p25": 519.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 90514.6640625, "completion_length/max": 1024.0, "completion_length/median": 533.0, "completion_length/min": 12.0, "completion_length/p25": 379.5, "completion_length/p75": 777.75, "completion_length/var": 69976.6640625, "epoch": 0.17721518987341772, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 824444.0, "grad_norm": 0.2801690399646759, "kl": 0.004913330078125, "learning_rate": 9.819814303479268e-06, "len_reward": 0.07311590760946274, "len_reward/correct": 0.159356489777565, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.19765864312648773, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.005923635791987181, "len_reward/correct/p75": 0.3308439254760742, "len_reward/correct/var": 0.05828801542520523, "len_reward/incorrect": -0.15306219458580017, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.09063062816858292, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.2879432737827301, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.02899157628417015, "len_reward/max": 0.5, "len_reward/median": 0.05495617166161537, "len_reward/min": -0.5, "len_reward/p25": -0.07770079374313354, "len_reward/p75": 0.2734633982181549, "len_reward/var": 0.0696147084236145, "loss": -0.7238, "mean_logprobs": -0.04443359375, "mean_logprobs/var": 0.006378173828125, "missing_response_penalty": -0.1197916716337204, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.43396228551864624, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.2503628432750702, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.10599368065595627, "non_smoothed_logprobs/min": -19.5625, "num_completions/total": 1344, "policy_loss": -0.723761796951294, "policy_loss/max": 0.00042992449016310275, "policy_loss/median": -0.9997755289077759, "policy_loss/min": -0.9999435544013977, "policy_loss/p25": -0.9998378753662109, "policy_loss/p75": 8.520297706127167e-05, "policy_loss/var": 0.20088227093219757, "pred_value_log_probs": -0.6857487559318542, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.7265625, "pred_value_log_probs/min": -1.2109375, "pred_value_log_probs/p25": -0.7265625, "pred_value_log_probs/p75": -0.7265625, "pred_value_log_probs/var": 0.032485608011484146, "reward": 0.7239583730697632, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.2008889615535736, "rewards/accuracy_reward": 0.7239583730697632, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.2008889615535736, "rewards/answer_logprob_reward": 2.985330104827881, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.3862943649291992, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.2183794379234314, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.07311590760946274, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.05495617166161537, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.07770079374313354, "rewards/len_reward/p75": 0.2734633982181549, "rewards/len_reward/var": 0.0696147084236145, "rewards/missing_response_penalty": -0.1197916716337204, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.10599368065595627, "smoothed_logprobs/min": -1.0, "step": 7, "value_loss": 2.8671875, "values": -0.7097371816635132, "values/max": 2.0, "values/median": -0.73388671875, "values/min": -2.0, "values/p25": -0.73388671875, "values/p75": -0.73388671875, "values/var": 0.18856659531593323 }, { "adam_stats/lr_effective_max": 6.182432116474956e-05, "adam_stats/lr_effective_mean": 4.6742158754664587e-11, "adam_stats/lr_effective_min": -6.246730481507257e-05, "adam_stats/m_t_max": 0.009906132705509663, "adam_stats/m_t_mean": -5.941885872218222e-11, "adam_stats/m_t_min": -0.006426454987376928, "adam_stats/v_t_max": 2.111867979692761e-05, "adam_stats/v_t_mean": 1.2852747295274969e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.17721518987341772, "eval_accuracy_reward": 0.6500000357627869, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.2282608449459076, "eval_advantages": 0.6500000357627869, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.2282608449459076, "eval_all_logprobs": -0.034109558910131454, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -9.625, "eval_all_logprobs/p25": -5.960464477539062e-07, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.0519154891371727, "eval_answer_logprob_reward": 2.9738736152648926, "eval_answer_logprob_reward/correct": 3.1169309616088867, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.07894976437091827, "eval_answer_logprob_reward/incorrect": 2.708195686340332, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.1394343376159668, "eval_answer_logprob_reward/incorrect/p25": 2.1202635765075684, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.45827212929725647, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.1394343376159668, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.24875830113887787, "eval_completion_length": 580.086669921875, "eval_completion_length/correct": 498.4154052734375, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 462.0, "eval_completion_length/correct/min": 180.0, "eval_completion_length/correct/p25": 337.5, "eval_completion_length/correct/p75": 625.0, "eval_completion_length/correct/var": 42704.66796875, "eval_completion_length/incorrect": 731.761962890625, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 750.0, "eval_completion_length/incorrect/min": 53.0, "eval_completion_length/incorrect/p25": 481.0, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 78969.8046875, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 526.0, "eval_completion_length/min": 53.0, "eval_completion_length/p25": 373.0, "eval_completion_length/p75": 771.5, "eval_completion_length/var": 67604.7421875, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 1279827.0, "eval_len_reward": 0.04109089449048042, "eval_len_reward/correct": 0.14343681931495667, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.1515151560306549, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.0, "eval_len_reward/correct/p75": 0.3186544179916382, "eval_len_reward/correct/var": 0.053380273282527924, "eval_len_reward/incorrect": -0.14898011088371277, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.06081081181764603, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.2922284007072449, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03230222314596176, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.09275729954242706, "eval_len_reward/p75": 0.23071445524692535, "eval_len_reward/var": 0.06538829207420349, "eval_mean_logprobs": -0.03369140625, "eval_mean_logprobs/var": 0.0013427734375, "eval_missing_response_penalty": -0.12666666507720947, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.3619047701358795, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.2331501841545105, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.1109922006726265, "eval_num_completions/total": 2100, "eval_pred_value_log_probs": -0.6708398461341858, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.71484375, "eval_pred_value_log_probs/min": -1.0625, "eval_pred_value_log_probs/p25": -0.71484375, "eval_pred_value_log_probs/p75": -0.71484375, "eval_pred_value_log_probs/var": 0.025487350299954414, "eval_reward": 0.6500000357627869, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.2282608449459076, "eval_rewards/accuracy_reward": 0.6500000357627869, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.2282608449459076, "eval_rewards/answer_logprob_reward": 2.9738736152648926, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.1394343376159668, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.24875830113887787, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.04109089449048042, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.09275729954242706, "eval_rewards/len_reward/p75": 0.23071445524692535, "eval_rewards/len_reward/var": 0.06538829207420349, "eval_rewards/missing_response_penalty": -0.12666666507720947, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.1109922006726265, "eval_runtime": 186.1527, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.7883679270744324, "eval_values/max": 1.9758377075195312, "eval_values/median": -0.7548828125, "eval_values/min": -1.973602294921875, "eval_values/p25": -0.7548828125, "eval_values/p75": -0.7548828125, "eval_values/var": 0.11575037986040115, "step": 7 }, { "accuracy_reward": 0.7760416865348816, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.1747109442949295, "adam_stats/lr_effective_max": 6.243507232284173e-05, "adam_stats/lr_effective_mean": -1.6910949030002342e-10, "adam_stats/lr_effective_min": -6.287322321441025e-05, "adam_stats/m_t_max": 0.007395743392407894, "adam_stats/m_t_mean": -2.0997197036631832e-11, "adam_stats/m_t_min": -0.004758418537676334, "adam_stats/v_t_max": 2.132852932845708e-05, "adam_stats/v_t_mean": 1.3262481390885483e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7760416865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.1747109442949295, "all_logprobs": -0.025388197973370552, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -7.53125, "all_logprobs/p25": -1.1920928955078125e-07, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.037190306931734085, "answer_logprob_reward": 2.963308334350586, "answer_logprob_reward/correct": 3.0721981525421143, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.07161752879619598, "answer_logprob_reward/incorrect": 2.5859925746917725, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.5686159133911133, "answer_logprob_reward/incorrect/p25": 2.079441547393799, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.47603172063827515, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.5686159133911133, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.20147234201431274, "clip_ratio": 0.0, "completion_length": 567.984375, "completion_length/correct": 491.16778564453125, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 435.0, "completion_length/correct/min": 173.0, "completion_length/correct/p25": 325.0, "completion_length/correct/p75": 593.0, "completion_length/correct/var": 47529.54296875, "completion_length/incorrect": 834.1627807617188, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 1024.0, "completion_length/incorrect/min": 308.0, "completion_length/incorrect/p25": 657.5, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 65805.8125, "completion_length/max": 1024.0, "completion_length/median": 483.0, "completion_length/min": 173.0, "completion_length/p25": 357.5, "completion_length/p75": 758.75, "completion_length/var": 71853.5390625, "epoch": 0.20253164556962025, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 933497.0, "grad_norm": 0.3094690442085266, "kl": 0.006195068359375, "learning_rate": 9.681174353198687e-06, "len_reward": 0.1136961430311203, "len_reward/correct": 0.20485958456993103, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.2629537582397461, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.10438908636569977, "len_reward/correct/p75": 0.3595964014530182, "len_reward/correct/var": 0.04752177372574806, "len_reward/incorrect": -0.20219576358795166, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.15865178406238556, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.36952483654022217, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03490697592496872, "len_reward/max": 0.5, "len_reward/median": 0.16337041556835175, "len_reward/min": -0.5, "len_reward/p25": -0.06925267726182938, "len_reward/p75": 0.32950764894485474, "len_reward/var": 0.07344760000705719, "loss": -0.7758, "mean_logprobs": -0.023193359375, "mean_logprobs/var": 0.00034332275390625, "missing_response_penalty": -0.109375, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.4883720874786377, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.25581395626068115, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.09792213141918182, "non_smoothed_logprobs/min": -19.8125, "num_completions/total": 1536, "policy_loss": -0.7757939100265503, "policy_loss/max": 0.0003666747361421585, "policy_loss/median": -0.9997168779373169, "policy_loss/min": -0.9999092221260071, "policy_loss/p25": -0.999791145324707, "policy_loss/p75": -0.9995872378349304, "policy_loss/var": 0.1746944934129715, "pred_value_log_probs": -0.6520019769668579, "pred_value_log_probs/max": -0.0028533935546875, "pred_value_log_probs/median": -0.71484375, "pred_value_log_probs/min": -1.3125, "pred_value_log_probs/p25": -0.71484375, "pred_value_log_probs/p75": -0.71484375, "pred_value_log_probs/var": 0.051762331277132034, "reward": 0.7760416865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.1747109442949295, "rewards/accuracy_reward": 0.7760416865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.1747109442949295, "rewards/answer_logprob_reward": 2.963308334350586, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.5686159133911133, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.20147234201431274, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.1136961430311203, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.16337041556835175, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.06925267726182938, "rewards/len_reward/p75": 0.32950764894485474, "rewards/len_reward/var": 0.07344760000705719, "rewards/missing_response_penalty": -0.109375, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.09792213141918182, "smoothed_logprobs/min": -1.0, "step": 8, "value_loss": 2.640625, "values": -0.8079665899276733, "values/max": 1.77435302734375, "values/median": -0.7548828125, "values/min": -1.957733154296875, "values/p25": -0.7548828125, "values/p75": -0.7548828125, "values/var": 0.18137165904045105 }, { "adam_stats/lr_effective_max": 6.243507232284173e-05, "adam_stats/lr_effective_mean": -1.6910949030002342e-10, "adam_stats/lr_effective_min": -6.287322321441025e-05, "adam_stats/m_t_max": 0.007395743392407894, "adam_stats/m_t_mean": -2.0997197036631832e-11, "adam_stats/m_t_min": -0.004758418537676334, "adam_stats/v_t_max": 2.132852932845708e-05, "adam_stats/v_t_mean": 1.3262481390885483e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.20253164556962025, "eval_accuracy_reward": 0.6266666650772095, "eval_accuracy_reward/correct": 0.9999999403953552, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.23473802208900452, "eval_advantages": 0.6266666650772095, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.23473802208900452, "eval_all_logprobs": -0.026214059442281723, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -8.9375, "eval_all_logprobs/p25": -1.1920928955078125e-07, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.040528010576963425, "eval_answer_logprob_reward": 2.9724133014678955, "eval_answer_logprob_reward/correct": 3.1414499282836914, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.525728702545166, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.047927774488925934, "eval_answer_logprob_reward/incorrect": 2.688673973083496, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.272965669631958, "eval_answer_logprob_reward/incorrect/p25": 2.1202635765075684, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.4752677083015442, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.272965669631958, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.25453484058380127, "eval_completion_length": 592.52001953125, "eval_completion_length/correct": 477.4201965332031, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 454.0, "eval_completion_length/correct/min": 167.0, "eval_completion_length/correct/p25": 318.75, "eval_completion_length/correct/p75": 612.75, "eval_completion_length/correct/var": 40107.1875, "eval_completion_length/incorrect": 785.7232666015625, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 880.0, "eval_completion_length/incorrect/min": 167.0, "eval_completion_length/incorrect/p25": 580.25, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 69330.3671875, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 555.0, "eval_completion_length/min": 167.0, "eval_completion_length/p25": 357.0, "eval_completion_length/p75": 806.0, "eval_completion_length/var": 73133.8125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 1457583.0, "eval_len_reward": 0.044376272708177567, "eval_len_reward/correct": 0.17966805398464203, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.20304115116596222, "eval_len_reward/correct/min": -0.489673912525177, "eval_len_reward/correct/p25": 0.03819954767823219, "eval_len_reward/correct/p75": 0.3427225351333618, "eval_len_reward/correct/var": 0.04893714562058449, "eval_len_reward/incorrect": -0.18272064626216888, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.15007656812667847, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.3179081678390503, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03250543028116226, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.13343378901481628, "eval_len_reward/p75": 0.27335086464881897, "eval_len_reward/var": 0.07350050657987595, "eval_mean_logprobs": -0.025634765625, "eval_mean_logprobs/var": 0.0003719329833984375, "eval_missing_response_penalty": -0.1366666704416275, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.3660714328289032, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.23415379226207733, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.1183835044503212, "eval_num_completions/total": 2400, "eval_pred_value_log_probs": -0.662541925907135, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.70703125, "eval_pred_value_log_probs/min": -1.0546875, "eval_pred_value_log_probs/p25": -0.70703125, "eval_pred_value_log_probs/p75": -0.70703125, "eval_pred_value_log_probs/var": 0.02536529116332531, "eval_reward": 0.6266666650772095, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.23473802208900452, "eval_rewards/accuracy_reward": 0.6266666650772095, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.23473802208900452, "eval_rewards/answer_logprob_reward": 2.9724133014678955, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.272965669631958, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.25453484058380127, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.044376272708177567, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.13343378901481628, "eval_rewards/len_reward/p75": 0.27335086464881897, "eval_rewards/len_reward/var": 0.07350050657987595, "eval_rewards/missing_response_penalty": -0.1366666704416275, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.1183835044503212, "eval_runtime": 171.5273, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8006720542907715, "eval_values/max": 1.9754486083984375, "eval_values/median": -0.76513671875, "eval_values/min": -1.9915542602539062, "eval_values/p25": -0.76513671875, "eval_values/p75": -0.76513671875, "eval_values/var": 0.1165805384516716, "step": 8 }, { "accuracy_reward": 0.7916666865348816, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.16579407453536987, "adam_stats/lr_effective_max": 6.244866381166503e-05, "adam_stats/lr_effective_mean": -1.4529581704447736e-10, "adam_stats/lr_effective_min": -6.220162322279066e-05, "adam_stats/m_t_max": 0.004721947014331818, "adam_stats/m_t_mean": -2.0175234746733217e-11, "adam_stats/m_t_min": -0.004131162539124489, "adam_stats/v_t_max": 2.23145070776809e-05, "adam_stats/v_t_mean": 1.3832478999026288e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7916666865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.16579407453536987, "all_logprobs": -0.02305363304913044, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -9.125, "all_logprobs/p25": -1.1920928955078125e-07, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.03668288514018059, "answer_logprob_reward": 3.02274751663208, "answer_logprob_reward/correct": 3.061724901199341, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.08524507284164429, "answer_logprob_reward/incorrect": 2.8746325969696045, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 2.079441547393799, "answer_logprob_reward/incorrect/p25": 2.4849066734313965, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.18981382250785828, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 2.079441547393799, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.11195386201143265, "clip_ratio": 0.0, "completion_length": 533.984375, "completion_length/correct": 463.02630615234375, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 413.0, "completion_length/correct/min": 175.0, "completion_length/correct/p25": 323.75, "completion_length/correct/p75": 546.5, "completion_length/correct/var": 37277.74609375, "completion_length/incorrect": 803.625, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 804.0, "completion_length/incorrect/min": 259.0, "completion_length/incorrect/p25": 660.5, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 58065.87890625, "completion_length/max": 1024.0, "completion_length/median": 457.0, "completion_length/min": 175.0, "completion_length/p25": 344.0, "completion_length/p75": 686.75, "completion_length/var": 60560.6171875, "epoch": 0.22784810126582278, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 1036022.0, "grad_norm": 0.36979013681411743, "kl": 0.007781982421875, "learning_rate": 9.504844339512096e-06, "len_reward": 0.13034191727638245, "len_reward/correct": 0.21805612742900848, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.2795400619506836, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.10460343956947327, "len_reward/correct/p75": 0.3683248460292816, "len_reward/correct/var": 0.04483015090227127, "len_reward/incorrect": -0.2029721736907959, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.1942392885684967, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.3398860991001129, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03722740337252617, "len_reward/max": 0.5, "len_reward/median": 0.1930946260690689, "len_reward/min": -0.5, "len_reward/p25": -0.0009001440485008061, "len_reward/p75": 0.3444806933403015, "len_reward/var": 0.0724324956536293, "loss": -0.7914, "mean_logprobs": -0.022216796875, "mean_logprobs/var": 0.00019550323486328125, "missing_response_penalty": -0.0677083358168602, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.32500001788139343, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.22500000894069672, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.06345440447330475, "non_smoothed_logprobs/min": -19.9375, "num_completions/total": 1728, "policy_loss": -0.7913558483123779, "policy_loss/max": 0.0004762290045619011, "policy_loss/median": -0.9996488690376282, "policy_loss/min": -0.9999149441719055, "policy_loss/p25": -0.9997439980506897, "policy_loss/p75": -0.9994585514068604, "policy_loss/var": 0.16577255725860596, "pred_value_log_probs": -0.6520935297012329, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.70703125, "pred_value_log_probs/min": -1.140625, "pred_value_log_probs/p25": -0.70703125, "pred_value_log_probs/p75": -0.70703125, "pred_value_log_probs/var": 0.041593849658966064, "reward": 0.7916666865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.16579407453536987, "rewards/accuracy_reward": 0.7916666865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.16579407453536987, "rewards/answer_logprob_reward": 3.02274751663208, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 2.079441547393799, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.11195386201143265, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.13034191727638245, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.1930946260690689, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.0009001440485008061, "rewards/len_reward/p75": 0.3444806933403015, "rewards/len_reward/var": 0.0724324956536293, "rewards/missing_response_penalty": -0.0677083358168602, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.06345440447330475, "smoothed_logprobs/min": -1.0, "step": 9, "value_loss": 2.953125, "values": -0.8088086843490601, "values/max": 1.8154296875, "values/median": -0.76513671875, "values/min": -2.000594139099121, "values/p25": -0.76513671875, "values/p75": -0.76513671875, "values/var": 0.20579427480697632 }, { "adam_stats/lr_effective_max": 6.244866381166503e-05, "adam_stats/lr_effective_mean": -1.4529581704447736e-10, "adam_stats/lr_effective_min": -6.220162322279066e-05, "adam_stats/m_t_max": 0.004721947014331818, "adam_stats/m_t_mean": -2.0175234746733217e-11, "adam_stats/m_t_min": -0.004131162539124489, "adam_stats/v_t_max": 2.23145070776809e-05, "adam_stats/v_t_mean": 1.3832478999026288e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.22784810126582278, "eval_accuracy_reward": 0.6700000166893005, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.22183947265148163, "eval_advantages": 0.6700000166893005, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.22183947265148163, "eval_all_logprobs": -0.023103248327970505, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -10.375, "eval_all_logprobs/p25": -1.1920928955078125e-07, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.03600381314754486, "eval_answer_logprob_reward": 3.008866548538208, "eval_answer_logprob_reward/correct": 3.1257662773132324, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.525728702545166, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.0561484731733799, "eval_answer_logprob_reward/incorrect": 2.7715237140655518, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.4271163940429688, "eval_answer_logprob_reward/incorrect/p25": 2.1202635765075684, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.4142383635044098, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.4271163940429688, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.20116610825061798, "eval_completion_length": 577.9866943359375, "eval_completion_length/correct": 494.14923095703125, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 442.0, "eval_completion_length/correct/min": 155.0, "eval_completion_length/correct/p25": 323.0, "eval_completion_length/correct/p75": 641.0, "eval_completion_length/correct/var": 47330.5234375, "eval_completion_length/incorrect": 748.2020263671875, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 759.0, "eval_completion_length/incorrect/min": 187.0, "eval_completion_length/incorrect/p25": 563.5, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 66543.4140625, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 525.0, "eval_completion_length/min": 155.0, "eval_completion_length/p25": 365.75, "eval_completion_length/p75": 761.25, "eval_completion_length/var": 67787.5703125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 1630979.0, "eval_len_reward": 0.043535925447940826, "eval_len_reward/correct": 0.14882901310920715, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.1847875714302063, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.005553270224481821, "eval_len_reward/correct/p75": 0.32223182916641235, "eval_len_reward/correct/var": 0.058313969522714615, "eval_len_reward/incorrect": -0.17024099826812744, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.13499315083026886, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.30066239833831787, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.030154937878251076, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.005553270224481821, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.14721713960170746, "eval_len_reward/p75": 0.24546536803245544, "eval_len_reward/var": 0.07147407531738281, "eval_mean_logprobs": -0.0224609375, "eval_mean_logprobs/var": 0.0003147125244140625, "eval_missing_response_penalty": -0.11666666716337204, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.35353535413742065, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.2308802306652069, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.10340021550655365, "eval_num_completions/total": 2700, "eval_pred_value_log_probs": -0.6526988744735718, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.69921875, "eval_pred_value_log_probs/min": -0.9375, "eval_pred_value_log_probs/p25": -0.69921875, "eval_pred_value_log_probs/p75": -0.69921875, "eval_pred_value_log_probs/var": 0.026807580143213272, "eval_reward": 0.6700000166893005, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.22183947265148163, "eval_rewards/accuracy_reward": 0.6700000166893005, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.22183947265148163, "eval_rewards/answer_logprob_reward": 3.008866548538208, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.4271163940429688, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.20116610825061798, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.043535925447940826, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.005553270224481821, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.14721713960170746, "eval_rewards/len_reward/p75": 0.24546536803245544, "eval_rewards/len_reward/var": 0.07147407531738281, "eval_rewards/missing_response_penalty": -0.11666666716337204, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.10340021550655365, "eval_runtime": 172.2124, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.812279999256134, "eval_values/max": 1.9834823608398438, "eval_values/median": -0.77294921875, "eval_values/min": -2.0000133514404297, "eval_values/p25": -0.77294921875, "eval_values/p75": -0.77294921875, "eval_values/var": 0.12297380715608597, "step": 9 }, { "accuracy_reward": 0.7760416865348816, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.1747109591960907, "adam_stats/lr_effective_max": 6.146151281427592e-05, "adam_stats/lr_effective_mean": -6.28145313541495e-11, "adam_stats/lr_effective_min": -6.176823808345944e-05, "adam_stats/m_t_max": 0.0044099693186581135, "adam_stats/m_t_mean": -3.2349935230602256e-11, "adam_stats/m_t_min": -0.003992704674601555, "adam_stats/v_t_max": 2.233444683952257e-05, "adam_stats/v_t_mean": 1.3899149843218939e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7760416865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.1747109591960907, "all_logprobs": -0.017746614292263985, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -8.9375, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.02617828920483589, "answer_logprob_reward": 2.9540839195251465, "answer_logprob_reward/correct": 3.0880846977233887, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.06667309999465942, "answer_logprob_reward/incorrect": 2.4897546768188477, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 2.4849066734313965, "answer_logprob_reward/incorrect/min": 1.3862943649291992, "answer_logprob_reward/incorrect/p25": 2.079441547393799, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.43310627341270447, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.3862943649291992, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.20944732427597046, "clip_ratio": 0.0, "completion_length": 571.4896240234375, "completion_length/correct": 498.16778564453125, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 453.0, "completion_length/correct/min": 160.0, "completion_length/correct/p25": 347.0, "completion_length/correct/p75": 618.0, "completion_length/correct/var": 42527.9140625, "completion_length/incorrect": 825.5581665039062, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 1024.0, "completion_length/incorrect/min": 289.0, "completion_length/incorrect/p25": 640.5, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 60186.01953125, "completion_length/max": 1024.0, "completion_length/median": 501.0, "completion_length/min": 160.0, "completion_length/p25": 381.25, "completion_length/p75": 755.0, "completion_length/var": 64914.48046875, "epoch": 0.25316455696202533, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 1145748.0, "grad_norm": 0.1811712235212326, "kl": 0.00775146484375, "learning_rate": 9.292243968009332e-06, "len_reward": 0.09788952767848969, "len_reward/correct": 0.18229344487190247, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.2094140201807022, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.07336561381816864, "len_reward/correct/p75": 0.3368038833141327, "len_reward/correct/var": 0.04961439594626427, "len_reward/incorrect": -0.19457988440990448, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.16489103436470032, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.344448059797287, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03579093515872955, "len_reward/max": 0.5, "len_reward/median": 0.1380111575126648, "len_reward/min": -0.5, "len_reward/p25": -0.0582333579659462, "len_reward/p75": 0.2976534962654114, "len_reward/var": 0.07112973183393478, "loss": -0.7757, "mean_logprobs": -0.017333984375, "mean_logprobs/var": 0.00017452239990234375, "missing_response_penalty": -0.1145833358168602, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.5116279125213623, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": -1.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.25581395626068115, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.10198517143726349, "non_smoothed_logprobs/min": -20.1875, "num_completions/total": 1920, "policy_loss": -0.7757319808006287, "policy_loss/max": 0.000476124812848866, "policy_loss/median": -0.9996414184570312, "policy_loss/min": -0.9998821020126343, "policy_loss/p25": -0.9997469186782837, "policy_loss/p75": -0.9994195103645325, "policy_loss/var": 0.17468826472759247, "pred_value_log_probs": -0.6530084609985352, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.69921875, "pred_value_log_probs/min": -1.28125, "pred_value_log_probs/p25": -0.69921875, "pred_value_log_probs/p75": -0.69921875, "pred_value_log_probs/var": 0.0391046479344368, "reward": 0.7760416865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.1747109591960907, "rewards/accuracy_reward": 0.7760416865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.1747109591960907, "rewards/answer_logprob_reward": 2.9540839195251465, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.3862943649291992, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.20944732427597046, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.09788952767848969, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.1380111575126648, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.0582333579659462, "rewards/len_reward/p75": 0.2976534962654114, "rewards/len_reward/var": 0.07112973183393478, "rewards/missing_response_penalty": -0.1145833358168602, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.10198517143726349, "smoothed_logprobs/min": -1.0, "step": 10, "value_loss": 2.8359375, "values": -0.7617847323417664, "values/max": 2.0, "values/median": -0.77294921875, "values/min": -1.9896049499511719, "values/p25": -0.77294921875, "values/p75": -0.77294921875, "values/var": 0.22872459888458252 }, { "adam_stats/lr_effective_max": 6.146151281427592e-05, "adam_stats/lr_effective_mean": -6.28145313541495e-11, "adam_stats/lr_effective_min": -6.176823808345944e-05, "adam_stats/m_t_max": 0.0044099693186581135, "adam_stats/m_t_mean": -3.2349935230602256e-11, "adam_stats/m_t_min": -0.003992704674601555, "adam_stats/v_t_max": 2.233444683952257e-05, "adam_stats/v_t_mean": 1.3899149843218939e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.25316455696202533, "eval_accuracy_reward": 0.6200000047683716, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.2363879680633545, "eval_advantages": 0.6200000047683716, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.2363879680633545, "eval_all_logprobs": -0.017270084470510483, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -7.53125, "eval_all_logprobs/p25": -1.1920928955078125e-07, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.02553660236299038, "eval_answer_logprob_reward": 3.0562992095947266, "eval_answer_logprob_reward/correct": 3.1368906497955322, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.525728702545166, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.05037706345319748, "eval_answer_logprob_reward/incorrect": 2.9248080253601074, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.6094379425048828, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.2635025382041931, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.6094379425048828, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.14138683676719666, "eval_completion_length": 558.9967041015625, "eval_completion_length/correct": 473.81182861328125, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 434.0, "eval_completion_length/correct/min": 104.0, "eval_completion_length/correct/p25": 319.75, "eval_completion_length/correct/p75": 600.25, "eval_completion_length/correct/var": 41081.31640625, "eval_completion_length/incorrect": 697.9824829101562, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 662.0, "eval_completion_length/incorrect/min": 167.0, "eval_completion_length/incorrect/p25": 442.0, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 74562.5859375, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 490.0, "eval_completion_length/min": 104.0, "eval_completion_length/p25": 351.75, "eval_completion_length/p75": 730.25, "eval_completion_length/var": 65476.453125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 1798678.0, "eval_len_reward": 0.047263771295547485, "eval_len_reward/correct": 0.1697932779788971, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.17877094447612762, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.04317869991064072, "eval_len_reward/correct/p75": 0.3401416540145874, "eval_len_reward/correct/var": 0.0487174354493618, "eval_len_reward/incorrect": -0.1526528149843216, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.07299452275037766, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.2952413558959961, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03232886269688606, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.08298538625240326, "eval_len_reward/p75": 0.24518707394599915, "eval_len_reward/var": 0.06693842262029648, "eval_mean_logprobs": -0.0174560546875, "eval_mean_logprobs/var": 0.0001621246337890625, "eval_missing_response_penalty": -0.10000000149011612, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.2631579041481018, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.1956217885017395, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.09030100703239441, "eval_num_completions/total": 3000, "eval_pred_value_log_probs": -0.6529466509819031, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.703125, "eval_pred_value_log_probs/min": -1.0234375, "eval_pred_value_log_probs/p25": -0.703125, "eval_pred_value_log_probs/p75": -0.703125, "eval_pred_value_log_probs/var": 0.028898609802126884, "eval_reward": 0.6200000047683716, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.2363879680633545, "eval_rewards/accuracy_reward": 0.6200000047683716, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.2363879680633545, "eval_rewards/answer_logprob_reward": 3.0562992095947266, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.6094379425048828, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.14138683676719666, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.047263771295547485, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.08298538625240326, "eval_rewards/len_reward/p75": 0.24518707394599915, "eval_rewards/len_reward/var": 0.06693842262029648, "eval_rewards/missing_response_penalty": -0.10000000149011612, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.09030100703239441, "eval_runtime": 169.9197, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8177988529205322, "eval_values/max": 1.9912681579589844, "eval_values/median": -0.77294921875, "eval_values/min": -2.000148296356201, "eval_values/p25": -0.77294921875, "eval_values/p75": -0.77294921875, "eval_values/var": 0.13172577321529388, "step": 10 }, { "accuracy_reward": 0.765625, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.1803828477859497, "adam_stats/lr_effective_max": 6.084773485781625e-05, "adam_stats/lr_effective_mean": 3.8493843529385785e-11, "adam_stats/lr_effective_min": -6.085845598136075e-05, "adam_stats/m_t_max": 0.004168862476944923, "adam_stats/m_t_mean": -3.178726379338137e-11, "adam_stats/m_t_min": -0.003922298084944487, "adam_stats/v_t_max": 2.2330892534228042e-05, "adam_stats/v_t_mean": 1.4039654855357853e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.765625, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.1803828477859497, "all_logprobs": -0.018302688375115395, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -6.59375, "all_logprobs/p25": -2.384185791015625e-07, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.027053385972976685, "answer_logprob_reward": 2.915700912475586, "answer_logprob_reward/correct": 3.0188896656036377, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.09472610056400299, "answer_logprob_reward/incorrect": 2.578617811203003, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.2321436405181885, "answer_logprob_reward/incorrect/p25": 2.079441547393799, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.5363913178443909, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.2321436405181885, "answer_logprob_reward/p25": 2.4849066734313965, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.23094026744365692, "clip_ratio": 0.0, "completion_length": 548.84375, "completion_length/correct": 481.9727783203125, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 452.0, "completion_length/correct/min": 141.0, "completion_length/correct/p25": 315.0, "completion_length/correct/p75": 613.5, "completion_length/correct/var": 43214.8671875, "completion_length/incorrect": 767.2888793945312, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 864.0, "completion_length/incorrect/min": 261.0, "completion_length/incorrect/p25": 560.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 74908.5703125, "completion_length/max": 1024.0, "completion_length/median": 490.0, "completion_length/min": 141.0, "completion_length/p25": 334.0, "completion_length/p75": 717.25, "completion_length/var": 64973.890625, "epoch": 0.27848101265822783, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 1251126.0, "grad_norm": 0.23812517523765564, "kl": 0.0096435546875, "learning_rate": 9.045084971874738e-06, "len_reward": 0.056315742433071136, "len_reward/correct": 0.1266818791627884, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.20374064147472382, "len_reward/correct/min": -0.5, "len_reward/correct/p25": -0.09587077796459198, "len_reward/correct/p75": 0.3539902865886688, "len_reward/correct/var": 0.07949673384428024, "len_reward/incorrect": -0.1735469400882721, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.08390804380178452, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.34236451983451843, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.037365350872278214, "len_reward/max": 0.5, "len_reward/median": 0.0015723269898444414, "len_reward/min": -0.5, "len_reward/p25": -0.14850373566150665, "len_reward/p75": 0.31304460763931274, "len_reward/var": 0.08563410490751266, "loss": -0.7652, "mean_logprobs": -0.01708984375, "mean_logprobs/var": 0.00019168853759765625, "missing_response_penalty": -0.09375, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.4000000059604645, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.2454545497894287, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.08540575951337814, "non_smoothed_logprobs/min": -19.0, "num_completions/total": 2112, "policy_loss": -0.7652397155761719, "policy_loss/max": 0.0006840777932666242, "policy_loss/median": -0.999577522277832, "policy_loss/min": -0.9998697638511658, "policy_loss/p25": -0.999678373336792, "policy_loss/p75": -0.9991584420204163, "policy_loss/var": 0.18036982417106628, "pred_value_log_probs": -0.6432632207870483, "pred_value_log_probs/max": -0.0003204345703125, "pred_value_log_probs/median": -0.703125, "pred_value_log_probs/min": -1.3515625, "pred_value_log_probs/p25": -0.703125, "pred_value_log_probs/p75": -0.703125, "pred_value_log_probs/var": 0.03457346186041832, "reward": 0.765625, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.1803828477859497, "rewards/accuracy_reward": 0.765625, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.1803828477859497, "rewards/answer_logprob_reward": 2.915700912475586, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.2321436405181885, "rewards/answer_logprob_reward/p25": 2.4849066734313965, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.23094026744365692, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.056315742433071136, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.0015723269898444414, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.14850373566150665, "rewards/len_reward/p75": 0.31304460763931274, "rewards/len_reward/var": 0.08563410490751266, "rewards/missing_response_penalty": -0.09375, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.08540575951337814, "smoothed_logprobs/min": -1.0, "step": 11, "value_loss": 2.7578125, "values": -0.8260021209716797, "values/max": 1.342041015625, "values/median": -0.77294921875, "values/min": -1.9998247623443604, "values/p25": -0.77294921875, "values/p75": -0.77294921875, "values/var": 0.20274372398853302 }, { "adam_stats/lr_effective_max": 6.084773485781625e-05, "adam_stats/lr_effective_mean": 3.8493843529385785e-11, "adam_stats/lr_effective_min": -6.085845598136075e-05, "adam_stats/m_t_max": 0.004168862476944923, "adam_stats/m_t_mean": -3.178726379338137e-11, "adam_stats/m_t_min": -0.003922298084944487, "adam_stats/v_t_max": 2.2330892534228042e-05, "adam_stats/v_t_mean": 1.4039654855357853e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.27848101265822783, "eval_accuracy_reward": 0.6600000262260437, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.22515049576759338, "eval_advantages": 0.6600000262260437, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.22515049576759338, "eval_all_logprobs": -0.015983078628778458, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -7.0, "eval_all_logprobs/p25": -1.1920928955078125e-07, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.024440567940473557, "eval_answer_logprob_reward": 3.0050878524780273, "eval_answer_logprob_reward/correct": 3.1009719371795654, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.08625704050064087, "eval_answer_logprob_reward/incorrect": 2.8189597129821777, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.4271163940429688, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.36717504262924194, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.4271163940429688, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.1987670212984085, "eval_completion_length": 563.7766723632812, "eval_completion_length/correct": 480.99493408203125, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 439.0, "eval_completion_length/correct/min": 103.0, "eval_completion_length/correct/p25": 319.0, "eval_completion_length/correct/p75": 629.0, "eval_completion_length/correct/var": 41770.5625, "eval_completion_length/incorrect": 724.4706420898438, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 722.0, "eval_completion_length/incorrect/min": 203.0, "eval_completion_length/incorrect/p25": 480.25, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 77053.21875, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 508.0, "eval_completion_length/min": 103.0, "eval_completion_length/p25": 351.25, "eval_completion_length/p75": 733.25, "eval_completion_length/var": 66896.09375, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 1967811.0, "eval_len_reward": 0.057859741151332855, "eval_len_reward/correct": 0.16566860675811768, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.18490415811538696, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.004859698936343193, "eval_len_reward/correct/p75": 0.3304007947444916, "eval_len_reward/correct/var": 0.04873096942901611, "eval_len_reward/incorrect": -0.15141631662845612, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.0882587879896164, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.27317923307418823, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.031230084598064423, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.09351489692926407, "eval_len_reward/p75": 0.25090155005455017, "eval_len_reward/var": 0.06529360264539719, "eval_mean_logprobs": -0.01611328125, "eval_mean_logprobs/var": 0.00013828277587890625, "eval_missing_response_penalty": -0.10999999940395355, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.3235294222831726, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.2210250347852707, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.09822742640972137, "eval_num_completions/total": 3300, "eval_pred_value_log_probs": -0.6458382606506348, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.6953125, "eval_pred_value_log_probs/min": -1.0703125, "eval_pred_value_log_probs/p25": -0.6953125, "eval_pred_value_log_probs/p75": -0.6953125, "eval_pred_value_log_probs/var": 0.028838587924838066, "eval_reward": 0.6600000262260437, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.22515049576759338, "eval_rewards/accuracy_reward": 0.6600000262260437, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.22515049576759338, "eval_rewards/answer_logprob_reward": 3.0050878524780273, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.4271163940429688, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.1987670212984085, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.057859741151332855, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.09351489692926407, "eval_rewards/len_reward/p75": 0.25090155005455017, "eval_rewards/len_reward/var": 0.06529360264539719, "eval_rewards/missing_response_penalty": -0.10999999940395355, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.09822742640972137, "eval_runtime": 165.0613, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8255552649497986, "eval_values/max": 1.9914283752441406, "eval_values/median": -0.78125, "eval_values/min": -2.0001308917999268, "eval_values/p25": -0.78125, "eval_values/p75": -0.78125, "eval_values/var": 0.13591179251670837, "step": 11 }, { "accuracy_reward": 0.8020833730697632, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.15957680344581604, "adam_stats/lr_effective_max": 5.870972745469771e-05, "adam_stats/lr_effective_mean": -8.81928488571937e-11, "adam_stats/lr_effective_min": -5.807427078252658e-05, "adam_stats/m_t_max": 0.0037859270814806223, "adam_stats/m_t_mean": -7.138125594080558e-12, "adam_stats/m_t_min": -0.0038272289093583822, "adam_stats/v_t_max": 2.269002652610652e-05, "adam_stats/v_t_mean": 1.4284688883148267e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.8020833730697632, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.15957680344581604, "all_logprobs": -0.014600125141441822, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -6.75, "all_logprobs/p25": -1.1920928955078125e-07, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.020908931270241737, "answer_logprob_reward": 3.0444083213806152, "answer_logprob_reward/correct": 3.0883731842041016, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.06903017312288284, "answer_logprob_reward/incorrect": 2.8662333488464355, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 2.079441547393799, "answer_logprob_reward/incorrect/p25": 2.4849066734313965, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.2063932716846466, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 2.079441547393799, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.10315284132957458, "clip_ratio": 0.0, "completion_length": 543.3385620117188, "completion_length/correct": 488.8571472167969, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 474.0, "completion_length/correct/min": 160.0, "completion_length/correct/p25": 329.25, "completion_length/correct/p75": 601.0, "completion_length/correct/var": 33187.94140625, "completion_length/incorrect": 764.131591796875, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 741.0, "completion_length/incorrect/min": 243.0, "completion_length/incorrect/p25": 543.25, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 62576.92578125, "completion_length/max": 1024.0, "completion_length/median": 504.0, "completion_length/min": 160.0, "completion_length/p25": 372.75, "completion_length/p75": 662.0, "completion_length/var": 50799.4296875, "epoch": 0.3037974683544304, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 1355447.0, "grad_norm": 0.2768768072128296, "kl": 0.010009765625, "learning_rate": 8.765357330018056e-06, "len_reward": 0.07785402238368988, "len_reward/correct": 0.14999371767044067, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.17182274162769318, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.0, "len_reward/correct/p75": 0.3500424921512604, "len_reward/correct/var": 0.06163345277309418, "len_reward/incorrect": -0.21450167894363403, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.20623500645160675, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.3933675289154053, "len_reward/incorrect/p75": -0.006477926857769489, "len_reward/incorrect/var": 0.037543874233961105, "len_reward/max": 0.5, "len_reward/median": 0.07889688014984131, "len_reward/min": -0.5, "len_reward/p25": -0.08291997760534286, "len_reward/p75": 0.327726274728775, "len_reward/var": 0.07784508168697357, "loss": -0.8017, "mean_logprobs": -0.014404296875, "mean_logprobs/var": 0.00012302398681640625, "missing_response_penalty": -0.0729166716337204, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.3684210479259491, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.23897583782672882, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.06795375049114227, "non_smoothed_logprobs/min": -18.625, "num_completions/total": 2304, "policy_loss": -0.8016828298568726, "policy_loss/max": 0.0008261256152763963, "policy_loss/median": -0.9995740056037903, "policy_loss/min": -0.9999816417694092, "policy_loss/p25": -0.9996675252914429, "policy_loss/p75": -0.9993911981582642, "policy_loss/var": 0.1595751941204071, "pred_value_log_probs": -0.6363335251808167, "pred_value_log_probs/max": -0.0003261566162109375, "pred_value_log_probs/median": -0.6953125, "pred_value_log_probs/min": -1.109375, "pred_value_log_probs/p25": -0.6953125, "pred_value_log_probs/p75": -0.6953125, "pred_value_log_probs/var": 0.03291379660367966, "reward": 0.8020833730697632, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.15957680344581604, "rewards/accuracy_reward": 0.8020833730697632, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.15957680344581604, "rewards/answer_logprob_reward": 3.0444083213806152, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 2.079441547393799, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.10315284132957458, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.07785402238368988, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.07889688014984131, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.08291997760534286, "rewards/len_reward/p75": 0.327726274728775, "rewards/len_reward/var": 0.07784508168697357, "rewards/missing_response_penalty": -0.0729166716337204, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.06795375049114227, "smoothed_logprobs/min": -1.0, "step": 12, "value_loss": 2.765625, "values": -0.7823073267936707, "values/max": 1.7805557250976562, "values/median": -0.78125, "values/min": -1.9998421669006348, "values/p25": -0.78125, "values/p75": -0.78125, "values/var": 0.24507829546928406 }, { "adam_stats/lr_effective_max": 5.870972745469771e-05, "adam_stats/lr_effective_mean": -8.81928488571937e-11, "adam_stats/lr_effective_min": -5.807427078252658e-05, "adam_stats/m_t_max": 0.0037859270814806223, "adam_stats/m_t_mean": -7.138125594080558e-12, "adam_stats/m_t_min": -0.0038272289093583822, "adam_stats/v_t_max": 2.269002652610652e-05, "adam_stats/v_t_mean": 1.4284688883148267e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.3037974683544304, "eval_accuracy_reward": 0.6600000262260437, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.22515052556991577, "eval_advantages": 0.6600000262260437, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.22515052556991577, "eval_all_logprobs": -0.01577628403902054, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -10.125, "eval_all_logprobs/p25": -1.1920928955078125e-07, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.024152709171175957, "eval_answer_logprob_reward": 3.056529998779297, "eval_answer_logprob_reward/correct": 3.141859292984009, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.525728702545166, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.047693025320768356, "eval_answer_logprob_reward/incorrect": 2.890890598297119, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.8325815200805664, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.27481406927108765, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.8325815200805664, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.13843455910682678, "eval_completion_length": 568.433349609375, "eval_completion_length/correct": 497.3737487792969, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 461.0, "eval_completion_length/correct/min": 115.0, "eval_completion_length/correct/p25": 322.75, "eval_completion_length/correct/p75": 633.5, "eval_completion_length/correct/var": 48618.09765625, "eval_completion_length/incorrect": 706.37255859375, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 716.0, "eval_completion_length/incorrect/min": 172.0, "eval_completion_length/incorrect/p25": 452.25, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 82445.921875, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 524.0, "eval_completion_length/min": 115.0, "eval_completion_length/p25": 346.5, "eval_completion_length/p75": 750.75, "eval_completion_length/var": 69716.96875, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 2138341.0, "eval_len_reward": 0.030194932594895363, "eval_len_reward/correct": 0.13411900401115417, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.13924558460712433, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": -0.016798434779047966, "eval_len_reward/correct/p75": 0.3000715374946594, "eval_len_reward/correct/var": 0.05324889346957207, "eval_len_reward/incorrect": -0.1715400218963623, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.11954765766859055, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.3578011691570282, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03252788633108139, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.12422685325145721, "eval_len_reward/p75": 0.23109574615955353, "eval_len_reward/var": 0.06710663437843323, "eval_mean_logprobs": -0.0150146484375, "eval_mean_logprobs/var": 0.00012302398681640625, "eval_missing_response_penalty": -0.09666667133569717, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.28431373834609985, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.20549407601356506, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.0876142606139183, "eval_num_completions/total": 3600, "eval_pred_value_log_probs": -0.6469375491142273, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.6953125, "eval_pred_value_log_probs/min": -1.078125, "eval_pred_value_log_probs/p25": -0.6953125, "eval_pred_value_log_probs/p75": -0.6953125, "eval_pred_value_log_probs/var": 0.028731336817145348, "eval_reward": 0.6600000262260437, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.22515052556991577, "eval_rewards/accuracy_reward": 0.6600000262260437, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.22515052556991577, "eval_rewards/answer_logprob_reward": 3.056529998779297, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.8325815200805664, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.13843455910682678, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.030194932594895363, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.12422685325145721, "eval_rewards/len_reward/p75": 0.23109574615955353, "eval_rewards/len_reward/var": 0.06710663437843323, "eval_rewards/missing_response_penalty": -0.09666667133569717, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.0876142606139183, "eval_runtime": 175.9355, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8294759392738342, "eval_values/max": 1.9914474487304688, "eval_values/median": -0.78515625, "eval_values/min": -2.000096082687378, "eval_values/p25": -0.78515625, "eval_values/p75": -0.78515625, "eval_values/var": 0.13602778315544128, "step": 12 }, { "accuracy_reward": 0.7552083730697632, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.18583661317825317, "adam_stats/lr_effective_max": 5.697783126379363e-05, "adam_stats/lr_effective_mean": -2.277365795944064e-10, "adam_stats/lr_effective_min": -5.703235729015432e-05, "adam_stats/m_t_max": 0.0037888039369136095, "adam_stats/m_t_mean": -8.236701251607137e-12, "adam_stats/m_t_min": -0.003288755426183343, "adam_stats/v_t_max": 2.2761514628655277e-05, "adam_stats/v_t_mean": 1.4496625471219216e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7552083730697632, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.18583661317825317, "all_logprobs": -0.013648364692926407, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -7.5625, "all_logprobs/p25": -1.1920928955078125e-07, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.021080132573843002, "answer_logprob_reward": 2.9505443572998047, "answer_logprob_reward/correct": 3.039424419403076, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 1.7917594909667969, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.10409815609455109, "answer_logprob_reward/incorrect": 2.676339864730835, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.3862943649291992, "answer_logprob_reward/incorrect/p25": 2.2821741104125977, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.41298890113830566, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.3862943649291992, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.20244458317756653, "clip_ratio": 0.0, "completion_length": 550.9739990234375, "completion_length/correct": 476.0689697265625, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 440.0, "completion_length/correct/min": 156.0, "completion_length/correct/p25": 317.0, "completion_length/correct/p75": 609.0, "completion_length/correct/var": 40123.453125, "completion_length/incorrect": 782.0637817382812, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 906.0, "completion_length/incorrect/min": 238.0, "completion_length/incorrect/p25": 510.5, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 74556.109375, "completion_length/max": 1024.0, "completion_length/median": 485.0, "completion_length/min": 156.0, "completion_length/p25": 344.25, "completion_length/p75": 719.5, "completion_length/var": 65606.4765625, "epoch": 0.3291139240506329, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 1461234.0, "grad_norm": 0.2585687041282654, "kl": 0.01025390625, "learning_rate": 8.455313244934324e-06, "len_reward": 0.11482175439596176, "len_reward/correct": 0.20802628993988037, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.22021225094795227, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.11475150287151337, "len_reward/correct/p75": 0.3654124438762665, "len_reward/correct/var": 0.042087990790605545, "len_reward/incorrect": -0.172724187374115, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.1230769231915474, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.3250735104084015, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03574460372328758, "len_reward/max": 0.5, "len_reward/median": 0.17319704592227936, "len_reward/min": -0.5, "len_reward/p25": 0.0, "len_reward/p75": 0.2954641878604889, "len_reward/var": 0.06728082150220871, "loss": -0.7548, "mean_logprobs": -0.0135498046875, "mean_logprobs/var": 8.96453857421875e-05, "missing_response_penalty": -0.1041666716337204, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.42553189396858215, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.24976873397827148, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.09380454570055008, "non_smoothed_logprobs/min": -18.875, "num_completions/total": 2496, "policy_loss": -0.7547990679740906, "policy_loss/max": 0.0008481450495310128, "policy_loss/median": -0.9995503425598145, "policy_loss/min": -0.9998537302017212, "policy_loss/p25": -0.9996625185012817, "policy_loss/p75": -0.9991025924682617, "policy_loss/var": 0.18582956492900848, "pred_value_log_probs": -0.6478489637374878, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.6953125, "pred_value_log_probs/min": -1.3671875, "pred_value_log_probs/p25": -0.6953125, "pred_value_log_probs/p75": -0.6953125, "pred_value_log_probs/var": 0.06093765050172806, "reward": 0.7552083730697632, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.18583661317825317, "rewards/accuracy_reward": 0.7552083730697632, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.18583661317825317, "rewards/answer_logprob_reward": 2.9505443572998047, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.3862943649291992, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.20244458317756653, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.11482175439596176, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.17319704592227936, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": 0.0, "rewards/len_reward/p75": 0.2954641878604889, "rewards/len_reward/var": 0.06728082150220871, "rewards/missing_response_penalty": -0.1041666716337204, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.09380454570055008, "smoothed_logprobs/min": -1.0, "step": 13, "value_loss": 3.1015625, "values": -0.8192496299743652, "values/max": 1.9883270263671875, "values/median": -0.78515625, "values/min": -2.0010595321655273, "values/p25": -0.78515625, "values/p75": -0.78515625, "values/var": 0.2388850301504135 }, { "adam_stats/lr_effective_max": 5.697783126379363e-05, "adam_stats/lr_effective_mean": -2.277365795944064e-10, "adam_stats/lr_effective_min": -5.703235729015432e-05, "adam_stats/m_t_max": 0.0037888039369136095, "adam_stats/m_t_mean": -8.236701251607137e-12, "adam_stats/m_t_min": -0.003288755426183343, "adam_stats/v_t_max": 2.2761514628655277e-05, "adam_stats/v_t_mean": 1.4496625471219216e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.3291139240506329, "eval_accuracy_reward": 0.6333333253860474, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.23299884796142578, "eval_advantages": 0.6333333253860474, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.23299884796142578, "eval_all_logprobs": -0.015315653756260872, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -9.875, "eval_all_logprobs/p25": -1.1920928955078125e-07, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.024531463161110878, "eval_answer_logprob_reward": 3.0196218490600586, "eval_answer_logprob_reward/correct": 3.109705924987793, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.07587447762489319, "eval_answer_logprob_reward/incorrect": 2.8640220165252686, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.6094379425048828, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.32031023502349854, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.6094379425048828, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.1787933111190796, "eval_completion_length": 557.0900268554688, "eval_completion_length/correct": 483.5684509277344, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 453.0, "eval_completion_length/correct/min": 121.0, "eval_completion_length/correct/p25": 328.0, "eval_completion_length/correct/p75": 615.0, "eval_completion_length/correct/var": 44640.59765625, "eval_completion_length/incorrect": 684.081787109375, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 665.0, "eval_completion_length/incorrect/min": 192.0, "eval_completion_length/incorrect/p25": 433.5, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 75831.328125, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 512.0, "eval_completion_length/min": 121.0, "eval_completion_length/p25": 345.75, "eval_completion_length/p75": 743.0, "eval_completion_length/var": 65229.69921875, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 2305468.0, "eval_len_reward": 0.020261196419596672, "eval_len_reward/correct": 0.1251496970653534, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.1495969593524933, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": -0.024701278656721115, "eval_len_reward/correct/p75": 0.3076336979866028, "eval_len_reward/correct/var": 0.060734037309885025, "eval_len_reward/incorrect": -0.16090983152389526, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.08825862407684326, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.29965516924858093, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.033284444361925125, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.1348545253276825, "eval_len_reward/p75": 0.21349772810935974, "eval_len_reward/var": 0.06959052383899689, "eval_mean_logprobs": -0.0152587890625, "eval_mean_logprobs/var": 0.0001125335693359375, "eval_missing_response_penalty": -0.10000000149011612, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.27272725105285645, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.20016680657863617, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.09030100703239441, "eval_num_completions/total": 3900, "eval_pred_value_log_probs": -0.6456806659698486, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.6953125, "eval_pred_value_log_probs/min": -0.9921875, "eval_pred_value_log_probs/p25": -0.6953125, "eval_pred_value_log_probs/p75": -0.6953125, "eval_pred_value_log_probs/var": 0.02859373576939106, "eval_reward": 0.6333333253860474, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.23299884796142578, "eval_rewards/accuracy_reward": 0.6333333253860474, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.23299884796142578, "eval_rewards/answer_logprob_reward": 3.0196218490600586, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.6094379425048828, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.1787933111190796, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.020261196419596672, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.1348545253276825, "eval_rewards/len_reward/p75": 0.21349772810935974, "eval_rewards/len_reward/var": 0.06959052383899689, "eval_rewards/missing_response_penalty": -0.10000000149011612, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.09030100703239441, "eval_runtime": 165.6153, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8315897583961487, "eval_values/max": 1.9913291931152344, "eval_values/median": -0.78662109375, "eval_values/min": -2.000145435333252, "eval_values/p25": -0.78662109375, "eval_values/p75": -0.78662109375, "eval_values/var": 0.1379442811012268, "step": 13 }, { "accuracy_reward": 0.78125, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.17179319262504578, "adam_stats/lr_effective_max": 5.388263889472e-05, "adam_stats/lr_effective_mean": -2.1835391827984552e-10, "adam_stats/lr_effective_min": -5.37708583578933e-05, "adam_stats/m_t_max": 0.003445042297244072, "adam_stats/m_t_mean": -2.166033671868739e-11, "adam_stats/m_t_min": -0.0031208600848913193, "adam_stats/v_t_max": 2.2922347852727398e-05, "adam_stats/v_t_mean": 1.4587817715946971e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.78125, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.17179319262504578, "all_logprobs": -0.01267217192798853, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -6.125, "all_logprobs/p25": -1.1920928955078125e-07, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.017811747267842293, "answer_logprob_reward": 3.0137572288513184, "answer_logprob_reward/correct": 3.1133601665496826, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.4849066734313965, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.04092986136674881, "answer_logprob_reward/incorrect": 2.658031940460205, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.7917594909667969, "answer_logprob_reward/incorrect/p25": 2.079441547393799, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.3449063003063202, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.7917594909667969, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.14158384501934052, "clip_ratio": 0.0, "completion_length": 541.9583740234375, "completion_length/correct": 475.7400207519531, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 441.0, "completion_length/correct/min": 161.0, "completion_length/correct/p25": 324.5, "completion_length/correct/p75": 563.75, "completion_length/correct/var": 42762.76171875, "completion_length/incorrect": 778.452392578125, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 895.0, "completion_length/incorrect/min": 261.0, "completion_length/incorrect/p25": 518.25, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 73976.6953125, "completion_length/max": 1024.0, "completion_length/median": 473.0, "completion_length/min": 161.0, "completion_length/p25": 339.75, "completion_length/p75": 705.25, "completion_length/var": 64981.4765625, "epoch": 0.35443037974683544, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 1565290.0, "grad_norm": 0.22427216172218323, "kl": 0.01043701171875, "learning_rate": 8.117449009293668e-06, "len_reward": 0.08795583248138428, "len_reward/correct": 0.17518459260463715, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.21158888936042786, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.03775645047426224, "len_reward/correct/p75": 0.3615916669368744, "len_reward/correct/var": 0.054440684616565704, "len_reward/incorrect": -0.22357545793056488, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.20162558555603027, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.47105860710144043, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.046373918652534485, "len_reward/max": 0.5, "len_reward/median": 0.10566334426403046, "len_reward/min": -0.5, "len_reward/p25": -0.0661887377500534, "len_reward/p75": 0.3085714280605316, "len_reward/var": 0.07974081486463547, "loss": -0.7808, "mean_logprobs": -0.01275634765625, "mean_logprobs/var": 7.867813110351562e-05, "missing_response_penalty": -0.0989583358168602, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.4523809552192688, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.2537747025489807, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.08963242173194885, "non_smoothed_logprobs/min": -18.0, "num_completions/total": 2688, "policy_loss": -0.7808328866958618, "policy_loss/max": 0.0009762375266291201, "policy_loss/median": -0.9995401501655579, "policy_loss/min": -0.9998199343681335, "policy_loss/p25": -0.9996650815010071, "policy_loss/p75": -0.9993335604667664, "policy_loss/var": 0.17179079353809357, "pred_value_log_probs": -0.6513863801956177, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.6953125, "pred_value_log_probs/min": -1.1015625, "pred_value_log_probs/p25": -0.6953125, "pred_value_log_probs/p75": -0.6953125, "pred_value_log_probs/var": 0.03190627694129944, "reward": 0.78125, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.17179319262504578, "rewards/accuracy_reward": 0.78125, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.17179319262504578, "rewards/answer_logprob_reward": 3.0137572288513184, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.7917594909667969, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.14158384501934052, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.08795583248138428, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.10566334426403046, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.0661887377500534, "rewards/len_reward/p75": 0.3085714280605316, "rewards/len_reward/var": 0.07974081486463547, "rewards/missing_response_penalty": -0.0989583358168602, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.08963242173194885, "smoothed_logprobs/min": -1.0, "step": 14, "value_loss": 2.875, "values": -0.7671071887016296, "values/max": 1.9346656799316406, "values/median": -0.78662109375, "values/min": -1.9921550750732422, "values/p25": -0.78662109375, "values/p75": -0.78662109375, "values/var": 0.2054801881313324 }, { "adam_stats/lr_effective_max": 5.388263889472e-05, "adam_stats/lr_effective_mean": -2.1835391827984552e-10, "adam_stats/lr_effective_min": -5.37708583578933e-05, "adam_stats/m_t_max": 0.003445042297244072, "adam_stats/m_t_mean": -2.166033671868739e-11, "adam_stats/m_t_min": -0.0031208600848913193, "adam_stats/v_t_max": 2.2922347852727398e-05, "adam_stats/v_t_mean": 1.4587817715946971e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.35443037974683544, "eval_accuracy_reward": 0.6800000071525574, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.21832777559757233, "eval_advantages": 0.6800000071525574, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.21832777559757233, "eval_all_logprobs": -0.016139881685376167, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -7.90625, "eval_all_logprobs/p25": -1.1920928955078125e-07, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.02557363547384739, "eval_answer_logprob_reward": 3.0449326038360596, "eval_answer_logprob_reward/correct": 3.1273913383483887, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.06628566235303879, "eval_answer_logprob_reward/incorrect": 2.8697075843811035, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.4271163940429688, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.3330424129962921, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.4271163940429688, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.16531668603420258, "eval_completion_length": 548.9466552734375, "eval_completion_length/correct": 483.0833435058594, "eval_completion_length/correct/max": 1002.0, "eval_completion_length/correct/median": 452.0, "eval_completion_length/correct/min": 103.0, "eval_completion_length/correct/p25": 339.25, "eval_completion_length/correct/p75": 605.0, "eval_completion_length/correct/var": 40634.56640625, "eval_completion_length/incorrect": 688.90625, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 644.0, "eval_completion_length/incorrect/min": 200.0, "eval_completion_length/incorrect/p25": 424.0, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 80034.359375, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 511.0, "eval_completion_length/min": 103.0, "eval_completion_length/p25": 352.0, "eval_completion_length/p75": 686.5, "eval_completion_length/var": 62266.02734375, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 2470152.0, "eval_len_reward": 0.044923048466444016, "eval_len_reward/correct": 0.14118969440460205, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.15497076511383057, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.005123578943312168, "eval_len_reward/correct/p75": 0.30648672580718994, "eval_len_reward/correct/var": 0.05321267619729042, "eval_len_reward/incorrect": -0.15964357554912567, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.10513035953044891, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.2941738963127136, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03156324848532677, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.006738544441759586, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.11079559475183487, "eval_len_reward/p75": 0.22145964205265045, "eval_len_reward/var": 0.06591493636369705, "eval_mean_logprobs": -0.01544189453125, "eval_mean_logprobs/var": 0.0003566741943359375, "eval_missing_response_penalty": -0.09333333373069763, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.2916666865348816, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.20877192914485931, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.084905244410038, "eval_num_completions/total": 4200, "eval_pred_value_log_probs": -0.6421944499015808, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.69140625, "eval_pred_value_log_probs/min": -0.9375, "eval_pred_value_log_probs/p25": -0.69140625, "eval_pred_value_log_probs/p75": -0.69140625, "eval_pred_value_log_probs/var": 0.0282747782766819, "eval_reward": 0.6800000071525574, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.21832777559757233, "eval_rewards/accuracy_reward": 0.6800000071525574, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.21832777559757233, "eval_rewards/answer_logprob_reward": 3.0449326038360596, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.4271163940429688, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.16531668603420258, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.044923048466444016, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.006738544441759586, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.11079559475183487, "eval_rewards/len_reward/p75": 0.22145964205265045, "eval_rewards/len_reward/var": 0.06591493636369705, "eval_rewards/missing_response_penalty": -0.09333333373069763, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.084905244410038, "eval_runtime": 176.2593, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8357735276222229, "eval_values/max": 1.9913787841796875, "eval_values/median": -0.7900390625, "eval_values/min": -1.9919319152832031, "eval_values/p25": -0.7900390625, "eval_values/p75": -0.7900390625, "eval_values/var": 0.1369415521621704, "step": 14 }, { "accuracy_reward": 0.78125, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.17179320752620697, "adam_stats/lr_effective_max": 5.17590124218259e-05, "adam_stats/lr_effective_mean": -1.4752990495914275e-10, "adam_stats/lr_effective_min": -5.170961594558321e-05, "adam_stats/m_t_max": 0.004034376237541437, "adam_stats/m_t_mean": -2.961150075142527e-11, "adam_stats/m_t_min": -0.002845394890755415, "adam_stats/v_t_max": 2.2986629119259305e-05, "adam_stats/v_t_mean": 1.4717971849945166e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.78125, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.17179320752620697, "all_logprobs": -0.012538185343146324, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -10.625, "all_logprobs/p25": -1.1920928955078125e-07, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.019076447933912277, "answer_logprob_reward": 2.933018207550049, "answer_logprob_reward/correct": 3.085634469985962, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.4849066734313965, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.055891625583171844, "answer_logprob_reward/incorrect": 2.387960195541382, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 2.079441547393799, "answer_logprob_reward/incorrect/min": 1.5686159133911133, "answer_logprob_reward/incorrect/p25": 1.7917594909667969, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.38665327429771423, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.5686159133911133, "answer_logprob_reward/p25": 3.0047669410705566, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.21022039651870728, "clip_ratio": 0.0, "completion_length": 554.1979370117188, "completion_length/correct": 472.3266906738281, "completion_length/correct/max": 993.0, "completion_length/correct/median": 429.0, "completion_length/correct/min": 121.0, "completion_length/correct/p25": 320.0, "completion_length/correct/p75": 575.75, "completion_length/correct/var": 38527.4296875, "completion_length/incorrect": 846.5952758789062, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 1024.0, "completion_length/incorrect/min": 313.0, "completion_length/incorrect/p25": 709.5, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 60216.390625, "completion_length/max": 1024.0, "completion_length/median": 496.0, "completion_length/min": 121.0, "completion_length/p25": 345.75, "completion_length/p75": 758.25, "completion_length/var": 67045.7265625, "epoch": 0.379746835443038, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 1671696.0, "grad_norm": 0.23879557847976685, "kl": 0.010284423828125, "learning_rate": 7.754484907260513e-06, "len_reward": 0.07622233033180237, "len_reward/correct": 0.16397397220134735, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.1973039209842682, "len_reward/correct/min": -0.45950794219970703, "len_reward/correct/p25": 0.03084648586809635, "len_reward/correct/p75": 0.35621941089630127, "len_reward/correct/var": 0.05613316223025322, "len_reward/incorrect": -0.23717637360095978, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.2298821061849594, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.4051016569137573, "len_reward/incorrect/p75": -0.0940435603260994, "len_reward/incorrect/var": 0.03346768394112587, "len_reward/max": 0.5, "len_reward/median": 0.12388059496879578, "len_reward/min": -0.5, "len_reward/p25": -0.11832349002361298, "len_reward/p75": 0.30537301301956177, "len_reward/var": 0.07861914485692978, "loss": -0.7808, "mean_logprobs": -0.012451171875, "mean_logprobs/var": 8.869171142578125e-05, "missing_response_penalty": -0.125, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.5714285969734192, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": -1.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.25087106227874756, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.10994765162467957, "non_smoothed_logprobs/min": -17.6875, "num_completions/total": 2880, "policy_loss": -0.7808372378349304, "policy_loss/max": 0.0005718395113945007, "policy_loss/median": -0.9995436668395996, "policy_loss/min": -0.9998787045478821, "policy_loss/p25": -0.9996510744094849, "policy_loss/p75": -0.9992718696594238, "policy_loss/var": 0.17176659405231476, "pred_value_log_probs": -0.6482011079788208, "pred_value_log_probs/max": -0.00087738037109375, "pred_value_log_probs/median": -0.69140625, "pred_value_log_probs/min": -1.3203125, "pred_value_log_probs/p25": -0.69140625, "pred_value_log_probs/p75": -0.69140625, "pred_value_log_probs/var": 0.0415121465921402, "reward": 0.78125, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.17179320752620697, "rewards/accuracy_reward": 0.78125, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.17179320752620697, "rewards/answer_logprob_reward": 2.933018207550049, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.5686159133911133, "rewards/answer_logprob_reward/p25": 3.0047669410705566, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.21022039651870728, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.07622233033180237, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.12388059496879578, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.11832349002361298, "rewards/len_reward/p75": 0.30537301301956177, "rewards/len_reward/var": 0.07861914485692978, "rewards/missing_response_penalty": -0.125, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.10994765162467957, "smoothed_logprobs/min": -1.0, "step": 15, "value_loss": 2.6875, "values": -0.7949997782707214, "values/max": 1.22149658203125, "values/median": -0.7900390625, "values/min": -1.9999971389770508, "values/p25": -0.7900390625, "values/p75": -0.7900390625, "values/var": 0.19552254676818848 }, { "adam_stats/lr_effective_max": 5.17590124218259e-05, "adam_stats/lr_effective_mean": -1.4752990495914275e-10, "adam_stats/lr_effective_min": -5.170961594558321e-05, "adam_stats/m_t_max": 0.004034376237541437, "adam_stats/m_t_mean": -2.961150075142527e-11, "adam_stats/m_t_min": -0.002845394890755415, "adam_stats/v_t_max": 2.2986629119259305e-05, "adam_stats/v_t_mean": 1.4717971849945166e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.379746835443038, "eval_accuracy_reward": 0.6533333659172058, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.22724638879299164, "eval_advantages": 0.6533333659172058, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.22724638879299164, "eval_all_logprobs": -0.014098653569817543, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -7.375, "eval_all_logprobs/p25": -1.1920928955078125e-07, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.02137773297727108, "eval_answer_logprob_reward": 3.027865409851074, "eval_answer_logprob_reward/correct": 3.1033051013946533, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.08536802232265472, "eval_answer_logprob_reward/incorrect": 2.8856899738311768, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.272965669631958, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.34138745069503784, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.272965669631958, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.18403808772563934, "eval_completion_length": 546.8500366210938, "eval_completion_length/correct": 476.3520202636719, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 442.0, "eval_completion_length/correct/min": 102.0, "eval_completion_length/correct/p25": 305.0, "eval_completion_length/correct/p75": 609.0, "eval_completion_length/correct/var": 44579.4296875, "eval_completion_length/incorrect": 679.7115478515625, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 664.0, "eval_completion_length/incorrect/min": 161.0, "eval_completion_length/incorrect/p25": 398.75, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 86873.515625, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 484.0, "eval_completion_length/min": 102.0, "eval_completion_length/p25": 337.75, "eval_completion_length/p75": 736.25, "eval_completion_length/var": 68397.6484375, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 2634207.0, "eval_len_reward": 0.05325581505894661, "eval_len_reward/correct": 0.16086210310459137, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.19661563634872437, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.03216128051280975, "eval_len_reward/correct/p75": 0.3326359987258911, "eval_len_reward/correct/var": 0.047571588307619095, "eval_len_reward/incorrect": -0.14954069256782532, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.04630269482731819, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.2933231592178345, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03319944441318512, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.08192966878414154, "eval_len_reward/p75": 0.2583830654621124, "eval_len_reward/var": 0.0643567144870758, "eval_mean_logprobs": -0.01409912109375, "eval_mean_logprobs/var": 0.000118255615234375, "eval_missing_response_penalty": -0.09000000357627869, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.25961539149284363, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.1940814107656479, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.0821739137172699, "eval_num_completions/total": 4500, "eval_pred_value_log_probs": -0.6381974220275879, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.6875, "eval_pred_value_log_probs/min": -0.94921875, "eval_pred_value_log_probs/p25": -0.6875, "eval_pred_value_log_probs/p75": -0.6875, "eval_pred_value_log_probs/var": 0.027867738157510757, "eval_reward": 0.6533333659172058, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.22724638879299164, "eval_rewards/accuracy_reward": 0.6533333659172058, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.22724638879299164, "eval_rewards/answer_logprob_reward": 3.027865409851074, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.272965669631958, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.18403808772563934, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.05325581505894661, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.08192966878414154, "eval_rewards/len_reward/p75": 0.2583830654621124, "eval_rewards/len_reward/var": 0.0643567144870758, "eval_rewards/missing_response_penalty": -0.09000000357627869, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.0821739137172699, "eval_runtime": 169.6683, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8470864295959473, "eval_values/max": 1.9908599853515625, "eval_values/median": -0.8037109375, "eval_values/min": -1.9913673400878906, "eval_values/p25": -0.8037109375, "eval_values/p75": -0.8037109375, "eval_values/var": 0.13635355234146118, "step": 15 }, { "accuracy_reward": 0.7760416865348816, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.1747109591960907, "adam_stats/lr_effective_max": 4.9730995669960976e-05, "adam_stats/lr_effective_mean": 2.7487906048562216e-11, "adam_stats/lr_effective_min": -4.8396959755336866e-05, "adam_stats/m_t_max": 0.004705157596617937, "adam_stats/m_t_mean": -2.7418390474709398e-11, "adam_stats/m_t_min": -0.0033178571611642838, "adam_stats/v_t_max": 2.307903559994884e-05, "adam_stats/v_t_mean": 1.4827556500326966e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7760416865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.1747109591960907, "all_logprobs": -0.012429504655301571, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -6.5, "all_logprobs/p25": -1.1920928955078125e-07, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.018746089190244675, "answer_logprob_reward": 2.9532697200775146, "answer_logprob_reward/correct": 3.0818519592285156, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.08189358562231064, "answer_logprob_reward/incorrect": 2.507716655731201, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 2.4849066734313965, "answer_logprob_reward/incorrect/min": 1.7917594909667969, "answer_logprob_reward/incorrect/p25": 1.7917594909667969, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.39231982827186584, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.7917594909667969, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.2073162943124771, "clip_ratio": 0.0, "completion_length": 547.2864990234375, "completion_length/correct": 464.0603942871094, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 408.0, "completion_length/correct/min": 131.0, "completion_length/correct/p25": 322.0, "completion_length/correct/p75": 553.0, "completion_length/correct/var": 41388.8671875, "completion_length/incorrect": 835.6744384765625, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 1024.0, "completion_length/incorrect/min": 267.0, "completion_length/incorrect/p25": 634.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 65544.5546875, "completion_length/max": 1024.0, "completion_length/median": 454.0, "completion_length/min": 131.0, "completion_length/p25": 346.25, "completion_length/p75": 718.75, "completion_length/var": 70610.9453125, "epoch": 0.4050632911392405, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 1776775.0, "grad_norm": 0.23507139086723328, "kl": 0.0107421875, "learning_rate": 7.369343312364994e-06, "len_reward": 0.09303848445415497, "len_reward/correct": 0.19003821909427643, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.23529411852359772, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.09877175092697144, "len_reward/correct/p75": 0.3461134433746338, "len_reward/correct/var": 0.04861581698060036, "len_reward/incorrect": -0.2430768609046936, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.26805612444877625, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.3951869010925293, "len_reward/incorrect/p75": -0.06606549769639969, "len_reward/incorrect/var": 0.032611481845378876, "len_reward/max": 0.5, "len_reward/median": 0.1441424936056137, "len_reward/min": -0.5, "len_reward/p25": -0.0984899252653122, "len_reward/p75": 0.3025240898132324, "len_reward/var": 0.07761579006910324, "loss": -0.7756, "mean_logprobs": -0.01239013671875, "mean_logprobs/var": 8.7738037109375e-05, "missing_response_penalty": -0.1197916716337204, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.5348837375640869, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": -1.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.2547065317630768, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.10599368065595627, "non_smoothed_logprobs/min": -17.3125, "num_completions/total": 3072, "policy_loss": -0.7756123542785645, "policy_loss/max": 0.0010877820895984769, "policy_loss/median": -0.9995073676109314, "policy_loss/min": -0.9998544454574585, "policy_loss/p25": -0.9996453523635864, "policy_loss/p75": -0.999257504940033, "policy_loss/var": 0.17469975352287292, "pred_value_log_probs": -0.6218863129615784, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.6875, "pred_value_log_probs/min": -1.2734375, "pred_value_log_probs/p25": -0.6875, "pred_value_log_probs/p75": -0.6875, "pred_value_log_probs/var": 0.055848974734544754, "reward": 0.7760416865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.1747109591960907, "rewards/accuracy_reward": 0.7760416865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.1747109591960907, "rewards/answer_logprob_reward": 2.9532697200775146, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.7917594909667969, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.2073162943124771, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.09303848445415497, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.1441424936056137, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.0984899252653122, "rewards/len_reward/p75": 0.3025240898132324, "rewards/len_reward/var": 0.07761579006910324, "rewards/missing_response_penalty": -0.1197916716337204, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.10599368065595627, "smoothed_logprobs/min": -1.0, "step": 16, "value_loss": 3.2265625, "values": -0.7615966796875, "values/max": 1.9897232055664062, "values/median": -0.8037109375, "values/min": -2.001072883605957, "values/p25": -0.8037109375, "values/p75": -0.8037109375, "values/var": 0.32145166397094727 }, { "adam_stats/lr_effective_max": 4.9730995669960976e-05, "adam_stats/lr_effective_mean": 2.7487906048562216e-11, "adam_stats/lr_effective_min": -4.8396959755336866e-05, "adam_stats/m_t_max": 0.004705157596617937, "adam_stats/m_t_mean": -2.7418390474709398e-11, "adam_stats/m_t_min": -0.0033178571611642838, "adam_stats/v_t_max": 2.307903559994884e-05, "adam_stats/v_t_mean": 1.4827556500326966e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.4050632911392405, "eval_accuracy_reward": 0.6600000262260437, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.22515049576759338, "eval_advantages": 0.6600000262260437, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.22515049576759338, "eval_all_logprobs": -0.013608547858893871, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -10.0, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.021750587970018387, "eval_answer_logprob_reward": 3.0337154865264893, "eval_answer_logprob_reward/correct": 3.1085681915283203, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.07818499207496643, "eval_answer_logprob_reward/incorrect": 2.888413429260254, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.6094379425048828, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.24697057902812958, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.6094379425048828, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.14585068821907043, "eval_completion_length": 551.8099975585938, "eval_completion_length/correct": 485.5404052734375, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 444.0, "eval_completion_length/correct/min": 102.0, "eval_completion_length/correct/p25": 322.25, "eval_completion_length/correct/p75": 637.0, "eval_completion_length/correct/var": 45999.62890625, "eval_completion_length/incorrect": 680.4509887695312, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 638.0, "eval_completion_length/incorrect/min": 172.0, "eval_completion_length/incorrect/p25": 433.0, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 82459.515625, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 489.0, "eval_completion_length/min": 102.0, "eval_completion_length/p25": 344.75, "eval_completion_length/p75": 733.25, "eval_completion_length/var": 66715.1640625, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 2799750.0, "eval_len_reward": 0.04988803341984749, "eval_len_reward/correct": 0.15069575607776642, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.1733633577823639, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.002489530947059393, "eval_len_reward/correct/p75": 0.3220340609550476, "eval_len_reward/correct/var": 0.05049169436097145, "eval_len_reward/incorrect": -0.14579756557941437, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.041744664311409, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.3054758310317993, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03288053721189499, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.09548091888427734, "eval_len_reward/p75": 0.23046810925006866, "eval_len_reward/var": 0.06416650861501694, "eval_mean_logprobs": -0.01409912109375, "eval_mean_logprobs/var": 9.72747802734375e-05, "eval_missing_response_penalty": -0.09666667133569717, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.28431373834609985, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.20549406111240387, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.0876142606139183, "eval_num_completions/total": 4800, "eval_pred_value_log_probs": -0.6389652490615845, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.6875, "eval_pred_value_log_probs/min": -0.91015625, "eval_pred_value_log_probs/p25": -0.6875, "eval_pred_value_log_probs/p75": -0.6875, "eval_pred_value_log_probs/var": 0.027192160487174988, "eval_reward": 0.6600000262260437, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.22515049576759338, "eval_rewards/accuracy_reward": 0.6600000262260437, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.22515049576759338, "eval_rewards/answer_logprob_reward": 3.0337154865264893, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.6094379425048828, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.14585068821907043, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.04988803341984749, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.09548091888427734, "eval_rewards/len_reward/p75": 0.23046810925006866, "eval_rewards/len_reward/var": 0.06416650861501694, "eval_rewards/missing_response_penalty": -0.09666667133569717, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.0876142606139183, "eval_runtime": 165.2241, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8467819094657898, "eval_values/max": 1.99127197265625, "eval_values/median": -0.8046875, "eval_values/min": -1.990447998046875, "eval_values/p25": -0.8046875, "eval_values/p75": -0.8046875, "eval_values/var": 0.1333002895116806, "step": 16 }, { "accuracy_reward": 0.7135416865348816, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.20547011494636536, "adam_stats/lr_effective_max": 4.561969035421498e-05, "adam_stats/lr_effective_mean": 3.8943893249099304e-11, "adam_stats/lr_effective_min": -4.658138641389087e-05, "adam_stats/m_t_max": 0.004057639744132757, "adam_stats/m_t_mean": -1.9962500402703753e-11, "adam_stats/m_t_min": -0.0026576807722449303, "adam_stats/v_t_max": 2.3059088562149554e-05, "adam_stats/v_t_mean": 1.4879517973645506e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7135416865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.20547011494636536, "all_logprobs": -0.014197630807757378, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -8.75, "all_logprobs/p25": -1.1920928955078125e-07, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.022814959287643433, "answer_logprob_reward": 2.971250057220459, "answer_logprob_reward/correct": 3.0861237049102783, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.06222710758447647, "answer_logprob_reward/incorrect": 2.6851089000701904, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.3862943649291992, "answer_logprob_reward/incorrect/p25": 2.079441547393799, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.426470547914505, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.3862943649291992, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.19792331755161285, "clip_ratio": 0.0, "completion_length": 513.0208740234375, "completion_length/correct": 430.98541259765625, "completion_length/correct/max": 943.0, "completion_length/correct/median": 411.0, "completion_length/correct/min": 153.0, "completion_length/correct/p25": 275.0, "completion_length/correct/p75": 552.0, "completion_length/correct/var": 37020.87890625, "completion_length/incorrect": 717.3635864257812, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 715.0, "completion_length/incorrect/min": 237.0, "completion_length/incorrect/p25": 474.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 70372.3125, "completion_length/max": 1024.0, "completion_length/median": 460.0, "completion_length/min": 153.0, "completion_length/p25": 298.5, "completion_length/p75": 665.25, "completion_length/var": 63107.37109375, "epoch": 0.43037974683544306, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 1875275.0, "grad_norm": 0.2200733721256256, "kl": 0.012603759765625, "learning_rate": 6.965125158269619e-06, "len_reward": 0.07363046705722809, "len_reward/correct": 0.16898098587989807, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.2068466693162918, "len_reward/correct/min": -0.5, "len_reward/correct/p25": -0.022179363295435905, "len_reward/correct/p75": 0.350384384393692, "len_reward/correct/var": 0.05886177346110344, "len_reward/incorrect": -0.16387902200222015, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.07846952229738235, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.2992038130760193, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03208499401807785, "len_reward/max": 0.5, "len_reward/median": 0.061749570071697235, "len_reward/min": -0.5, "len_reward/p25": -0.09734134376049042, "len_reward/p75": 0.3237048089504242, "len_reward/var": 0.07374843209981918, "loss": -0.713, "mean_logprobs": -0.01446533203125, "mean_logprobs/var": 9.965896606445312e-05, "missing_response_penalty": -0.0989583358168602, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.34545454382896423, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.23030301928520203, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.08963242173194885, "non_smoothed_logprobs/min": -17.6875, "num_completions/total": 3264, "policy_loss": -0.7130376100540161, "policy_loss/max": 0.001001933473162353, "policy_loss/median": -0.9994107484817505, "policy_loss/min": -0.9998096227645874, "policy_loss/p25": -0.9995970726013184, "policy_loss/p75": 0.00025670218747109175, "policy_loss/var": 0.20545172691345215, "pred_value_log_probs": -0.6223148107528687, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.6875, "pred_value_log_probs/min": -1.1953125, "pred_value_log_probs/p25": -0.6875, "pred_value_log_probs/p75": -0.6875, "pred_value_log_probs/var": 0.03758833557367325, "reward": 0.7135416865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.20547011494636536, "rewards/accuracy_reward": 0.7135416865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.20547011494636536, "rewards/answer_logprob_reward": 2.971250057220459, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.3862943649291992, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.19792331755161285, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.07363046705722809, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.061749570071697235, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.09734134376049042, "rewards/len_reward/p75": 0.3237048089504242, "rewards/len_reward/var": 0.07374843209981918, "rewards/missing_response_penalty": -0.0989583358168602, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.08963242173194885, "smoothed_logprobs/min": -1.0, "step": 17, "value_loss": 2.796875, "values": -0.8071715235710144, "values/max": 1.830169677734375, "values/median": -0.8046875, "values/min": -1.9844542741775513, "values/p25": -0.8046875, "values/p75": -0.8046875, "values/var": 0.2470327913761139 }, { "adam_stats/lr_effective_max": 4.561969035421498e-05, "adam_stats/lr_effective_mean": 3.8943893249099304e-11, "adam_stats/lr_effective_min": -4.658138641389087e-05, "adam_stats/m_t_max": 0.004057639744132757, "adam_stats/m_t_mean": -1.9962500402703753e-11, "adam_stats/m_t_min": -0.0026576807722449303, "adam_stats/v_t_max": 2.3059088562149554e-05, "adam_stats/v_t_mean": 1.4879517973645506e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.43037974683544306, "eval_accuracy_reward": 0.643333375453949, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.23022297024726868, "eval_advantages": 0.643333375453949, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.23022297024726868, "eval_all_logprobs": -0.014583272859454155, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -11.0, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.02367628738284111, "eval_answer_logprob_reward": 3.040311574935913, "eval_answer_logprob_reward/correct": 3.108691930770874, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.08312822133302689, "eval_answer_logprob_reward/incorrect": 2.91697096824646, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.4271163940429688, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.29633092880249023, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.4271163940429688, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.1668960005044937, "eval_completion_length": 538.5833740234375, "eval_completion_length/correct": 470.3575134277344, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 434.0, "eval_completion_length/correct/min": 123.0, "eval_completion_length/correct/p25": 315.0, "eval_completion_length/correct/p75": 574.0, "eval_completion_length/correct/var": 41447.609375, "eval_completion_length/incorrect": 661.6448364257812, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 637.0, "eval_completion_length/incorrect/min": 172.0, "eval_completion_length/incorrect/p25": 423.5, "eval_completion_length/incorrect/p75": 995.0, "eval_completion_length/incorrect/var": 80487.4375, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 474.0, "eval_completion_length/min": 123.0, "eval_completion_length/p25": 330.75, "eval_completion_length/p75": 704.0, "eval_completion_length/var": 63573.23828125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 2961325.0, "eval_len_reward": 0.022688377648591995, "eval_len_reward/correct": 0.12160007655620575, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.1525157243013382, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": -0.034984275698661804, "eval_len_reward/correct/p75": 0.28127622604370117, "eval_len_reward/correct/var": 0.058058690279722214, "eval_len_reward/incorrect": -0.1557224690914154, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.06534695625305176, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.2915356159210205, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03322194516658783, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.15201473236083984, "eval_len_reward/p75": 0.210999995470047, "eval_len_reward/var": 0.0667654499411583, "eval_mean_logprobs": -0.0146484375, "eval_mean_logprobs/var": 0.00010395050048828125, "eval_missing_response_penalty": -0.0833333358168602, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.23364485800266266, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": 0.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.18074415624141693, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.07664437592029572, "eval_num_completions/total": 5100, "eval_pred_value_log_probs": -0.6259163618087769, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.671875, "eval_pred_value_log_probs/min": -0.94921875, "eval_pred_value_log_probs/p25": -0.671875, "eval_pred_value_log_probs/p75": -0.671875, "eval_pred_value_log_probs/var": 0.025486089289188385, "eval_reward": 0.643333375453949, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.23022297024726868, "eval_rewards/accuracy_reward": 0.643333375453949, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.23022297024726868, "eval_rewards/answer_logprob_reward": 3.040311574935913, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.4271163940429688, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.1668960005044937, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.022688377648591995, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.15201473236083984, "eval_rewards/len_reward/p75": 0.210999995470047, "eval_rewards/len_reward/var": 0.0667654499411583, "eval_rewards/missing_response_penalty": -0.0833333358168602, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.07664437592029572, "eval_runtime": 163.4822, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8655804395675659, "eval_values/max": 1.990936279296875, "eval_values/median": -0.8271484375, "eval_values/min": -1.9853706359863281, "eval_values/p25": -0.8271484375, "eval_values/p75": -0.8271484375, "eval_values/var": 0.12937664985656738, "step": 17 }, { "accuracy_reward": 0.7552083730697632, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.18583661317825317, "adam_stats/lr_effective_max": 4.375962816993706e-05, "adam_stats/lr_effective_mean": 2.8192944506733042e-11, "adam_stats/lr_effective_min": -4.2835705244215205e-05, "adam_stats/m_t_max": 0.005055684596300125, "adam_stats/m_t_mean": -6.154668441560318e-11, "adam_stats/m_t_min": -0.0036076109390705824, "adam_stats/v_t_max": 2.3233094907482155e-05, "adam_stats/v_t_mean": 1.5197829973667698e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7552083730697632, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.18583661317825317, "all_logprobs": -0.013125843368470669, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -7.5, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.02088928036391735, "answer_logprob_reward": 2.927567481994629, "answer_logprob_reward/correct": 3.0681064128875732, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.4849066734313965, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.06456664204597473, "answer_logprob_reward/incorrect": 2.493990182876587, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 2.4849066734313965, "answer_logprob_reward/incorrect/min": 1.3862943649291992, "answer_logprob_reward/incorrect/p25": 2.079441547393799, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.5105509757995605, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.3862943649291992, "answer_logprob_reward/p25": 3.0047669410705566, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.23289190232753754, "clip_ratio": 0.0, "completion_length": 553.984375, "completion_length/correct": 463.2896423339844, "completion_length/correct/max": 1017.0, "completion_length/correct/median": 418.0, "completion_length/correct/min": 143.0, "completion_length/correct/p25": 334.0, "completion_length/correct/p75": 561.0, "completion_length/correct/var": 37730.87109375, "completion_length/incorrect": 833.7872314453125, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 1024.0, "completion_length/incorrect/min": 244.0, "completion_length/incorrect/p25": 661.5, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 62925.0390625, "completion_length/max": 1024.0, "completion_length/median": 472.0, "completion_length/min": 143.0, "completion_length/p25": 353.0, "completion_length/p75": 737.25, "completion_length/var": 69110.5390625, "epoch": 0.45569620253164556, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 1981640.0, "grad_norm": 0.31346508860588074, "kl": 0.011932373046875, "learning_rate": 6.545084971874738e-06, "len_reward": 0.06259474903345108, "len_reward/correct": 0.16533038020133972, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.2032332569360733, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.04145798459649086, "len_reward/correct/p75": 0.3477851152420044, "len_reward/correct/var": 0.061689212918281555, "len_reward/incorrect": -0.25435560941696167, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.24641422927379608, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.4109390676021576, "len_reward/incorrect/p75": -0.1067853569984436, "len_reward/incorrect/var": 0.03014647774398327, "len_reward/max": 0.5, "len_reward/median": 0.08783120661973953, "len_reward/min": -0.5, "len_reward/p25": -0.1487763375043869, "len_reward/p75": 0.2968714237213135, "len_reward/var": 0.08650212734937668, "loss": -0.7547, "mean_logprobs": -0.01318359375, "mean_logprobs/var": 0.0001850128173828125, "missing_response_penalty": -0.125, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.5106382966041565, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": -1.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.25531914830207825, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.10994765162467957, "non_smoothed_logprobs/min": -17.625, "num_completions/total": 3456, "policy_loss": -0.754731297492981, "policy_loss/max": 0.0008864151895977557, "policy_loss/median": -0.99945467710495, "policy_loss/min": -0.9998028874397278, "policy_loss/p25": -0.999605655670166, "policy_loss/p75": -0.9989223480224609, "policy_loss/var": 0.18581964075565338, "pred_value_log_probs": -0.6210931539535522, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.671875, "pred_value_log_probs/min": -1.421875, "pred_value_log_probs/p25": -0.671875, "pred_value_log_probs/p75": -0.671875, "pred_value_log_probs/var": 0.04601706191897392, "reward": 0.7552083730697632, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.18583661317825317, "rewards/accuracy_reward": 0.7552083730697632, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.18583661317825317, "rewards/answer_logprob_reward": 2.927567481994629, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.3862943649291992, "rewards/answer_logprob_reward/p25": 3.0047669410705566, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.23289190232753754, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.06259474903345108, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.08783120661973953, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.1487763375043869, "rewards/len_reward/p75": 0.2968714237213135, "rewards/len_reward/var": 0.08650212734937668, "rewards/missing_response_penalty": -0.125, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.10994765162467957, "smoothed_logprobs/min": -1.0, "step": 18, "value_loss": 2.96875, "values": -0.8621608018875122, "values/max": 2.0, "values/median": -0.8271484375, "values/min": -2.0, "values/p25": -0.8271484375, "values/p75": -0.8271484375, "values/var": 0.20169949531555176 }, { "adam_stats/lr_effective_max": 4.375962816993706e-05, "adam_stats/lr_effective_mean": 2.8192944506733042e-11, "adam_stats/lr_effective_min": -4.2835705244215205e-05, "adam_stats/m_t_max": 0.005055684596300125, "adam_stats/m_t_mean": -6.154668441560318e-11, "adam_stats/m_t_min": -0.0036076109390705824, "adam_stats/v_t_max": 2.3233094907482155e-05, "adam_stats/v_t_mean": 1.5197829973667698e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.45569620253164556, "eval_accuracy_reward": 0.6399999856948853, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.23117054998874664, "eval_advantages": 0.6399999856948853, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.23117054998874664, "eval_all_logprobs": -0.014533042907714844, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -12.875, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.023786023259162903, "eval_answer_logprob_reward": 3.0196218490600586, "eval_answer_logprob_reward/correct": 3.1066195964813232, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.08187548816204071, "eval_answer_logprob_reward/incorrect": 2.864959478378296, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.6094379425048828, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.35166385769844055, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.6094379425048828, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.1916482299566269, "eval_completion_length": 538.77001953125, "eval_completion_length/correct": 471.7864685058594, "eval_completion_length/correct/max": 1021.0, "eval_completion_length/correct/median": 434.0, "eval_completion_length/correct/min": 102.0, "eval_completion_length/correct/p25": 312.5, "eval_completion_length/correct/p75": 603.25, "eval_completion_length/correct/var": 43407.1328125, "eval_completion_length/incorrect": 657.8518676757812, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 593.0, "eval_completion_length/incorrect/min": 163.0, "eval_completion_length/incorrect/p25": 432.5, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 80225.25, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 491.0, "eval_completion_length/min": 102.0, "eval_completion_length/p25": 339.75, "eval_completion_length/p75": 686.75, "eval_completion_length/var": 64440.875, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 3122956.0, "eval_len_reward": 0.04496534913778305, "eval_len_reward/correct": 0.1389261931180954, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.16355518996715546, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": -0.011785062029957771, "eval_len_reward/correct/p75": 0.31801870465278625, "eval_len_reward/correct/var": 0.05499494448304176, "eval_len_reward/incorrect": -0.12207616120576859, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.0275835283100605, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.20396654307842255, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.027708251029253006, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.08302466571331024, "eval_len_reward/p75": 0.23195244371891022, "eval_len_reward/var": 0.06079406663775444, "eval_mean_logprobs": -0.0146484375, "eval_mean_logprobs/var": 0.0001277923583984375, "eval_missing_response_penalty": -0.09666667133569717, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.26851850748062134, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.19825200736522675, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.0876142606139183, "eval_num_completions/total": 5400, "eval_pred_value_log_probs": -0.6362558007240295, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.68359375, "eval_pred_value_log_probs/min": -0.96875, "eval_pred_value_log_probs/p25": -0.68359375, "eval_pred_value_log_probs/p75": -0.68359375, "eval_pred_value_log_probs/var": 0.026474738493561745, "eval_reward": 0.6399999856948853, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.23117054998874664, "eval_rewards/accuracy_reward": 0.6399999856948853, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.23117054998874664, "eval_rewards/answer_logprob_reward": 3.0196218490600586, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.6094379425048828, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.1916482299566269, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.04496534913778305, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.08302466571331024, "eval_rewards/len_reward/p75": 0.23195244371891022, "eval_rewards/len_reward/var": 0.06079406663775444, "eval_rewards/missing_response_penalty": -0.09666667133569717, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.0876142606139183, "eval_runtime": 168.3289, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8485709428787231, "eval_values/max": 1.9911270141601562, "eval_values/median": -0.80859375, "eval_values/min": -1.9788665771484375, "eval_values/p25": -0.80859375, "eval_values/p75": -0.80859375, "eval_values/var": 0.129899263381958, "step": 18 }, { "accuracy_reward": 0.8020833730697632, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.15957678854465485, "adam_stats/lr_effective_max": 3.9845628634793684e-05, "adam_stats/lr_effective_mean": 2.0078984308558034e-10, "adam_stats/lr_effective_min": -3.964529969380237e-05, "adam_stats/m_t_max": 0.00423883693292737, "adam_stats/m_t_mean": -4.4684693445429247e-11, "adam_stats/m_t_min": -0.003130326746031642, "adam_stats/v_t_max": 2.3219552531372756e-05, "adam_stats/v_t_mean": 1.53001916691764e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.8020833730697632, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.15957678854465485, "all_logprobs": -0.01230152789503336, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -7.84375, "all_logprobs/p25": -1.1920928955078125e-07, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.018796944990754128, "answer_logprob_reward": 3.048903226852417, "answer_logprob_reward/correct": 3.123277425765991, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.04101899638772011, "answer_logprob_reward/incorrect": 2.747490644454956, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 2.079441547393799, "answer_logprob_reward/incorrect/p25": 2.079441547393799, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.272623747587204, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 2.079441547393799, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.10820484906435013, "clip_ratio": 0.0, "completion_length": 529.3125, "completion_length/correct": 455.3246765136719, "completion_length/correct/max": 913.0, "completion_length/correct/median": 423.0, "completion_length/correct/min": 146.0, "completion_length/correct/p25": 313.25, "completion_length/correct/p75": 583.25, "completion_length/correct/var": 33110.40234375, "completion_length/incorrect": 829.1578979492188, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 1003.0, "completion_length/incorrect/min": 350.0, "completion_length/incorrect/p25": 621.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 59654.7890625, "completion_length/max": 1024.0, "completion_length/median": 466.0, "completion_length/min": 146.0, "completion_length/p25": 338.0, "completion_length/p75": 675.25, "completion_length/var": 60380.21484375, "epoch": 0.4810126582278481, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 2083268.0, "grad_norm": 0.24223755300045013, "kl": 0.01190185546875, "learning_rate": 6.112604669781572e-06, "len_reward": 0.09958770126104355, "len_reward/correct": 0.18344828486442566, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.2180722951889038, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.020527295768260956, "len_reward/correct/p75": 0.3609812259674072, "len_reward/correct/var": 0.04795420542359352, "len_reward/incorrect": -0.24026833474636078, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.2524407207965851, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.462963342666626, "len_reward/incorrect/p75": -0.009423407725989819, "len_reward/incorrect/var": 0.04041590169072151, "len_reward/max": 0.5, "len_reward/median": 0.12857143580913544, "len_reward/min": -0.5, "len_reward/p25": -0.06484229117631912, "len_reward/p75": 0.3338591456413269, "len_reward/var": 0.07489258050918579, "loss": -0.8016, "mean_logprobs": -0.01177978515625, "mean_logprobs/var": 8.440017700195312e-05, "missing_response_penalty": -0.0833333358168602, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.42105263471603394, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.25035563111305237, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.07678882777690887, "non_smoothed_logprobs/min": -17.25, "num_completions/total": 3648, "policy_loss": -0.8016080260276794, "policy_loss/max": 0.0008301552152261138, "policy_loss/median": -0.9994857907295227, "policy_loss/min": -0.9998277425765991, "policy_loss/p25": -0.9996201395988464, "policy_loss/p75": -0.9992146492004395, "policy_loss/var": 0.15955623984336853, "pred_value_log_probs": -0.634164035320282, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.68359375, "pred_value_log_probs/min": -1.3125, "pred_value_log_probs/p25": -0.68359375, "pred_value_log_probs/p75": -0.68359375, "pred_value_log_probs/var": 0.04011666402220726, "reward": 0.8020833730697632, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.15957678854465485, "rewards/accuracy_reward": 0.8020833730697632, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.15957678854465485, "rewards/answer_logprob_reward": 3.048903226852417, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 2.079441547393799, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.10820484906435013, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.09958770126104355, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.12857143580913544, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.06484229117631912, "rewards/len_reward/p75": 0.3338591456413269, "rewards/len_reward/var": 0.07489258050918579, "rewards/missing_response_penalty": -0.0833333358168602, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.07678882777690887, "smoothed_logprobs/min": -1.0, "step": 19, "value_loss": 2.953125, "values": -0.7967839241027832, "values/max": 2.0, "values/median": -0.80859375, "values/min": -2.0, "values/p25": -0.80859375, "values/p75": -0.80859375, "values/var": 0.2700471878051758 }, { "adam_stats/lr_effective_max": 3.9845628634793684e-05, "adam_stats/lr_effective_mean": 2.0078984308558034e-10, "adam_stats/lr_effective_min": -3.964529969380237e-05, "adam_stats/m_t_max": 0.00423883693292737, "adam_stats/m_t_mean": -4.4684693445429247e-11, "adam_stats/m_t_min": -0.003130326746031642, "adam_stats/v_t_max": 2.3219552531372756e-05, "adam_stats/v_t_mean": 1.53001916691764e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.4810126582278481, "eval_accuracy_reward": 0.6866666674613953, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.21587511897087097, "eval_advantages": 0.6866666674613953, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.21587511897087097, "eval_all_logprobs": -0.013197014108300209, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -12.5, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.019858872517943382, "eval_answer_logprob_reward": 3.0339996814727783, "eval_answer_logprob_reward/correct": 3.1187570095062256, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.06618177890777588, "eval_answer_logprob_reward/incorrect": 2.848254919052124, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.272965669631958, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.3835936486721039, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.272965669631958, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.18048304319381714, "eval_completion_length": 546.316650390625, "eval_completion_length/correct": 492.4563293457031, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 452.0, "eval_completion_length/correct/min": 102.0, "eval_completion_length/correct/p25": 313.5, "eval_completion_length/correct/p75": 630.25, "eval_completion_length/correct/var": 49226.74609375, "eval_completion_length/incorrect": 664.3510131835938, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 637.0, "eval_completion_length/incorrect/min": 170.0, "eval_completion_length/incorrect/p25": 403.75, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 88437.84375, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 496.0, "eval_completion_length/min": 102.0, "eval_completion_length/p25": 332.75, "eval_completion_length/p75": 720.0, "eval_completion_length/var": 67636.8359375, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 3286851.0, "eval_len_reward": 0.03408301994204521, "eval_len_reward/correct": 0.1176217645406723, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.15866558253765106, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": -0.02469135820865631, "eval_len_reward/correct/p75": 0.30703064799308777, "eval_len_reward/correct/var": 0.06151648238301277, "eval_len_reward/incorrect": -0.14899122714996338, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.07014767825603485, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.2524755299091339, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.031159698963165283, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.11927083134651184, "eval_len_reward/p75": 0.221572607755661, "eval_len_reward/var": 0.06721360236406326, "eval_mean_logprobs": -0.01361083984375, "eval_mean_logprobs/var": 0.00014781951904296875, "eval_missing_response_penalty": -0.08666666597127914, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.27659574151039124, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.20224204659461975, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.0794202908873558, "eval_num_completions/total": 5700, "eval_pred_value_log_probs": -0.6355226039886475, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.68359375, "eval_pred_value_log_probs/min": -0.953125, "eval_pred_value_log_probs/p25": -0.68359375, "eval_pred_value_log_probs/p75": -0.68359375, "eval_pred_value_log_probs/var": 0.026992160826921463, "eval_reward": 0.6866666674613953, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.21587511897087097, "eval_rewards/accuracy_reward": 0.6866666674613953, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.21587511897087097, "eval_rewards/answer_logprob_reward": 3.0339996814727783, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.272965669631958, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.18048304319381714, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.03408301994204521, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.11927083134651184, "eval_rewards/len_reward/p75": 0.221572607755661, "eval_rewards/len_reward/var": 0.06721360236406326, "eval_rewards/missing_response_penalty": -0.08666666597127914, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.0794202908873558, "eval_runtime": 176.7268, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8475617170333862, "eval_values/max": 1.9909133911132812, "eval_values/median": -0.8076171875, "eval_values/min": -1.98785400390625, "eval_values/p25": -0.8076171875, "eval_values/p75": -0.8076171875, "eval_values/var": 0.13097244501113892, "step": 19 }, { "accuracy_reward": 0.703125, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.20983311533927917, "adam_stats/lr_effective_max": 3.718593143275939e-05, "adam_stats/lr_effective_mean": 1.5384329921097617e-10, "adam_stats/lr_effective_min": -3.6357450881041586e-05, "adam_stats/m_t_max": 0.004364269785583019, "adam_stats/m_t_mean": -3.4422544675782873e-11, "adam_stats/m_t_min": -0.003123733215034008, "adam_stats/v_t_max": 2.3226508346851915e-05, "adam_stats/v_t_mean": 1.5332153949221272e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.703125, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.20983311533927917, "all_logprobs": -0.011873790062963963, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -6.75, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.01732758805155754, "answer_logprob_reward": 3.01825213432312, "answer_logprob_reward/correct": 3.1318440437316895, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.4849066734313965, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.030117960646748543, "answer_logprob_reward/incorrect": 2.7492189407348633, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.7917594909667969, "answer_logprob_reward/incorrect/p25": 2.079441547393799, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.32423263788223267, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.7917594909667969, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.1469128578901291, "clip_ratio": 0.0, "completion_length": 576.8854370117188, "completion_length/correct": 482.5555419921875, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 436.0, "completion_length/correct/min": 173.0, "completion_length/correct/p25": 307.5, "completion_length/correct/p75": 634.5, "completion_length/correct/var": 41892.47265625, "completion_length/incorrect": 800.2982788085938, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 901.0, "completion_length/incorrect/min": 206.0, "completion_length/incorrect/p25": 574.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 66785.7890625, "completion_length/max": 1024.0, "completion_length/median": 544.0, "completion_length/min": 173.0, "completion_length/p25": 349.25, "completion_length/p75": 785.0, "completion_length/var": 70156.5390625, "epoch": 0.5063291139240507, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 2194030.0, "grad_norm": 0.19211068749427795, "kl": 0.011749267578125, "learning_rate": 5.671166329088278e-06, "len_reward": 0.08219938725233078, "len_reward/correct": 0.19225099682807922, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.24640198051929474, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.04950905218720436, "len_reward/correct/p75": 0.37484481930732727, "len_reward/correct/var": 0.05272315815091133, "len_reward/incorrect": -0.17844919860363007, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.13193117082118988, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.293021023273468, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.033872731029987335, "len_reward/max": 0.5, "len_reward/median": 0.0772036463022232, "len_reward/min": -0.5, "len_reward/p25": -0.08989867568016052, "len_reward/p75": 0.3173549175262451, "len_reward/var": 0.075755275785923, "loss": -0.7027, "mean_logprobs": -0.01153564453125, "mean_logprobs/var": 5.7220458984375e-05, "missing_response_penalty": -0.109375, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.3684210479259491, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.2368421107530594, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.09792213141918182, "non_smoothed_logprobs/min": -17.1875, "num_completions/total": 3840, "policy_loss": -0.7026552557945251, "policy_loss/max": 0.0008396075572818518, "policy_loss/median": -0.9994357228279114, "policy_loss/min": -0.99984210729599, "policy_loss/p25": -0.9995936155319214, "policy_loss/p75": 0.0003041678573936224, "policy_loss/var": 0.20981955528259277, "pred_value_log_probs": -0.6398011445999146, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.68359375, "pred_value_log_probs/min": -1.3125, "pred_value_log_probs/p25": -0.68359375, "pred_value_log_probs/p75": -0.68359375, "pred_value_log_probs/var": 0.04309263452887535, "reward": 0.703125, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.20983311533927917, "rewards/accuracy_reward": 0.703125, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.20983311533927917, "rewards/answer_logprob_reward": 3.01825213432312, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.7917594909667969, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.1469128578901291, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.08219938725233078, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.0772036463022232, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.08989867568016052, "rewards/len_reward/p75": 0.3173549175262451, "rewards/len_reward/var": 0.075755275785923, "rewards/missing_response_penalty": -0.109375, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.09792213141918182, "smoothed_logprobs/min": -1.0, "step": 20, "value_loss": 3.015625, "values": -0.8254277110099792, "values/max": 2.000021457672119, "values/median": -0.8076171875, "values/min": -1.94500732421875, "values/p25": -0.8076171875, "values/p75": -0.8076171875, "values/var": 0.1680065244436264 }, { "adam_stats/lr_effective_max": 3.718593143275939e-05, "adam_stats/lr_effective_mean": 1.5384329921097617e-10, "adam_stats/lr_effective_min": -3.6357450881041586e-05, "adam_stats/m_t_max": 0.004364269785583019, "adam_stats/m_t_mean": -3.4422544675782873e-11, "adam_stats/m_t_min": -0.003123733215034008, "adam_stats/v_t_max": 2.3226508346851915e-05, "adam_stats/v_t_mean": 1.5332153949221272e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.5063291139240507, "eval_accuracy_reward": 0.6566666960716248, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.2262096107006073, "eval_advantages": 0.6566666960716248, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.2262096107006073, "eval_all_logprobs": -0.013537714257836342, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -7.4375, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.02079758048057556, "eval_answer_logprob_reward": 3.0227293968200684, "eval_answer_logprob_reward/correct": 3.1032936573028564, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.0898236483335495, "eval_answer_logprob_reward/incorrect": 2.8686399459838867, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.4271163940429688, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.3311939537525177, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.4271163940429688, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.18431931734085083, "eval_completion_length": 537.4733276367188, "eval_completion_length/correct": 474.22332763671875, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 448.0, "eval_completion_length/correct/min": 102.0, "eval_completion_length/correct/p25": 304.0, "eval_completion_length/correct/p75": 600.0, "eval_completion_length/correct/var": 44748.88671875, "eval_completion_length/incorrect": 658.4465942382812, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 591.0, "eval_completion_length/incorrect/min": 167.0, "eval_completion_length/incorrect/p25": 411.0, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 84326.3359375, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 481.0, "eval_completion_length/min": 102.0, "eval_completion_length/p25": 327.0, "eval_completion_length/p75": 702.0, "eval_completion_length/var": 65777.703125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 3448093.0, "eval_len_reward": 0.04659387096762657, "eval_len_reward/correct": 0.14398375153541565, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.18306010961532593, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.012527751736342907, "eval_len_reward/correct/p75": 0.2897033095359802, "eval_len_reward/correct/var": 0.053421687334775925, "eval_len_reward/incorrect": -0.1396760791540146, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.024859286844730377, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.2602172791957855, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.033676546066999435, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.08028224855661392, "eval_len_reward/p75": 0.2392531782388687, "eval_len_reward/var": 0.06470870971679688, "eval_mean_logprobs": -0.01361083984375, "eval_mean_logprobs/var": 0.00010442733764648438, "eval_missing_response_penalty": -0.0833333358168602, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.24271844327449799, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": 0.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.18560825288295746, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.07664437592029572, "eval_num_completions/total": 6000, "eval_pred_value_log_probs": -0.63873291015625, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.6875, "eval_pred_value_log_probs/min": -0.97265625, "eval_pred_value_log_probs/p25": -0.6875, "eval_pred_value_log_probs/p75": -0.6875, "eval_pred_value_log_probs/var": 0.027704287320375443, "eval_reward": 0.6566666960716248, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.2262096107006073, "eval_rewards/accuracy_reward": 0.6566666960716248, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.2262096107006073, "eval_rewards/answer_logprob_reward": 3.0227293968200684, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.4271163940429688, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.18431931734085083, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.04659387096762657, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.08028224855661392, "eval_rewards/len_reward/p75": 0.2392531782388687, "eval_rewards/len_reward/var": 0.06470870971679688, "eval_rewards/missing_response_penalty": -0.0833333358168602, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.07664437592029572, "eval_runtime": 178.2592, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8452271819114685, "eval_values/max": 1.9911384582519531, "eval_values/median": -0.8037109375, "eval_values/min": -1.9876670837402344, "eval_values/p25": -0.8037109375, "eval_values/p75": -0.8037109375, "eval_values/var": 0.1308903843164444, "step": 20 }, { "accuracy_reward": 0.734375, "accuracy_reward/correct": 0.9999999403953552, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.19608965516090393, "adam_stats/lr_effective_max": 3.338371971040033e-05, "adam_stats/lr_effective_mean": 1.3322469516463542e-10, "adam_stats/lr_effective_min": -3.350756742293015e-05, "adam_stats/m_t_max": 0.0037767805624753237, "adam_stats/m_t_mean": -3.4601509157905497e-11, "adam_stats/m_t_min": -0.002743084216490388, "adam_stats/v_t_max": 2.3205564502859488e-05, "adam_stats/v_t_mean": 1.5373737439344781e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.734375, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.19608965516090393, "all_logprobs": -0.013131962157785892, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -12.25, "all_logprobs/p25": -1.1920928955078125e-07, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.021637005731463432, "answer_logprob_reward": 3.0625293254852295, "answer_logprob_reward/correct": 3.1485581398010254, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.4849066734313965, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.019714640453457832, "answer_logprob_reward/incorrect": 2.8246846199035645, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.7917594909667969, "answer_logprob_reward/incorrect/p25": 2.4849066734313965, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.2762133777141571, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.7917594909667969, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.10732635110616684, "clip_ratio": 0.0, "completion_length": 531.1614990234375, "completion_length/correct": 446.39715576171875, "completion_length/correct/max": 1020.0, "completion_length/correct/median": 412.0, "completion_length/correct/min": 135.0, "completion_length/correct/p25": 285.0, "completion_length/correct/p75": 579.0, "completion_length/correct/var": 38654.16796875, "completion_length/incorrect": 765.5098266601562, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 888.0, "completion_length/incorrect/min": 280.0, "completion_length/incorrect/p25": 501.5, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 76984.9765625, "completion_length/max": 1024.0, "completion_length/median": 462.0, "completion_length/min": 135.0, "completion_length/p25": 320.25, "completion_length/p75": 666.75, "completion_length/var": 68454.40625, "epoch": 0.5316455696202531, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 2296013.0, "grad_norm": 0.20857343077659607, "kl": 0.0126953125, "learning_rate": 5.224324151752575e-06, "len_reward": 0.09813292324542999, "len_reward/correct": 0.20786333084106445, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.2633753716945648, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.052776601165533066, "len_reward/correct/p75": 0.363953173160553, "len_reward/correct/var": 0.04442443698644638, "len_reward/incorrect": -0.2052394300699234, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.1717289686203003, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.3497586250305176, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03765484690666199, "len_reward/max": 0.5, "len_reward/median": 0.13923776149749756, "len_reward/min": -0.5, "len_reward/p25": -0.07858544588088989, "len_reward/p75": 0.32227522134780884, "len_reward/var": 0.0758831650018692, "loss": -0.7339, "mean_logprobs": -0.01287841796875, "mean_logprobs/var": 0.00011491775512695312, "missing_response_penalty": -0.09375, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.3529411852359772, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.2329411655664444, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.08540575951337814, "non_smoothed_logprobs/min": -17.4375, "num_completions/total": 4032, "policy_loss": -0.7338664531707764, "policy_loss/max": 0.000989806023426354, "policy_loss/median": -0.9994437098503113, "policy_loss/min": -0.9997986555099487, "policy_loss/p25": -0.9995540380477905, "policy_loss/p75": 0.0002309512346982956, "policy_loss/var": 0.19607244431972504, "pred_value_log_probs": -0.6659946441650391, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.6875, "pred_value_log_probs/min": -1.2265625, "pred_value_log_probs/p25": -0.6875, "pred_value_log_probs/p75": -0.6875, "pred_value_log_probs/var": 0.025847474113106728, "reward": 0.734375, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.19608965516090393, "rewards/accuracy_reward": 0.734375, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.19608965516090393, "rewards/answer_logprob_reward": 3.0625293254852295, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.7917594909667969, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.10732635110616684, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.09813292324542999, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.13923776149749756, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.07858544588088989, "rewards/len_reward/p75": 0.32227522134780884, "rewards/len_reward/var": 0.0758831650018692, "rewards/missing_response_penalty": -0.09375, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.08540575951337814, "smoothed_logprobs/min": -1.0, "step": 21, "value_loss": 2.7421875, "values": -0.7915716171264648, "values/max": 1.9507980346679688, "values/median": -0.8037109375, "values/min": -1.9878120422363281, "values/p25": -0.8037109375, "values/p75": -0.8037109375, "values/var": 0.12283516675233841 }, { "adam_stats/lr_effective_max": 3.338371971040033e-05, "adam_stats/lr_effective_mean": 1.3322469516463542e-10, "adam_stats/lr_effective_min": -3.350756742293015e-05, "adam_stats/m_t_max": 0.0037767805624753237, "adam_stats/m_t_mean": -3.4601509157905497e-11, "adam_stats/m_t_min": -0.002743084216490388, "adam_stats/v_t_max": 2.3205564502859488e-05, "adam_stats/v_t_mean": 1.5373737439344781e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.5316455696202531, "eval_accuracy_reward": 0.6633333563804626, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.22406914830207825, "eval_advantages": 0.6633333563804626, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.22406914830207825, "eval_all_logprobs": -0.014047027565538883, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -10.4375, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.023722998797893524, "eval_answer_logprob_reward": 3.0409231185913086, "eval_answer_logprob_reward/correct": 3.118717908859253, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.0664416253566742, "eval_answer_logprob_reward/incorrect": 2.8876430988311768, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.8325815200805664, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.27647578716278076, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.8325815200805664, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.14842931926250458, "eval_completion_length": 549.856689453125, "eval_completion_length/correct": 477.6783752441406, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 448.0, "eval_completion_length/correct/min": 114.0, "eval_completion_length/correct/p25": 300.5, "eval_completion_length/correct/p75": 617.5, "eval_completion_length/correct/var": 49479.5703125, "eval_completion_length/incorrect": 692.0692749023438, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 661.0, "eval_completion_length/incorrect/min": 180.0, "eval_completion_length/incorrect/p25": 440.0, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 83685.796875, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 504.0, "eval_completion_length/min": 114.0, "eval_completion_length/p25": 321.75, "eval_completion_length/p75": 742.75, "eval_completion_length/var": 71053.2890625, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 3613050.0, "eval_len_reward": 0.05123516172170639, "eval_len_reward/correct": 0.15512846410274506, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.19577352702617645, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.010729953646659851, "eval_len_reward/correct/p75": 0.3160973787307739, "eval_len_reward/correct/var": 0.054460883140563965, "eval_len_reward/incorrect": -0.15346552431583405, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.07097288966178894, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.27089521288871765, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03267636150121689, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.008183632977306843, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.09504184871912003, "eval_len_reward/p75": 0.2602965235710144, "eval_len_reward/var": 0.06833110004663467, "eval_mean_logprobs": -0.0135498046875, "eval_mean_logprobs/var": 0.000118255615234375, "eval_missing_response_penalty": -0.10000000149011612, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.2970297038555145, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.21089109778404236, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.09030100703239441, "eval_num_completions/total": 6300, "eval_pred_value_log_probs": -0.6350853443145752, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.68359375, "eval_pred_value_log_probs/min": -0.98046875, "eval_pred_value_log_probs/p25": -0.68359375, "eval_pred_value_log_probs/p75": -0.68359375, "eval_pred_value_log_probs/var": 0.027328863739967346, "eval_reward": 0.6633333563804626, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.22406914830207825, "eval_rewards/accuracy_reward": 0.6633333563804626, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.22406914830207825, "eval_rewards/answer_logprob_reward": 3.0409231185913086, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.8325815200805664, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.14842931926250458, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.05123516172170639, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.008183632977306843, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.09504184871912003, "eval_rewards/len_reward/p75": 0.2602965235710144, "eval_rewards/len_reward/var": 0.06833110004663467, "eval_rewards/missing_response_penalty": -0.10000000149011612, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.09030100703239441, "eval_runtime": 175.1157, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8495923280715942, "eval_values/max": 1.9913177490234375, "eval_values/median": -0.80859375, "eval_values/min": -1.9885406494140625, "eval_values/p25": -0.80859375, "eval_values/p75": -0.80859375, "eval_values/var": 0.12950313091278076, "step": 21 }, { "accuracy_reward": 0.7604166865348816, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.18313699960708618, "adam_stats/lr_effective_max": 3.071120227104984e-05, "adam_stats/lr_effective_mean": 5.2620817425630406e-11, "adam_stats/lr_effective_min": -3.083813135162927e-05, "adam_stats/m_t_max": 0.003942315466701984, "adam_stats/m_t_mean": -2.766375670204546e-11, "adam_stats/m_t_min": -0.002786028664559126, "adam_stats/v_t_max": 2.3211867301142775e-05, "adam_stats/v_t_mean": 1.5432596606884674e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7604166865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.18313699960708618, "all_logprobs": -0.012917487882077694, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -7.75, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.018993373960256577, "answer_logprob_reward": 3.0326926708221436, "answer_logprob_reward/correct": 3.098508596420288, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.05836192145943642, "answer_logprob_reward/incorrect": 2.823798894882202, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.7917594909667969, "answer_logprob_reward/incorrect/p25": 2.4849066734313965, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.2675291895866394, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.7917594909667969, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.12115713208913803, "clip_ratio": 0.0, "completion_length": 491.4895935058594, "completion_length/correct": 417.28765869140625, "completion_length/correct/max": 1021.0, "completion_length/correct/median": 386.0, "completion_length/correct/min": 138.0, "completion_length/correct/p25": 299.25, "completion_length/correct/p75": 500.5, "completion_length/correct/var": 28962.716796875, "completion_length/incorrect": 727.0, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 738.0, "completion_length/incorrect/min": 217.0, "completion_length/incorrect/p25": 458.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 83329.6875, "completion_length/max": 1024.0, "completion_length/median": 425.0, "completion_length/min": 138.0, "completion_length/p25": 316.0, "completion_length/p75": 594.5, "completion_length/var": 59186.8671875, "epoch": 0.5569620253164557, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 2390379.0, "grad_norm": 0.22681620717048645, "kl": 0.01324462890625, "learning_rate": 4.775675848247427e-06, "len_reward": 0.09454120695590973, "len_reward/correct": 0.18561342358589172, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.2097862809896469, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.03180480748414993, "len_reward/correct/p75": 0.36853471398353577, "len_reward/correct/var": 0.04988931864500046, "len_reward/incorrect": -0.1945141702890396, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.12032489478588104, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.43185335397720337, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.04208043962717056, "len_reward/max": 0.5, "len_reward/median": 0.08360301703214645, "len_reward/min": -0.5, "len_reward/p25": -0.0493905171751976, "len_reward/p75": 0.31959444284439087, "len_reward/var": 0.07425107806921005, "loss": -0.7599, "mean_logprobs": -0.01251220703125, "mean_logprobs/var": 7.963180541992188e-05, "missing_response_penalty": -0.0885416716337204, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.3695652186870575, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.2381642609834671, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.08112456649541855, "non_smoothed_logprobs/min": -17.5625, "num_completions/total": 4224, "policy_loss": -0.759887158870697, "policy_loss/max": 0.0012284929398447275, "policy_loss/median": -0.9994112253189087, "policy_loss/min": -0.9997952580451965, "policy_loss/p25": -0.9995701313018799, "policy_loss/p75": -0.9989782571792603, "policy_loss/var": 0.18312634527683258, "pred_value_log_probs": -0.6359685659408569, "pred_value_log_probs/max": -0.00982666015625, "pred_value_log_probs/median": -0.68359375, "pred_value_log_probs/min": -1.0625, "pred_value_log_probs/p25": -0.68359375, "pred_value_log_probs/p75": -0.68359375, "pred_value_log_probs/var": 0.028932176530361176, "reward": 0.7604166865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.18313699960708618, "rewards/accuracy_reward": 0.7604166865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.18313699960708618, "rewards/answer_logprob_reward": 3.0326926708221436, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.7917594909667969, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.12115713208913803, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.09454120695590973, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.08360301703214645, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.0493905171751976, "rewards/len_reward/p75": 0.31959444284439087, "rewards/len_reward/var": 0.07425107806921005, "rewards/missing_response_penalty": -0.0885416716337204, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.08112456649541855, "smoothed_logprobs/min": -1.0, "step": 22, "value_loss": 2.7421875, "values": -0.7965652346611023, "values/max": 1.78765869140625, "values/median": -0.80859375, "values/min": -1.9856138229370117, "values/p25": -0.80859375, "values/p75": -0.80859375, "values/var": 0.180428147315979 }, { "adam_stats/lr_effective_max": 3.071120227104984e-05, "adam_stats/lr_effective_mean": 5.2620817425630406e-11, "adam_stats/lr_effective_min": -3.083813135162927e-05, "adam_stats/m_t_max": 0.003942315466701984, "adam_stats/m_t_mean": -2.766375670204546e-11, "adam_stats/m_t_min": -0.002786028664559126, "adam_stats/v_t_max": 2.3211867301142775e-05, "adam_stats/v_t_mean": 1.5432596606884674e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.5569620253164557, "eval_accuracy_reward": 0.6333333253860474, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.23299887776374817, "eval_advantages": 0.6333333253860474, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.23299887776374817, "eval_all_logprobs": -0.01301631610840559, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -8.125, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.01946735754609108, "eval_answer_logprob_reward": 3.0356905460357666, "eval_answer_logprob_reward/correct": 3.106058120727539, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.07760242372751236, "eval_answer_logprob_reward/incorrect": 2.914146661758423, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.8325815200805664, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.22009778022766113, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.8325815200805664, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.137870654463768, "eval_completion_length": 538.27001953125, "eval_completion_length/correct": 465.2684326171875, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 454.0, "eval_completion_length/correct/min": 119.0, "eval_completion_length/correct/p25": 311.25, "eval_completion_length/correct/p75": 597.75, "eval_completion_length/correct/var": 40518.1484375, "eval_completion_length/incorrect": 664.3635864257812, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 619.0, "eval_completion_length/incorrect/min": 168.0, "eval_completion_length/incorrect/p25": 422.5, "eval_completion_length/incorrect/p75": 1009.5, "eval_completion_length/incorrect/var": 80659.25, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 484.0, "eval_completion_length/min": 119.0, "eval_completion_length/p25": 331.25, "eval_completion_length/p75": 686.0, "eval_completion_length/var": 64251.83203125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 3774531.0, "eval_len_reward": 0.03958354517817497, "eval_len_reward/correct": 0.15136002004146576, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.17757193744182587, "eval_len_reward/correct/min": -0.46847546100616455, "eval_len_reward/correct/p25": 0.008404039777815342, "eval_len_reward/correct/p75": 0.32655590772628784, "eval_len_reward/correct/var": 0.05165595933794975, "eval_len_reward/incorrect": -0.15348489582538605, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.060496795922517776, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.28577178716659546, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03317929431796074, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.12148172408342361, "eval_len_reward/p75": 0.23447690904140472, "eval_len_reward/var": 0.06640024483203888, "eval_mean_logprobs": -0.01336669921875, "eval_mean_logprobs/var": 9.918212890625e-05, "eval_missing_response_penalty": -0.08666666597127914, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.23636363446712494, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": 0.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.18215179443359375, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.07942028343677521, "eval_num_completions/total": 6600, "eval_pred_value_log_probs": -0.6274722814559937, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.67578125, "eval_pred_value_log_probs/min": -0.99609375, "eval_pred_value_log_probs/p25": -0.67578125, "eval_pred_value_log_probs/p75": -0.67578125, "eval_pred_value_log_probs/var": 0.02641473338007927, "eval_reward": 0.6333333253860474, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.23299887776374817, "eval_rewards/accuracy_reward": 0.6333333253860474, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.23299887776374817, "eval_rewards/answer_logprob_reward": 3.0356905460357666, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.8325815200805664, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.137870654463768, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.03958354517817497, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.12148172408342361, "eval_rewards/len_reward/p75": 0.23447690904140472, "eval_rewards/len_reward/var": 0.06640024483203888, "eval_rewards/missing_response_penalty": -0.08666666597127914, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.07942028343677521, "eval_runtime": 172.5601, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8610339760780334, "eval_values/max": 1.9912223815917969, "eval_values/median": -0.8212890625, "eval_values/min": -1.9895038604736328, "eval_values/p25": -0.8212890625, "eval_values/p75": -0.8212890625, "eval_values/var": 0.1292973905801773, "step": 22 }, { "accuracy_reward": 0.7395833730697632, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.19360819458961487, "adam_stats/lr_effective_max": 2.803716233756859e-05, "adam_stats/lr_effective_mean": 1.5093860883386157e-10, "adam_stats/lr_effective_min": -2.8081658456358127e-05, "adam_stats/m_t_max": 0.0031940797343850136, "adam_stats/m_t_mean": -1.9256929384425803e-11, "adam_stats/m_t_min": -0.0024208729155361652, "adam_stats/v_t_max": 2.3201186195365153e-05, "adam_stats/v_t_mean": 1.5473259609363743e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7395833730697632, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.19360819458961487, "all_logprobs": -0.01277267187833786, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -12.75, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.020812146365642548, "answer_logprob_reward": 2.9443728923797607, "answer_logprob_reward/correct": 3.101149082183838, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.057205647230148315, "answer_logprob_reward/incorrect": 2.4991276264190674, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 0.8754687309265137, "answer_logprob_reward/incorrect/p25": 1.7917594909667969, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.8718653917312622, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 0.8754687309265137, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.33607202768325806, "clip_ratio": 0.0, "completion_length": 554.6510620117188, "completion_length/correct": 474.6056213378906, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 385.0, "completion_length/correct/min": 155.0, "completion_length/correct/p25": 311.5, "completion_length/correct/p75": 595.25, "completion_length/correct/var": 50056.6640625, "completion_length/incorrect": 781.97998046875, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 827.0, "completion_length/incorrect/min": 205.0, "completion_length/incorrect/p25": 569.5, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 64115.1640625, "completion_length/max": 1024.0, "completion_length/median": 468.0, "completion_length/min": 155.0, "completion_length/p25": 344.0, "completion_length/p75": 758.5, "completion_length/var": 71693.1328125, "epoch": 0.5822784810126582, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 2496872.0, "grad_norm": 0.21534405648708344, "kl": 0.01239013671875, "learning_rate": 4.3288336709117246e-06, "len_reward": 0.09688381850719452, "len_reward/correct": 0.19501151144504547, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.22932861745357513, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.04732845723628998, "len_reward/correct/p75": 0.36181944608688354, "len_reward/correct/var": 0.05219462141394615, "len_reward/incorrect": -0.18179886043071747, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.14137931168079376, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.3377705514431, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03440873324871063, "len_reward/max": 0.5, "len_reward/median": 0.11601041257381439, "len_reward/min": -0.5, "len_reward/p25": -0.05947368964552879, "len_reward/p75": 0.3168730139732361, "len_reward/var": 0.07484813779592514, "loss": -0.7391, "mean_logprobs": -0.0126953125, "mean_logprobs/var": 8.630752563476562e-05, "missing_response_penalty": -0.1145833358168602, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.4399999976158142, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.2514285743236542, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.10198517888784409, "non_smoothed_logprobs/min": -17.3125, "num_completions/total": 4416, "policy_loss": -0.739088237285614, "policy_loss/max": 0.0009990279795601964, "policy_loss/median": -0.9994358420372009, "policy_loss/min": -0.9997886419296265, "policy_loss/p25": -0.9995859861373901, "policy_loss/p75": 0.0001322152093052864, "policy_loss/var": 0.19358791410923004, "pred_value_log_probs": -0.6088201403617859, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.67578125, "pred_value_log_probs/min": -1.234375, "pred_value_log_probs/p25": -0.67578125, "pred_value_log_probs/p75": -0.67578125, "pred_value_log_probs/var": 0.03758978098630905, "reward": 0.7395833730697632, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.19360819458961487, "rewards/accuracy_reward": 0.7395833730697632, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.19360819458961487, "rewards/answer_logprob_reward": 2.9443728923797607, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 0.8754687309265137, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.33607202768325806, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.09688381850719452, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.11601041257381439, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.05947368964552879, "rewards/len_reward/p75": 0.3168730139732361, "rewards/len_reward/var": 0.07484813779592514, "rewards/missing_response_penalty": -0.1145833358168602, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.10198517888784409, "smoothed_logprobs/min": -1.0, "step": 23, "value_loss": 2.96875, "values": -0.8436825275421143, "values/max": 1.195037841796875, "values/median": -0.8212890625, "values/min": -2.0, "values/p25": -0.8212890625, "values/p75": -0.8212890625, "values/var": 0.17299538850784302 }, { "adam_stats/lr_effective_max": 2.803716233756859e-05, "adam_stats/lr_effective_mean": 1.5093860883386157e-10, "adam_stats/lr_effective_min": -2.8081658456358127e-05, "adam_stats/m_t_max": 0.0031940797343850136, "adam_stats/m_t_mean": -1.9256929384425803e-11, "adam_stats/m_t_min": -0.0024208729155361652, "adam_stats/v_t_max": 2.3201186195365153e-05, "adam_stats/v_t_mean": 1.5473259609363743e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.5822784810126582, "eval_accuracy_reward": 0.6366666555404663, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.2320958524942398, "eval_advantages": 0.6366666555404663, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.2320958524942398, "eval_all_logprobs": -0.013486981391906738, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -6.5, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.01943836733698845, "eval_answer_logprob_reward": 3.047809362411499, "eval_answer_logprob_reward/correct": 3.12178111076355, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.06521151214838028, "eval_answer_logprob_reward/incorrect": 2.9181885719299316, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.4271163940429688, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.2903265655040741, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.4271163940429688, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.15592624247074127, "eval_completion_length": 526.106689453125, "eval_completion_length/correct": 457.0052490234375, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 426.0, "eval_completion_length/correct/min": 102.0, "eval_completion_length/correct/p25": 301.0, "eval_completion_length/correct/p75": 587.5, "eval_completion_length/correct/var": 41787.1015625, "eval_completion_length/incorrect": 647.192626953125, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 592.0, "eval_completion_length/incorrect/min": 178.0, "eval_completion_length/incorrect/p25": 402.0, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 80823.3046875, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 463.0, "eval_completion_length/min": 102.0, "eval_completion_length/p25": 329.75, "eval_completion_length/p75": 684.25, "eval_completion_length/var": 64142.58203125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 3932363.0, "eval_len_reward": 0.04952501505613327, "eval_len_reward/correct": 0.15692557394504547, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.17270317673683167, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.0006625441601499915, "eval_len_reward/correct/p75": 0.32966095209121704, "eval_len_reward/correct/var": 0.051744624972343445, "eval_len_reward/incorrect": -0.13867229223251343, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.014140008948743343, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.2403801530599594, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03106965683400631, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.09050571173429489, "eval_len_reward/p75": 0.2296256422996521, "eval_len_reward/var": 0.06438378244638443, "eval_mean_logprobs": -0.013671875, "eval_mean_logprobs/var": 0.00010061264038085938, "eval_missing_response_penalty": -0.09666667133569717, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.26605504751205444, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.19707782566547394, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.0876142680644989, "eval_num_completions/total": 6900, "eval_pred_value_log_probs": -0.624318540096283, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.671875, "eval_pred_value_log_probs/min": -0.984375, "eval_pred_value_log_probs/p25": -0.671875, "eval_pred_value_log_probs/p75": -0.671875, "eval_pred_value_log_probs/var": 0.02601056918501854, "eval_reward": 0.6366666555404663, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.2320958524942398, "eval_rewards/accuracy_reward": 0.6366666555404663, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.2320958524942398, "eval_rewards/answer_logprob_reward": 3.047809362411499, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.4271163940429688, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.15592624247074127, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.04952501505613327, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.09050571173429489, "eval_rewards/len_reward/p75": 0.2296256422996521, "eval_rewards/len_reward/var": 0.06438378244638443, "eval_rewards/missing_response_penalty": -0.09666667133569717, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.0876142680644989, "eval_runtime": 168.7882, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8640043139457703, "eval_values/max": 1.9913520812988281, "eval_values/median": -0.82568359375, "eval_values/min": -1.9897537231445312, "eval_values/p25": -0.82568359375, "eval_values/p75": -0.82568359375, "eval_values/var": 0.12934480607509613, "step": 23 }, { "accuracy_reward": 0.7916666865348816, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.16579407453536987, "adam_stats/lr_effective_max": 2.521297210478224e-05, "adam_stats/lr_effective_mean": 6.384962003558314e-11, "adam_stats/lr_effective_min": -2.474541906849481e-05, "adam_stats/m_t_max": 0.002732001943513751, "adam_stats/m_t_mean": -1.6766574240101306e-11, "adam_stats/m_t_min": -0.0020878624636679888, "adam_stats/v_t_max": 2.3180020434665494e-05, "adam_stats/v_t_mean": 1.5528077955406783e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7916666865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.16579407453536987, "all_logprobs": -0.014185355976223946, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -6.84375, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.022149622440338135, "answer_logprob_reward": 3.0299673080444336, "answer_logprob_reward/correct": 3.1062088012695312, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.0537831075489521, "answer_logprob_reward/incorrect": 2.740250587463379, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.7917594909667969, "answer_logprob_reward/incorrect/p25": 2.079441547393799, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.316007137298584, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.7917594909667969, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.1292486935853958, "clip_ratio": 0.0, "completion_length": 504.203125, "completion_length/correct": 437.631591796875, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 388.0, "completion_length/correct/min": 127.0, "completion_length/correct/p25": 291.0, "completion_length/correct/p75": 518.75, "completion_length/correct/var": 45088.0546875, "completion_length/incorrect": 757.1749877929688, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 835.0, "completion_length/incorrect/min": 223.0, "completion_length/incorrect/p25": 451.5, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 84737.8984375, "completion_length/max": 1024.0, "completion_length/median": 422.0, "completion_length/min": 127.0, "completion_length/p25": 307.75, "completion_length/p75": 661.5, "completion_length/var": 69876.921875, "epoch": 0.6075949367088608, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 2593679.0, "grad_norm": 0.23429417610168457, "kl": 0.013427734375, "learning_rate": 3.887395330218429e-06, "len_reward": 0.10179547965526581, "len_reward/correct": 0.18434910476207733, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.22507065534591675, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.0629768818616867, "len_reward/correct/p75": 0.3572998344898224, "len_reward/correct/var": 0.055906061083078384, "len_reward/incorrect": -0.21190838515758514, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.2201894074678421, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.3721355199813843, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.037402331829071045, "len_reward/max": 0.5, "len_reward/median": 0.1524021029472351, "len_reward/min": -0.5, "len_reward/p25": -0.05030941963195801, "len_reward/p75": 0.3255058526992798, "len_reward/var": 0.07786809653043747, "loss": -0.7911, "mean_logprobs": -0.0133056640625, "mean_logprobs/var": 0.00011014938354492188, "missing_response_penalty": -0.0833333358168602, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.4000000059604645, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.24615386128425598, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.07678883522748947, "non_smoothed_logprobs/min": -17.25, "num_completions/total": 4608, "policy_loss": -0.7911297678947449, "policy_loss/max": 0.0009242998785339296, "policy_loss/median": -0.999406099319458, "policy_loss/min": -0.9998186826705933, "policy_loss/p25": -0.999532163143158, "policy_loss/p75": -0.9991399645805359, "policy_loss/var": 0.1657712310552597, "pred_value_log_probs": -0.605526328086853, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.671875, "pred_value_log_probs/min": -1.21875, "pred_value_log_probs/p25": -0.671875, "pred_value_log_probs/p75": -0.671875, "pred_value_log_probs/var": 0.03997424617409706, "reward": 0.7916666865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.16579407453536987, "rewards/accuracy_reward": 0.7916666865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.16579407453536987, "rewards/answer_logprob_reward": 3.0299673080444336, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.7917594909667969, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.1292486935853958, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.10179547965526581, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.1524021029472351, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.05030941963195801, "rewards/len_reward/p75": 0.3255058526992798, "rewards/len_reward/var": 0.07786809653043747, "rewards/missing_response_penalty": -0.0833333358168602, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.07678883522748947, "smoothed_logprobs/min": -1.0, "step": 24, "value_loss": 3.0546875, "values": -0.873255729675293, "values/max": 1.30712890625, "values/median": -0.82568359375, "values/min": -1.9909396171569824, "values/p25": -0.82568359375, "values/p75": -0.82568359375, "values/var": 0.20940503478050232 }, { "adam_stats/lr_effective_max": 2.521297210478224e-05, "adam_stats/lr_effective_mean": 6.384962003558314e-11, "adam_stats/lr_effective_min": -2.474541906849481e-05, "adam_stats/m_t_max": 0.002732001943513751, "adam_stats/m_t_mean": -1.6766574240101306e-11, "adam_stats/m_t_min": -0.0020878624636679888, "adam_stats/v_t_max": 2.3180020434665494e-05, "adam_stats/v_t_mean": 1.5528077955406783e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.6075949367088608, "eval_accuracy_reward": 0.6466667056083679, "eval_accuracy_reward/correct": 0.9999999403953552, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.22925306856632233, "eval_advantages": 0.6466667056083679, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.22925306856632233, "eval_all_logprobs": -0.01458882074803114, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -8.75, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.023142628371715546, "eval_answer_logprob_reward": 3.075014114379883, "eval_answer_logprob_reward/correct": 3.126248359680176, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.06745291501283646, "eval_answer_logprob_reward/incorrect": 2.9812450408935547, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.8325815200805664, "eval_answer_logprob_reward/incorrect/p25": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.19945694506168365, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.8325815200805664, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.11840354651212692, "eval_completion_length": 535.1699829101562, "eval_completion_length/correct": 472.83502197265625, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 435.0, "eval_completion_length/correct/min": 118.0, "eval_completion_length/correct/p25": 302.0, "eval_completion_length/correct/p75": 603.75, "eval_completion_length/correct/var": 48038.92578125, "eval_completion_length/incorrect": 649.2546997070312, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 628.0, "eval_completion_length/incorrect/min": 170.0, "eval_completion_length/incorrect/p25": 416.5, "eval_completion_length/incorrect/p75": 942.25, "eval_completion_length/incorrect/var": 77611.6640625, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 483.0, "eval_completion_length/min": 118.0, "eval_completion_length/p25": 321.0, "eval_completion_length/p75": 684.5, "eval_completion_length/var": 65398.58203125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 4092914.0, "eval_len_reward": 0.061788979917764664, "eval_len_reward/correct": 0.15681885182857513, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.1812652051448822, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.0326542966067791, "eval_len_reward/correct/p75": 0.34234336018562317, "eval_len_reward/correct/var": 0.05734947696328163, "eval_len_reward/incorrect": -0.11213362216949463, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": 0.0, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.15305142104625702, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.028778472915291786, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.005759736523032188, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.0507662259042263, "eval_len_reward/p75": 0.2570588290691376, "eval_len_reward/var": 0.06370748579502106, "eval_mean_logprobs": -0.013671875, "eval_mean_logprobs/var": 0.0004291534423828125, "eval_missing_response_penalty": -0.07999999821186066, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.22641509771347046, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": 0.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.17681941390037537, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.07384615391492844, "eval_num_completions/total": 7200, "eval_pred_value_log_probs": -0.620497465133667, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.66796875, "eval_pred_value_log_probs/min": -1.0234375, "eval_pred_value_log_probs/p25": -0.66796875, "eval_pred_value_log_probs/p75": -0.66796875, "eval_pred_value_log_probs/var": 0.026062848046422005, "eval_reward": 0.6466667056083679, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.22925306856632233, "eval_rewards/accuracy_reward": 0.6466667056083679, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.22925306856632233, "eval_rewards/answer_logprob_reward": 3.075014114379883, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.8325815200805664, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.11840354651212692, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.061788979917764664, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.005759736523032188, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.0507662259042263, "eval_rewards/len_reward/p75": 0.2570588290691376, "eval_rewards/len_reward/var": 0.06370748579502106, "eval_rewards/missing_response_penalty": -0.07999999821186066, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.07384615391492844, "eval_runtime": 165.1828, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8670269250869751, "eval_values/max": 1.991119384765625, "eval_values/median": -0.82958984375, "eval_values/min": -1.997873306274414, "eval_values/p25": -0.82958984375, "eval_values/p75": -0.82958984375, "eval_values/var": 0.12986624240875244, "step": 24 }, { "accuracy_reward": 0.7864583730697632, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.16882090270519257, "adam_stats/lr_effective_max": 2.1769656086689793e-05, "adam_stats/lr_effective_mean": 5.913265710422166e-11, "adam_stats/lr_effective_min": -2.242199843749404e-05, "adam_stats/m_t_max": 0.0026235966943204403, "adam_stats/m_t_mean": -2.350836886266805e-11, "adam_stats/m_t_min": -0.0016450312687084079, "adam_stats/v_t_max": 2.3159556803875603e-05, "adam_stats/v_t_mean": 1.558106725238484e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7864583730697632, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.16882090270519257, "all_logprobs": -0.013827561400830746, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -7.3125, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.022680440917611122, "answer_logprob_reward": 3.028197765350342, "answer_logprob_reward/correct": 3.0954270362854004, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.4849066734313965, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.05078166723251343, "answer_logprob_reward/incorrect": 2.7805967330932617, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.7917594909667969, "answer_logprob_reward/incorrect/p25": 2.4849066734313965, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.2833702266216278, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.7917594909667969, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.11595863103866577, "clip_ratio": 0.0, "completion_length": 514.9739990234375, "completion_length/correct": 456.0331115722656, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 396.0, "completion_length/correct/min": 120.0, "completion_length/correct/p25": 304.5, "completion_length/correct/p75": 590.0, "completion_length/correct/var": 43072.3671875, "completion_length/incorrect": 732.0487670898438, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 735.0, "completion_length/incorrect/min": 254.0, "completion_length/incorrect/p25": 435.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 86421.203125, "completion_length/max": 1024.0, "completion_length/median": 435.0, "completion_length/min": 120.0, "completion_length/p25": 317.75, "completion_length/p75": 653.0, "completion_length/var": 64786.703125, "epoch": 0.6329113924050633, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 2692554.0, "grad_norm": 0.22061339020729065, "kl": 0.01324462890625, "learning_rate": 3.4549150281252635e-06, "len_reward": 0.1224907785654068, "len_reward/correct": 0.20522892475128174, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.2400950938463211, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.07591985166072845, "len_reward/correct/p75": 0.36015838384628296, "len_reward/correct/var": 0.04885726049542427, "len_reward/incorrect": -0.18222780525684357, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.1176169365644455, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.3276824951171875, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.035905227065086365, "len_reward/max": 0.5, "len_reward/median": 0.17908082902431488, "len_reward/min": -0.5, "len_reward/p25": -0.03324875980615616, "len_reward/p75": 0.34264105558395386, "len_reward/var": 0.07123284786939621, "loss": -0.7859, "mean_logprobs": -0.0135498046875, "mean_logprobs/var": 9.870529174804688e-05, "missing_response_penalty": -0.0833333358168602, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.39024388790130615, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.24390244483947754, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.07678883522748947, "non_smoothed_logprobs/min": -17.25, "num_completions/total": 4800, "policy_loss": -0.7859289646148682, "policy_loss/max": 0.0010629133321344852, "policy_loss/median": -0.9994313716888428, "policy_loss/min": -0.9997806549072266, "policy_loss/p25": -0.9995715618133545, "policy_loss/p75": -0.9991579055786133, "policy_loss/var": 0.1688242405653, "pred_value_log_probs": -0.6209506988525391, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.66796875, "pred_value_log_probs/min": -1.1328125, "pred_value_log_probs/p25": -0.66796875, "pred_value_log_probs/p75": -0.66796875, "pred_value_log_probs/var": 0.032860416918992996, "reward": 0.7864583730697632, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.16882090270519257, "rewards/accuracy_reward": 0.7864583730697632, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.16882090270519257, "rewards/answer_logprob_reward": 3.028197765350342, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.7917594909667969, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.11595863103866577, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.1224907785654068, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.17908082902431488, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.03324875980615616, "rewards/len_reward/p75": 0.34264105558395386, "rewards/len_reward/var": 0.07123284786939621, "rewards/missing_response_penalty": -0.0833333358168602, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.07678883522748947, "smoothed_logprobs/min": -1.0, "step": 25, "value_loss": 3.015625, "values": -0.8420132994651794, "values/max": 1.9203028678894043, "values/median": -0.82958984375, "values/min": -2.0, "values/p25": -0.82958984375, "values/p75": -0.82958984375, "values/var": 0.1504109799861908 }, { "adam_stats/lr_effective_max": 2.1769656086689793e-05, "adam_stats/lr_effective_mean": 5.913265710422166e-11, "adam_stats/lr_effective_min": -2.242199843749404e-05, "adam_stats/m_t_max": 0.0026235966943204403, "adam_stats/m_t_mean": -2.350836886266805e-11, "adam_stats/m_t_min": -0.0016450312687084079, "adam_stats/v_t_max": 2.3159556803875603e-05, "adam_stats/v_t_mean": 1.558106725238484e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.6329113924050633, "eval_accuracy_reward": 0.643333375453949, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.23022297024726868, "eval_advantages": 0.643333375453949, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.23022297024726868, "eval_all_logprobs": -0.012885897420346737, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -6.25, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.01903092861175537, "eval_answer_logprob_reward": 3.0509700775146484, "eval_answer_logprob_reward/correct": 3.1185855865478516, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.07136764377355576, "eval_answer_logprob_reward/incorrect": 2.929009199142456, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.272965669631958, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.3084012567996979, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.272965669631958, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.1634349822998047, "eval_completion_length": 524.856689453125, "eval_completion_length/correct": 461.5336608886719, "eval_completion_length/correct/max": 1008.0, "eval_completion_length/correct/median": 441.0, "eval_completion_length/correct/min": 119.0, "eval_completion_length/correct/p25": 307.0, "eval_completion_length/correct/p75": 582.0, "eval_completion_length/correct/var": 40730.69140625, "eval_completion_length/incorrect": 639.0747680664062, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 580.0, "eval_completion_length/incorrect/min": 159.0, "eval_completion_length/incorrect/p25": 381.0, "eval_completion_length/incorrect/p75": 997.5, "eval_completion_length/incorrect/var": 86812.4375, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 467.0, "eval_completion_length/min": 119.0, "eval_completion_length/p25": 332.0, "eval_completion_length/p75": 670.0, "eval_completion_length/var": 64187.9609375, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 4250371.0, "eval_len_reward": 0.037196140736341476, "eval_len_reward/correct": 0.13615931570529938, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.14834243059158325, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": -0.023277467116713524, "eval_len_reward/correct/p75": 0.3070557415485382, "eval_len_reward/correct/var": 0.05235496163368225, "eval_len_reward/incorrect": -0.1413075029850006, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": 0.0, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.2787383198738098, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03431173041462898, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.10086379945278168, "eval_len_reward/p75": 0.21484653651714325, "eval_len_reward/var": 0.06350763142108917, "eval_mean_logprobs": -0.01312255859375, "eval_mean_logprobs/var": 9.107589721679688e-05, "eval_missing_response_penalty": -0.07999999821186066, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.22429905831813812, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": 0.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.17563040554523468, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.07384615391492844, "eval_num_completions/total": 7500, "eval_pred_value_log_probs": -0.624106228351593, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.671875, "eval_pred_value_log_probs/min": -1.015625, "eval_pred_value_log_probs/p25": -0.671875, "eval_pred_value_log_probs/p75": -0.671875, "eval_pred_value_log_probs/var": 0.026507392525672913, "eval_reward": 0.643333375453949, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.23022297024726868, "eval_rewards/accuracy_reward": 0.643333375453949, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.23022297024726868, "eval_rewards/answer_logprob_reward": 3.0509700775146484, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.272965669631958, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.1634349822998047, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.037196140736341476, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.10086379945278168, "eval_rewards/len_reward/p75": 0.21484653651714325, "eval_rewards/len_reward/var": 0.06350763142108917, "eval_rewards/missing_response_penalty": -0.07999999821186066, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.07384615391492844, "eval_runtime": 166.0003, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8647846579551697, "eval_values/max": 1.9911270141601562, "eval_values/median": -0.82568359375, "eval_values/min": -1.9895076751708984, "eval_values/p25": -0.82568359375, "eval_values/p75": -0.82568359375, "eval_values/var": 0.1284439116716385, "step": 25 }, { "accuracy_reward": 0.7916666865348816, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.16579407453536987, "adam_stats/lr_effective_max": 1.9133012756356038e-05, "adam_stats/lr_effective_mean": 6.889214893002205e-11, "adam_stats/lr_effective_min": -1.8895842003985308e-05, "adam_stats/m_t_max": 0.00241082813590765, "adam_stats/m_t_mean": -2.992975312032797e-11, "adam_stats/m_t_min": -0.0015358412638306618, "adam_stats/v_t_max": 2.3136642994359136e-05, "adam_stats/v_t_mean": 1.5624625074664444e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7916666865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.16579407453536987, "all_logprobs": -0.01206886488944292, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -9.0, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.02019508183002472, "answer_logprob_reward": 3.0381431579589844, "answer_logprob_reward/correct": 3.1016485691070557, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.05628438666462898, "answer_logprob_reward/incorrect": 2.7968227863311768, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.7917594909667969, "answer_logprob_reward/incorrect/p25": 2.4849066734313965, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.319070041179657, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.7917594909667969, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.12505289912223816, "clip_ratio": 0.0, "completion_length": 506.890625, "completion_length/correct": 430.7105407714844, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 408.0, "completion_length/correct/min": 122.0, "completion_length/correct/p25": 309.25, "completion_length/correct/p75": 521.5, "completion_length/correct/var": 30985.37109375, "completion_length/incorrect": 796.375, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 898.0, "completion_length/incorrect/min": 337.0, "completion_length/incorrect/p25": 561.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 63653.1171875, "completion_length/max": 1024.0, "completion_length/median": 431.0, "completion_length/min": 122.0, "completion_length/p25": 334.25, "completion_length/p75": 656.25, "completion_length/var": 59661.93359375, "epoch": 0.6582278481012658, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 2789877.0, "grad_norm": 0.22772380709648132, "kl": 0.013031005859375, "learning_rate": 3.0348748417303826e-06, "len_reward": 0.11382603645324707, "len_reward/correct": 0.20018017292022705, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.2098509967327118, "len_reward/correct/min": -0.48064124584198, "len_reward/correct/p25": 0.10171883553266525, "len_reward/correct/p75": 0.37242966890335083, "len_reward/correct/var": 0.04144986718893051, "len_reward/incorrect": -0.21431966125965118, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.1825396865606308, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.38078171014785767, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03748637065291405, "len_reward/max": 0.5, "len_reward/median": 0.1651490032672882, "len_reward/min": -0.5, "len_reward/p25": -0.010968567803502083, "len_reward/p75": 0.3135233521461487, "len_reward/var": 0.06890864670276642, "loss": -0.7911, "mean_logprobs": -0.01165771484375, "mean_logprobs/var": 7.104873657226562e-05, "missing_response_penalty": -0.078125, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.375, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.24038463830947876, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.07239855825901031, "non_smoothed_logprobs/min": -17.4375, "num_completions/total": 4992, "policy_loss": -0.7911450266838074, "policy_loss/max": 0.0009366496815346181, "policy_loss/median": -0.9994315505027771, "policy_loss/min": -0.9998270869255066, "policy_loss/p25": -0.9995665550231934, "policy_loss/p75": -0.9991076588630676, "policy_loss/var": 0.1657780408859253, "pred_value_log_probs": -0.6421036720275879, "pred_value_log_probs/max": -5.960464477539062e-07, "pred_value_log_probs/median": -0.671875, "pred_value_log_probs/min": -1.3203125, "pred_value_log_probs/p25": -0.671875, "pred_value_log_probs/p75": -0.671875, "pred_value_log_probs/var": 0.029975350946187973, "reward": 0.7916666865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.16579407453536987, "rewards/accuracy_reward": 0.7916666865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.16579407453536987, "rewards/answer_logprob_reward": 3.0381431579589844, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.7917594909667969, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.12505289912223816, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.11382603645324707, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.1651490032672882, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.010968567803502083, "rewards/len_reward/p75": 0.3135233521461487, "rewards/len_reward/var": 0.06890864670276642, "rewards/missing_response_penalty": -0.078125, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.07239855825901031, "smoothed_logprobs/min": -1.0, "step": 26, "value_loss": 2.7890625, "values": -0.8395852446556091, "values/max": 1.30419921875, "values/median": -0.82568359375, "values/min": -2.000000476837158, "values/p25": -0.82568359375, "values/p75": -0.82568359375, "values/var": 0.12676073610782623 }, { "adam_stats/lr_effective_max": 1.9133012756356038e-05, "adam_stats/lr_effective_mean": 6.889214893002205e-11, "adam_stats/lr_effective_min": -1.8895842003985308e-05, "adam_stats/m_t_max": 0.00241082813590765, "adam_stats/m_t_mean": -2.992975312032797e-11, "adam_stats/m_t_min": -0.0015358412638306618, "adam_stats/v_t_max": 2.3136642994359136e-05, "adam_stats/v_t_mean": 1.5624625074664444e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.6582278481012658, "eval_accuracy_reward": 0.643333375453949, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.23022295534610748, "eval_advantages": 0.643333375453949, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.23022295534610748, "eval_all_logprobs": -0.013339114375412464, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -6.65625, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.019734643399715424, "eval_answer_logprob_reward": 3.0400807857513428, "eval_answer_logprob_reward/correct": 3.112013101577759, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.06993844360113144, "eval_answer_logprob_reward/incorrect": 2.9103331565856934, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.4271163940429688, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.32598403096199036, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.4271163940429688, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.16984079778194427, "eval_completion_length": 524.2467041015625, "eval_completion_length/correct": 455.0051574707031, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 408.0, "eval_completion_length/correct/min": 102.0, "eval_completion_length/correct/p25": 289.0, "eval_completion_length/correct/p75": 595.0, "eval_completion_length/correct/var": 42489.26171875, "eval_completion_length/incorrect": 649.14013671875, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 606.0, "eval_completion_length/incorrect/min": 159.0, "eval_completion_length/incorrect/p25": 428.5, "eval_completion_length/incorrect/p75": 994.5, "eval_completion_length/incorrect/var": 76860.953125, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 467.0, "eval_completion_length/min": 102.0, "eval_completion_length/p25": 318.5, "eval_completion_length/p75": 691.25, "eval_completion_length/var": 63209.1640625, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 4407645.0, "eval_len_reward": 0.04974249377846718, "eval_len_reward/correct": 0.1533033549785614, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.1989329308271408, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.006053268909454346, "eval_len_reward/correct/p75": 0.3202614486217499, "eval_len_reward/correct/var": 0.05350358784198761, "eval_len_reward/incorrect": -0.1370541900396347, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.019054878503084183, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.2640671133995056, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03244072198867798, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.08369508385658264, "eval_len_reward/p75": 0.2448240965604782, "eval_len_reward/var": 0.06526707112789154, "eval_mean_logprobs": -0.01336669921875, "eval_mean_logprobs/var": 9.584426879882812e-05, "eval_missing_response_penalty": -0.0833333358168602, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.23364485800266266, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": 0.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.18074414134025574, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.07664437592029572, "eval_num_completions/total": 7800, "eval_pred_value_log_probs": -0.6271107792854309, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.67578125, "eval_pred_value_log_probs/min": -1.0546875, "eval_pred_value_log_probs/p25": -0.67578125, "eval_pred_value_log_probs/p75": -0.67578125, "eval_pred_value_log_probs/var": 0.026666205376386642, "eval_reward": 0.643333375453949, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.23022295534610748, "eval_rewards/accuracy_reward": 0.643333375453949, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.23022295534610748, "eval_rewards/answer_logprob_reward": 3.0400807857513428, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.4271163940429688, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.16984079778194427, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.04974249377846718, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.08369508385658264, "eval_rewards/len_reward/p75": 0.2448240965604782, "eval_rewards/len_reward/var": 0.06526707112789154, "eval_rewards/missing_response_penalty": -0.0833333358168602, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.07664437592029572, "eval_runtime": 164.0283, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8602791428565979, "eval_values/max": 1.9909400939941406, "eval_values/median": -0.8203125, "eval_values/min": -1.9979867935180664, "eval_values/p25": -0.8203125, "eval_values/p75": -0.8203125, "eval_values/var": 0.128428652882576, "step": 26 }, { "accuracy_reward": 0.75, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.75, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.1884816735982895, "adam_stats/lr_effective_max": 1.6164200133061968e-05, "adam_stats/lr_effective_mean": 8.531608752804232e-11, "adam_stats/lr_effective_min": -1.6297251931973733e-05, "adam_stats/m_t_max": 0.0021214892622083426, "adam_stats/m_t_mean": -2.8601409435280445e-11, "adam_stats/m_t_min": -0.001608126563951373, "adam_stats/v_t_max": 2.311374009877909e-05, "adam_stats/v_t_mean": 1.5665066899900326e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.75, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.75, "advantages/p75": 1.0, "advantages/var": 0.1884816735982895, "all_logprobs": -0.01235611829906702, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -6.25, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.018512871116399765, "answer_logprob_reward": 2.921663284301758, "answer_logprob_reward/correct": 3.0877768993377686, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.06415103375911713, "answer_logprob_reward/incorrect": 2.4233226776123047, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 2.4849066734313965, "answer_logprob_reward/incorrect/min": 1.0986123085021973, "answer_logprob_reward/incorrect/p25": 2.007521152496338, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.6102849245071411, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.0986123085021973, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.28141871094703674, "clip_ratio": 0.0, "completion_length": 571.3177490234375, "completion_length/correct": 477.0069580078125, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 476.0, "completion_length/correct/min": 114.0, "completion_length/correct/p25": 303.5, "completion_length/correct/p75": 595.5, "completion_length/correct/var": 48252.68359375, "completion_length/incorrect": 854.25, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 1024.0, "completion_length/incorrect/min": 198.0, "completion_length/incorrect/p25": 720.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 59079.8515625, "completion_length/max": 1024.0, "completion_length/median": 522.0, "completion_length/min": 114.0, "completion_length/p25": 337.25, "completion_length/p75": 827.75, "completion_length/var": 77487.59375, "epoch": 0.6835443037974683, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 2899570.0, "grad_norm": 0.20783691108226776, "kl": 0.012664794921875, "learning_rate": 2.6306566876350072e-06, "len_reward": 0.06911170482635498, "len_reward/correct": 0.16831690073013306, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.20964567363262177, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.025169409811496735, "len_reward/correct/p75": 0.3304530382156372, "len_reward/correct/var": 0.05686808004975319, "len_reward/incorrect": -0.22850386798381805, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.24156877398490906, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.387567937374115, "len_reward/incorrect/p75": -0.07083163410425186, "len_reward/incorrect/var": 0.030994582921266556, "len_reward/max": 0.5, "len_reward/median": 0.10745261609554291, "len_reward/min": -0.5, "len_reward/p25": -0.12372851371765137, "len_reward/p75": 0.2982177138328552, "len_reward/var": 0.0798831582069397, "loss": -0.7495, "mean_logprobs": -0.012451171875, "mean_logprobs/var": 6.818771362304688e-05, "missing_response_penalty": -0.1302083432674408, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.5208333730697632, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": -1.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.25487589836120605, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.11384708434343338, "non_smoothed_logprobs/min": -17.75, "num_completions/total": 5184, "policy_loss": -0.749494194984436, "policy_loss/max": 0.0013085182290524244, "policy_loss/median": -0.9994063377380371, "policy_loss/min": -0.9998708367347717, "policy_loss/p25": -0.9995794296264648, "policy_loss/p75": -0.7491034865379333, "policy_loss/var": 0.18845784664154053, "pred_value_log_probs": -0.6104012727737427, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.67578125, "pred_value_log_probs/min": -1.1328125, "pred_value_log_probs/p25": -0.67578125, "pred_value_log_probs/p75": -0.67578125, "pred_value_log_probs/var": 0.04308498278260231, "reward": 0.75, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.75, "reward/p75": 1.0, "reward/var": 0.1884816735982895, "rewards/accuracy_reward": 0.75, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.75, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.1884816735982895, "rewards/answer_logprob_reward": 2.921663284301758, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.0986123085021973, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.28141871094703674, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.06911170482635498, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.10745261609554291, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.12372851371765137, "rewards/len_reward/p75": 0.2982177138328552, "rewards/len_reward/var": 0.0798831582069397, "rewards/missing_response_penalty": -0.1302083432674408, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.11384708434343338, "smoothed_logprobs/min": -1.0, "step": 27, "value_loss": 2.828125, "values": -0.8203917741775513, "values/max": 1.244140625, "values/median": -0.8203125, "values/min": -1.9633731842041016, "values/p25": -0.8203125, "values/p75": -0.8203125, "values/var": 0.22798120975494385 }, { "adam_stats/lr_effective_max": 1.6164200133061968e-05, "adam_stats/lr_effective_mean": 8.531608752804232e-11, "adam_stats/lr_effective_min": -1.6297251931973733e-05, "adam_stats/m_t_max": 0.0021214892622083426, "adam_stats/m_t_mean": -2.8601409435280445e-11, "adam_stats/m_t_min": -0.001608126563951373, "adam_stats/v_t_max": 2.311374009877909e-05, "adam_stats/v_t_mean": 1.5665066899900326e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.6835443037974683, "eval_accuracy_reward": 0.6366666555404663, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.2320958822965622, "eval_advantages": 0.6366666555404663, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.2320958822965622, "eval_all_logprobs": -0.012918404303491116, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -8.75, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.019511204212903976, "eval_answer_logprob_reward": 3.075014114379883, "eval_answer_logprob_reward/correct": 3.1354079246520996, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.525728702545166, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.051156532019376755, "eval_answer_logprob_reward/incorrect": 2.9691860675811768, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.4271163940429688, "eval_answer_logprob_reward/incorrect/p25": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.2690730094909668, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.4271163940429688, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.13611048460006714, "eval_completion_length": 522.8699951171875, "eval_completion_length/correct": 452.8638916015625, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 415.0, "eval_completion_length/correct/min": 119.0, "eval_completion_length/correct/p25": 292.0, "eval_completion_length/correct/p75": 570.0, "eval_completion_length/correct/var": 42531.328125, "eval_completion_length/incorrect": 645.541259765625, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 590.0, "eval_completion_length/incorrect/min": 170.0, "eval_completion_length/incorrect/p25": 412.0, "eval_completion_length/incorrect/p75": 924.0, "eval_completion_length/incorrect/var": 77136.40625, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 469.0, "eval_completion_length/min": 119.0, "eval_completion_length/p25": 314.75, "eval_completion_length/p75": 670.5, "eval_completion_length/var": 63505.03515625, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 4564506.0, "eval_len_reward": 0.04475758224725723, "eval_len_reward/correct": 0.15148313343524933, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.1757829785346985, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.015369649976491928, "eval_len_reward/correct/p75": 0.31522229313850403, "eval_len_reward/correct/var": 0.052471134811639786, "eval_len_reward/incorrect": -0.14225688576698303, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.006225680932402611, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.29405033588409424, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.031476959586143494, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.0766662061214447, "eval_len_reward/p75": 0.23568587005138397, "eval_len_reward/var": 0.0647384375333786, "eval_mean_logprobs": -0.0128173828125, "eval_mean_logprobs/var": 7.82012939453125e-05, "eval_missing_response_penalty": -0.0833333358168602, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.2293577939271927, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": 0.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.17838941514492035, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.07664437592029572, "eval_num_completions/total": 8100, "eval_pred_value_log_probs": -0.6281352639198303, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.67578125, "eval_pred_value_log_probs/min": -1.0625, "eval_pred_value_log_probs/p25": -0.67578125, "eval_pred_value_log_probs/p75": -0.67578125, "eval_pred_value_log_probs/var": 0.026235129684209824, "eval_reward": 0.6366666555404663, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.2320958822965622, "eval_rewards/accuracy_reward": 0.6366666555404663, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.2320958822965622, "eval_rewards/answer_logprob_reward": 3.075014114379883, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.4271163940429688, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.13611048460006714, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.04475758224725723, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.0766662061214447, "eval_rewards/len_reward/p75": 0.23568587005138397, "eval_rewards/len_reward/var": 0.0647384375333786, "eval_rewards/missing_response_penalty": -0.0833333358168602, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.07664437592029572, "eval_runtime": 172.4692, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8561747670173645, "eval_values/max": 1.9912757873535156, "eval_values/median": -0.81640625, "eval_values/min": -1.997899055480957, "eval_values/p25": -0.81640625, "eval_values/p75": -0.81640625, "eval_values/var": 0.1274670958518982, "step": 27 }, { "accuracy_reward": 0.7864583730697632, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.16882090270519257, "adam_stats/lr_effective_max": 1.4186392036208417e-05, "adam_stats/lr_effective_mean": 7.851148897675131e-11, "adam_stats/lr_effective_min": -1.3711563042306807e-05, "adam_stats/m_t_max": 0.002043617656454444, "adam_stats/m_t_mean": -2.1564961621978185e-11, "adam_stats/m_t_min": -0.0016183295520022511, "adam_stats/v_t_max": 2.3092428818927146e-05, "adam_stats/v_t_mean": 1.5696143386770278e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7864583730697632, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.16882090270519257, "all_logprobs": -0.011460821144282818, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -8.375, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.015566036105155945, "answer_logprob_reward": 3.047133207321167, "answer_logprob_reward/correct": 3.0919618606567383, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.061520010232925415, "answer_logprob_reward/incorrect": 2.882032871246338, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 2.079441547393799, "answer_logprob_reward/incorrect/p25": 2.4849066734313965, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.18731394410133362, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 2.079441547393799, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.09498215466737747, "clip_ratio": 0.0, "completion_length": 524.15625, "completion_length/correct": 471.0860900878906, "completion_length/correct/max": 998.0, "completion_length/correct/median": 439.0, "completion_length/correct/min": 71.0, "completion_length/correct/p25": 333.0, "completion_length/correct/p75": 582.0, "completion_length/correct/var": 40626.06640625, "completion_length/incorrect": 719.6097412109375, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 698.0, "completion_length/incorrect/min": 321.0, "completion_length/incorrect/p25": 435.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 74266.4921875, "completion_length/max": 1024.0, "completion_length/median": 471.0, "completion_length/min": 71.0, "completion_length/p25": 359.75, "completion_length/p75": 684.0, "completion_length/var": 57885.53515625, "epoch": 0.7088607594936709, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 3000208.0, "grad_norm": 0.22927089035511017, "kl": 0.0133056640625, "learning_rate": 2.245515092739488e-06, "len_reward": 0.05197674781084061, "len_reward/correct": 0.12103062123060226, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.16318106651306152, "len_reward/correct/min": -0.5, "len_reward/correct/p25": -0.08399999886751175, "len_reward/correct/p75": 0.3286933898925781, "len_reward/correct/var": 0.06610661745071411, "len_reward/incorrect": -0.20234361290931702, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.18196457624435425, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.37749120593070984, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03964154049754143, "len_reward/max": 0.5, "len_reward/median": 0.04279015213251114, "len_reward/min": -0.5, "len_reward/p25": -0.16771990060806274, "len_reward/p75": 0.2686336934566498, "len_reward/var": 0.07787184417247772, "loss": -0.7859, "mean_logprobs": -0.01220703125, "mean_logprobs/var": 8.20159912109375e-05, "missing_response_penalty": -0.078125, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.3658536374568939, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.23780488967895508, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.07239855825901031, "non_smoothed_logprobs/min": -17.375, "num_completions/total": 5376, "policy_loss": -0.785926342010498, "policy_loss/max": 0.0010675566736608744, "policy_loss/median": -0.9994320273399353, "policy_loss/min": -0.9997646808624268, "policy_loss/p25": -0.99957275390625, "policy_loss/p75": -0.9990478754043579, "policy_loss/var": 0.1688149869441986, "pred_value_log_probs": -0.602538526058197, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.67578125, "pred_value_log_probs/min": -1.03125, "pred_value_log_probs/p25": -0.67578125, "pred_value_log_probs/p75": -0.67578125, "pred_value_log_probs/var": 0.04960930347442627, "reward": 0.7864583730697632, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.16882090270519257, "rewards/accuracy_reward": 0.7864583730697632, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.16882090270519257, "rewards/answer_logprob_reward": 3.047133207321167, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 2.079441547393799, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.09498215466737747, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.05197674781084061, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.04279015213251114, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.16771990060806274, "rewards/len_reward/p75": 0.2686336934566498, "rewards/len_reward/var": 0.07787184417247772, "rewards/missing_response_penalty": -0.078125, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.07239855825901031, "smoothed_logprobs/min": -1.0, "step": 28, "value_loss": 2.8828125, "values": -0.8180979490280151, "values/max": 2.0, "values/median": -0.81640625, "values/min": -1.9906539916992188, "values/p25": -0.81640625, "values/p75": -0.81640625, "values/var": 0.294994056224823 }, { "adam_stats/lr_effective_max": 1.4186392036208417e-05, "adam_stats/lr_effective_mean": 7.851148897675131e-11, "adam_stats/lr_effective_min": -1.3711563042306807e-05, "adam_stats/m_t_max": 0.002043617656454444, "adam_stats/m_t_mean": -2.1564961621978185e-11, "adam_stats/m_t_min": -0.0016183295520022511, "adam_stats/v_t_max": 2.3092428818927146e-05, "adam_stats/v_t_mean": 1.5696143386770278e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.7088607594936709, "eval_accuracy_reward": 0.6600000262260437, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.22515049576759338, "eval_advantages": 0.6600000262260437, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.22515049576759338, "eval_all_logprobs": -0.013635395094752312, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -7.8125, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.022605856880545616, "eval_answer_logprob_reward": 3.0406923294067383, "eval_answer_logprob_reward/correct": 3.1021616458892822, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.07666410505771637, "eval_answer_logprob_reward/incorrect": 2.9213695526123047, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.6094379425048828, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.23875315487384796, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.6094379425048828, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.13851943612098694, "eval_completion_length": 522.07666015625, "eval_completion_length/correct": 465.57574462890625, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 419.0, "eval_completion_length/correct/min": 102.0, "eval_completion_length/correct/p25": 292.0, "eval_completion_length/correct/p75": 600.0, "eval_completion_length/correct/var": 45138.41015625, "eval_completion_length/incorrect": 631.7549438476562, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 577.0, "eval_completion_length/incorrect/min": 164.0, "eval_completion_length/incorrect/p25": 380.25, "eval_completion_length/incorrect/p75": 956.25, "eval_completion_length/incorrect/var": 84311.40625, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 476.0, "eval_completion_length/min": 102.0, "eval_completion_length/p25": 311.75, "eval_completion_length/p75": 667.75, "eval_completion_length/var": 64437.4375, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 4721129.0, "eval_len_reward": 0.07318604737520218, "eval_len_reward/correct": 0.17031632363796234, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.19773392379283905, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.037893105298280716, "eval_len_reward/correct/p75": 0.321541965007782, "eval_len_reward/correct/var": 0.044649749994277954, "eval_len_reward/incorrect": -0.11536100506782532, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": 0.0, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.20145376026630402, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.02995114028453827, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.026206322014331818, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.021526621654629707, "eval_len_reward/p75": 0.25351065397262573, "eval_len_reward/var": 0.05791022256016731, "eval_mean_logprobs": -0.0135498046875, "eval_mean_logprobs/var": 0.00012159347534179688, "eval_missing_response_penalty": -0.0833333358168602, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.2450980544090271, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": 0.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.1868569254875183, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.07664437592029572, "eval_num_completions/total": 8400, "eval_pred_value_log_probs": -0.6240668892860413, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.671875, "eval_pred_value_log_probs/min": -1.0546875, "eval_pred_value_log_probs/p25": -0.671875, "eval_pred_value_log_probs/p75": -0.671875, "eval_pred_value_log_probs/var": 0.026030590757727623, "eval_reward": 0.6600000262260437, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.22515049576759338, "eval_rewards/accuracy_reward": 0.6600000262260437, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.22515049576759338, "eval_rewards/answer_logprob_reward": 3.0406923294067383, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.6094379425048828, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.13851943612098694, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.07318604737520218, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.026206322014331818, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.021526621654629707, "eval_rewards/len_reward/p75": 0.25351065397262573, "eval_rewards/len_reward/var": 0.05791022256016731, "eval_rewards/missing_response_penalty": -0.0833333358168602, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.07664437592029572, "eval_runtime": 169.8675, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8697391748428345, "eval_values/max": 1.9914169311523438, "eval_values/median": -0.83154296875, "eval_values/min": -1.9982261657714844, "eval_values/p25": -0.83154296875, "eval_values/p75": -0.83154296875, "eval_values/var": 0.1275765597820282, "step": 28 }, { "accuracy_reward": 0.7291666865348816, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.1985165923833847, "adam_stats/lr_effective_max": 1.1584174899326172e-05, "adam_stats/lr_effective_mean": 5.095042443503672e-11, "adam_stats/lr_effective_min": -1.149139461631421e-05, "adam_stats/m_t_max": 0.0020437235943973064, "adam_stats/m_t_mean": -5.718687676181666e-12, "adam_stats/m_t_min": -0.0016945337411016226, "adam_stats/v_t_max": 2.307351678609848e-05, "adam_stats/v_t_mean": 1.5742044169944625e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7291666865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.1985165923833847, "all_logprobs": -0.012813528068363667, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -12.625, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.02218828722834587, "answer_logprob_reward": 2.97220516204834, "answer_logprob_reward/correct": 3.049698829650879, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 1.7917594909667969, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.09595181047916412, "answer_logprob_reward/incorrect": 2.763568639755249, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.7917594909667969, "answer_logprob_reward/incorrect/p25": 2.079441547393799, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.3107951283454895, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.7917594909667969, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.1690686196088791, "clip_ratio": 0.0, "completion_length": 541.4739990234375, "completion_length/correct": 475.8642883300781, "completion_length/correct/max": 1004.0, "completion_length/correct/median": 416.0, "completion_length/correct/min": 102.0, "completion_length/correct/p25": 314.75, "completion_length/correct/p75": 605.75, "completion_length/correct/var": 52804.36328125, "completion_length/incorrect": 718.1154174804688, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 700.0, "completion_length/incorrect/min": 203.0, "completion_length/incorrect/p25": 459.75, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 80780.6171875, "completion_length/max": 1024.0, "completion_length/median": 459.0, "completion_length/min": 102.0, "completion_length/p25": 336.75, "completion_length/p75": 749.25, "completion_length/var": 71648.0625, "epoch": 0.7341772151898734, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 3104171.0, "grad_norm": 0.22333048284053802, "kl": 0.01397705078125, "learning_rate": 1.8825509907063328e-06, "len_reward": 0.05953149124979973, "len_reward/correct": 0.14699342846870422, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.19756481051445007, "len_reward/correct/min": -0.5, "len_reward/correct/p25": -0.02327059395611286, "len_reward/correct/p75": 0.323266863822937, "len_reward/correct/var": 0.05901362746953964, "len_reward/incorrect": -0.17594295740127563, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.11352656781673431, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.3106473684310913, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03516513109207153, "len_reward/max": 0.5, "len_reward/median": 0.017633674666285515, "len_reward/min": -0.5, "len_reward/p25": -0.11306626349687576, "len_reward/p75": 0.2794128358364105, "len_reward/var": 0.07303960621356964, "loss": -0.7286, "mean_logprobs": -0.0126953125, "mean_logprobs/var": 8.249282836914062e-05, "missing_response_penalty": -0.1041666716337204, "missing_response_penalty/correct": -0.0071428571827709675, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": -1.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0071428571827709675, "missing_response_penalty/incorrect": -0.36538463830947876, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.23642534017562866, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.09380455315113068, "non_smoothed_logprobs/min": -17.625, "num_completions/total": 5568, "policy_loss": -0.728607177734375, "policy_loss/max": 0.0011987847974523902, "policy_loss/median": -0.9993556141853333, "policy_loss/min": -0.9998286366462708, "policy_loss/p25": -0.9995555877685547, "policy_loss/p75": 0.00018846243619918823, "policy_loss/var": 0.19850312173366547, "pred_value_log_probs": -0.6303989291191101, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.671875, "pred_value_log_probs/min": -1.21875, "pred_value_log_probs/p25": -0.671875, "pred_value_log_probs/p75": -0.671875, "pred_value_log_probs/var": 0.037602685391902924, "reward": 0.7291666865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.1985165923833847, "rewards/accuracy_reward": 0.7291666865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.1985165923833847, "rewards/answer_logprob_reward": 2.97220516204834, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.7917594909667969, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.1690686196088791, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.05953149124979973, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.017633674666285515, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.11306626349687576, "rewards/len_reward/p75": 0.2794128358364105, "rewards/len_reward/var": 0.07303960621356964, "rewards/missing_response_penalty": -0.1041666716337204, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.09380455315113068, "smoothed_logprobs/min": -1.0, "step": 29, "value_loss": 3.015625, "values": -0.7812522649765015, "values/max": 1.9557266235351562, "values/median": -0.83154296875, "values/min": -2.0, "values/p25": -0.83154296875, "values/p75": -0.83154296875, "values/var": 0.23247919976711273 }, { "adam_stats/lr_effective_max": 1.1584174899326172e-05, "adam_stats/lr_effective_mean": 5.095042443503672e-11, "adam_stats/lr_effective_min": -1.149139461631421e-05, "adam_stats/m_t_max": 0.0020437235943973064, "adam_stats/m_t_mean": -5.718687676181666e-12, "adam_stats/m_t_min": -0.0016945337411016226, "adam_stats/v_t_max": 2.307351678609848e-05, "adam_stats/v_t_mean": 1.5742044169944625e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.7341772151898734, "eval_accuracy_reward": 0.643333375453949, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.23022295534610748, "eval_advantages": 0.643333375453949, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.23022295534610748, "eval_all_logprobs": -0.013509598560631275, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -6.75, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.020643368363380432, "eval_answer_logprob_reward": 3.034558057785034, "eval_answer_logprob_reward/correct": 3.104830265045166, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.07334812730550766, "eval_answer_logprob_reward/incorrect": 2.907804489135742, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.4271163940429688, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.2928704023361206, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.4271163940429688, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.15986381471157074, "eval_completion_length": 523.1600341796875, "eval_completion_length/correct": 457.2124328613281, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 419.0, "eval_completion_length/correct/min": 119.0, "eval_completion_length/correct/p25": 288.0, "eval_completion_length/correct/p75": 598.0, "eval_completion_length/correct/var": 44095.76171875, "eval_completion_length/incorrect": 642.1121215820312, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 616.0, "eval_completion_length/incorrect/min": 159.0, "eval_completion_length/incorrect/p25": 397.5, "eval_completion_length/incorrect/p75": 970.5, "eval_completion_length/incorrect/var": 81015.578125, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 469.0, "eval_completion_length/min": 119.0, "eval_completion_length/p25": 313.5, "eval_completion_length/p75": 689.75, "eval_completion_length/var": 64907.7578125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 4878077.0, "eval_len_reward": 0.047344304621219635, "eval_len_reward/correct": 0.15471094846725464, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.1744791716337204, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": -0.0014566642930731177, "eval_len_reward/correct/p75": 0.3337520956993103, "eval_len_reward/correct/var": 0.05422097072005272, "eval_len_reward/incorrect": -0.14631706476211548, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.059578947722911835, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.2804742455482483, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03172319009900093, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.08517948538064957, "eval_len_reward/p75": 0.24898335337638855, "eval_len_reward/var": 0.06692613661289215, "eval_mean_logprobs": -0.0133056640625, "eval_mean_logprobs/var": 0.0001239776611328125, "eval_missing_response_penalty": -0.07666666805744171, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.2149532586336136, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": 0.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.17034032940864563, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.07102563977241516, "eval_num_completions/total": 8700, "eval_pred_value_log_probs": -0.6311616897583008, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.6796875, "eval_pred_value_log_probs/min": -1.09375, "eval_pred_value_log_probs/p25": -0.6796875, "eval_pred_value_log_probs/p75": -0.6796875, "eval_pred_value_log_probs/var": 0.026840047910809517, "eval_reward": 0.643333375453949, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.23022295534610748, "eval_rewards/accuracy_reward": 0.643333375453949, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.23022295534610748, "eval_rewards/answer_logprob_reward": 3.034558057785034, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.4271163940429688, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.15986381471157074, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.047344304621219635, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.08517948538064957, "eval_rewards/len_reward/p75": 0.24898335337638855, "eval_rewards/len_reward/var": 0.06692613661289215, "eval_rewards/missing_response_penalty": -0.07666666805744171, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.07102563977241516, "eval_runtime": 183.1127, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8550611734390259, "eval_values/max": 1.9912834167480469, "eval_values/median": -0.81640625, "eval_values/min": -1.9976587295532227, "eval_values/p25": -0.81640625, "eval_values/p75": -0.81640625, "eval_values/var": 0.1290351301431656, "step": 29 }, { "accuracy_reward": 0.75, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.75, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.1884816735982895, "adam_stats/lr_effective_max": 9.796427548280917e-06, "adam_stats/lr_effective_mean": 3.373353679725355e-11, "adam_stats/lr_effective_min": -9.434922503714915e-06, "adam_stats/m_t_max": 0.0022849079687148333, "adam_stats/m_t_mean": 3.377608522731057e-12, "adam_stats/m_t_min": -0.0018363596173003316, "adam_stats/v_t_max": 2.30702953558648e-05, "adam_stats/v_t_mean": 1.576915681367197e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.75, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.75, "advantages/p75": 1.0, "advantages/var": 0.1884816735982895, "all_logprobs": -0.012992546893656254, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -7.75, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.020218783989548683, "answer_logprob_reward": 3.098630905151367, "answer_logprob_reward/correct": 3.1202919483184814, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.4849066734313965, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.03695792704820633, "answer_logprob_reward/incorrect": 3.0336482524871826, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 2.4849066734313965, "answer_logprob_reward/incorrect/p25": 3.178053855895996, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.08092737942934036, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 2.4849066734313965, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.048999086022377014, "clip_ratio": 0.0, "completion_length": 497.91668701171875, "completion_length/correct": 439.1944580078125, "completion_length/correct/max": 993.0, "completion_length/correct/median": 390.0, "completion_length/correct/min": 161.0, "completion_length/correct/p25": 296.75, "completion_length/correct/p75": 547.0, "completion_length/correct/var": 37253.359375, "completion_length/incorrect": 674.0833740234375, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 682.0, "completion_length/incorrect/min": 202.0, "completion_length/incorrect/p25": 395.0, "completion_length/incorrect/p75": 980.5, "completion_length/incorrect/var": 76959.8671875, "completion_length/max": 1024.0, "completion_length/median": 436.0, "completion_length/min": 161.0, "completion_length/p25": 309.75, "completion_length/p75": 659.0, "completion_length/var": 57228.09375, "epoch": 0.759493670886076, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 3199771.0, "grad_norm": 0.21546195447444916, "kl": 0.0135498046875, "learning_rate": 1.544686755065677e-06, "len_reward": 0.11244390904903412, "len_reward/correct": 0.20514892041683197, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.23633748292922974, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.10495449602603912, "len_reward/correct/p75": 0.3914213478565216, "len_reward/correct/var": 0.05474286526441574, "len_reward/incorrect": -0.1656711846590042, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.08173579722642899, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.32875633239746094, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.037785883992910385, "len_reward/max": 0.5, "len_reward/median": 0.16490891575813293, "len_reward/min": -0.5, "len_reward/p25": -0.031448401510715485, "len_reward/p75": 0.343137264251709, "len_reward/var": 0.07620125263929367, "loss": -0.7495, "mean_logprobs": -0.0128173828125, "mean_logprobs/var": 0.00020694732666015625, "missing_response_penalty": -0.0625, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.25, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -0.25, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.19148936867713928, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.05890052393078804, "non_smoothed_logprobs/min": -17.375, "num_completions/total": 5760, "policy_loss": -0.7494577765464783, "policy_loss/max": 0.0012080126907676458, "policy_loss/median": -0.9994379281997681, "policy_loss/min": -0.999779462814331, "policy_loss/p25": -0.9995537996292114, "policy_loss/p75": -0.7489566802978516, "policy_loss/var": 0.18848688900470734, "pred_value_log_probs": -0.6227776408195496, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.6796875, "pred_value_log_probs/min": -1.2265625, "pred_value_log_probs/p25": -0.6796875, "pred_value_log_probs/p75": -0.6796875, "pred_value_log_probs/var": 0.03195691853761673, "reward": 0.75, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.75, "reward/p75": 1.0, "reward/var": 0.1884816735982895, "rewards/accuracy_reward": 0.75, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.75, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.1884816735982895, "rewards/answer_logprob_reward": 3.098630905151367, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 2.4849066734313965, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.048999086022377014, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.11244390904903412, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.16490891575813293, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.031448401510715485, "rewards/len_reward/p75": 0.343137264251709, "rewards/len_reward/var": 0.07620125263929367, "rewards/missing_response_penalty": -0.0625, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.05890052393078804, "smoothed_logprobs/min": -1.0, "step": 30, "value_loss": 2.828125, "values": -0.8329395651817322, "values/max": 1.816192626953125, "values/median": -0.81640625, "values/min": -2.0, "values/p25": -0.81640625, "values/p75": -0.81640625, "values/var": 0.17554526031017303 }, { "adam_stats/lr_effective_max": 9.796427548280917e-06, "adam_stats/lr_effective_mean": 3.373353679725355e-11, "adam_stats/lr_effective_min": -9.434922503714915e-06, "adam_stats/m_t_max": 0.0022849079687148333, "adam_stats/m_t_mean": 3.377608522731057e-12, "adam_stats/m_t_min": -0.0018363596173003316, "adam_stats/v_t_max": 2.30702953558648e-05, "adam_stats/v_t_mean": 1.576915681367197e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.759493670886076, "eval_accuracy_reward": 0.6166666746139526, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.2371794730424881, "eval_advantages": 0.6166666746139526, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.2371794730424881, "eval_all_logprobs": -0.015801355242729187, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -7.25, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.025429002940654755, "eval_answer_logprob_reward": 3.0262176990509033, "eval_answer_logprob_reward/correct": 3.117713689804077, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.525728702545166, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.060211893171072006, "eval_answer_logprob_reward/incorrect": 2.879028081893921, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.6094379425048828, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.27812936902046204, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.6094379425048828, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.15660837292671204, "eval_completion_length": 535.6366577148438, "eval_completion_length/correct": 458.4432678222656, "eval_completion_length/correct/max": 1005.0, "eval_completion_length/correct/median": 409.0, "eval_completion_length/correct/min": 120.0, "eval_completion_length/correct/p25": 296.0, "eval_completion_length/correct/p75": 599.0, "eval_completion_length/correct/var": 43624.09375, "eval_completion_length/incorrect": 659.8173828125, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 609.0, "eval_completion_length/incorrect/min": 161.0, "eval_completion_length/incorrect/p25": 406.5, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 85657.09375, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 465.0, "eval_completion_length/min": 120.0, "eval_completion_length/p25": 318.0, "eval_completion_length/p75": 715.5, "eval_completion_length/var": 69122.15625, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 5038768.0, "eval_len_reward": 0.019363053143024445, "eval_len_reward/correct": 0.13039062917232513, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.15446966886520386, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": -0.04123222827911377, "eval_len_reward/correct/p75": 0.3129418194293976, "eval_len_reward/correct/var": 0.05886157602071762, "eval_len_reward/incorrect": -0.1592465192079544, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.035070907324552536, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.30080854892730713, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03488873317837715, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.1431734263896942, "eval_len_reward/p75": 0.23216068744659424, "eval_len_reward/var": 0.06942147761583328, "eval_mean_logprobs": -0.01495361328125, "eval_mean_logprobs/var": 0.000659942626953125, "eval_missing_response_penalty": -0.10000000149011612, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.260869562625885, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.19450801610946655, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.09030099958181381, "eval_num_completions/total": 9000, "eval_pred_value_log_probs": -0.6379637122154236, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.6875, "eval_pred_value_log_probs/min": -1.0, "eval_pred_value_log_probs/p25": -0.6875, "eval_pred_value_log_probs/p75": -0.6875, "eval_pred_value_log_probs/var": 0.026982415467500687, "eval_reward": 0.6166666746139526, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.2371794730424881, "eval_rewards/accuracy_reward": 0.6166666746139526, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.2371794730424881, "eval_rewards/answer_logprob_reward": 3.0262176990509033, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.6094379425048828, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.15660837292671204, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.019363053143024445, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.1431734263896942, "eval_rewards/len_reward/p75": 0.23216068744659424, "eval_rewards/len_reward/var": 0.06942147761583328, "eval_rewards/missing_response_penalty": -0.10000000149011612, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.09030099958181381, "eval_runtime": 165.6023, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8475949168205261, "eval_values/max": 1.9912223815917969, "eval_values/median": -0.8076171875, "eval_values/min": -1.9976463317871094, "eval_values/p25": -0.8076171875, "eval_values/p75": -0.8076171875, "eval_values/var": 0.12839378416538239, "step": 30 }, { "accuracy_reward": 0.734375, "accuracy_reward/correct": 0.9999999403953552, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.19608965516090393, "adam_stats/lr_effective_max": 7.4884374043904245e-06, "adam_stats/lr_effective_mean": 2.6002009573256046e-11, "adam_stats/lr_effective_min": -7.4642371146182995e-06, "adam_stats/m_t_max": 0.002186116762459278, "adam_stats/m_t_mean": 1.3758061530311227e-11, "adam_stats/m_t_min": -0.0015871108043938875, "adam_stats/v_t_max": 2.304890767845791e-05, "adam_stats/v_t_mean": 1.5836876081365414e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.734375, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.19608965516090393, "all_logprobs": -0.017007239162921906, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -10.8125, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.029502060264348984, "answer_logprob_reward": 2.951953649520874, "answer_logprob_reward/correct": 3.1006038188934326, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.05757172405719757, "answer_logprob_reward/incorrect": 2.5409796237945557, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.3862943649291992, "answer_logprob_reward/incorrect/p25": 1.7917594909667969, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.5237216353416443, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.3862943649291992, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.2407103180885315, "clip_ratio": 0.0, "completion_length": 557.90625, "completion_length/correct": 470.1631164550781, "completion_length/correct/max": 999.0, "completion_length/correct/median": 438.0, "completion_length/correct/min": 109.0, "completion_length/correct/p25": 288.0, "completion_length/correct/p75": 632.0, "completion_length/correct/var": 48596.6484375, "completion_length/incorrect": 800.490234375, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 921.0, "completion_length/incorrect/min": 227.0, "completion_length/incorrect/p25": 573.5, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 67807.0546875, "completion_length/max": 1024.0, "completion_length/median": 493.0, "completion_length/min": 109.0, "completion_length/p25": 321.5, "completion_length/p75": 798.5, "completion_length/var": 74767.6328125, "epoch": 0.7848101265822784, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 3306889.0, "grad_norm": 0.24691376090049744, "kl": 0.013580322265625, "learning_rate": 1.234642669981946e-06, "len_reward": 0.08584549278020859, "len_reward/correct": 0.18430396914482117, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.22412508726119995, "len_reward/correct/min": -0.4765070974826813, "len_reward/correct/p25": 0.04035968333482742, "len_reward/correct/p75": 0.359270304441452, "len_reward/correct/var": 0.050723738968372345, "len_reward/incorrect": -0.18636326491832733, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.13509193062782288, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.33006739616394043, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.035194478929042816, "len_reward/max": 0.5, "len_reward/median": 0.07579787075519562, "len_reward/min": -0.5, "len_reward/p25": -0.08782581239938736, "len_reward/p75": 0.3012104034423828, "len_reward/var": 0.07333450019359589, "loss": -0.7338, "mean_logprobs": -0.0152587890625, "mean_logprobs/var": 0.000904083251953125, "missing_response_penalty": -0.125, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.4705882668495178, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.2541176676750183, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.10994765162467957, "non_smoothed_logprobs/min": -17.5, "num_completions/total": 5952, "policy_loss": -0.7338322401046753, "policy_loss/max": 0.0009722772520035505, "policy_loss/median": -0.9993570446968079, "policy_loss/min": -0.9997707605361938, "policy_loss/p25": -0.9995685815811157, "policy_loss/p75": 0.0002435501664876938, "policy_loss/var": 0.19606992602348328, "pred_value_log_probs": -0.6232706904411316, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.6875, "pred_value_log_probs/min": -1.3046875, "pred_value_log_probs/p25": -0.6875, "pred_value_log_probs/p75": -0.6875, "pred_value_log_probs/var": 0.04709911346435547, "reward": 0.734375, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.19608965516090393, "rewards/accuracy_reward": 0.734375, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.19608965516090393, "rewards/answer_logprob_reward": 2.951953649520874, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.3862943649291992, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.2407103180885315, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.08584549278020859, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.07579787075519562, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.08782581239938736, "rewards/len_reward/p75": 0.3012104034423828, "rewards/len_reward/var": 0.07333450019359589, "rewards/missing_response_penalty": -0.125, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.10994765162467957, "smoothed_logprobs/min": -1.0, "step": 31, "value_loss": 2.8203125, "values": -0.8314536809921265, "values/max": 1.558349609375, "values/median": -0.8076171875, "values/min": -2.0, "values/p25": -0.8076171875, "values/p75": -0.8076171875, "values/var": 0.21946369111537933 }, { "adam_stats/lr_effective_max": 7.4884374043904245e-06, "adam_stats/lr_effective_mean": 2.6002009573256046e-11, "adam_stats/lr_effective_min": -7.4642371146182995e-06, "adam_stats/m_t_max": 0.002186116762459278, "adam_stats/m_t_mean": 1.3758061530311227e-11, "adam_stats/m_t_min": -0.0015871108043938875, "adam_stats/v_t_max": 2.304890767845791e-05, "adam_stats/v_t_mean": 1.5836876081365414e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.7848101265822784, "eval_accuracy_reward": 0.6466667056083679, "eval_accuracy_reward/correct": 0.9999999403953552, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.22925306856632233, "eval_advantages": 0.6466667056083679, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.22925306856632233, "eval_all_logprobs": -0.013432860374450684, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -9.0, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.02059178426861763, "eval_answer_logprob_reward": 3.0655412673950195, "eval_answer_logprob_reward/correct": 3.1116878986358643, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.525728702545166, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.06313309818506241, "eval_answer_logprob_reward/incorrect": 2.981083869934082, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.6094379425048828, "eval_answer_logprob_reward/incorrect/p25": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.22886857390403748, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.6094379425048828, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.12503387033939362, "eval_completion_length": 517.8833618164062, "eval_completion_length/correct": 459.0257568359375, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 432.0, "eval_completion_length/correct/min": 123.0, "eval_completion_length/correct/p25": 295.75, "eval_completion_length/correct/p75": 579.75, "eval_completion_length/correct/var": 42457.7578125, "eval_completion_length/incorrect": 625.603759765625, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 579.0, "eval_completion_length/incorrect/min": 161.0, "eval_completion_length/incorrect/p25": 396.25, "eval_completion_length/incorrect/p75": 965.75, "eval_completion_length/incorrect/var": 77964.5078125, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 474.0, "eval_completion_length/min": 123.0, "eval_completion_length/p25": 313.0, "eval_completion_length/p75": 649.25, "eval_completion_length/var": 61146.046875, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 5194133.0, "eval_len_reward": 0.059598326683044434, "eval_len_reward/correct": 0.15640662610530853, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.17131148278713226, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.025762856006622314, "eval_len_reward/correct/p75": 0.3303588032722473, "eval_len_reward/correct/var": 0.04956372454762459, "eval_len_reward/incorrect": -0.1175791323184967, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.016791781410574913, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.19059230387210846, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.027757765725255013, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.004508196841925383, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.06665706634521484, "eval_len_reward/p75": 0.245398610830307, "eval_len_reward/var": 0.05894996598362923, "eval_mean_logprobs": -0.0135498046875, "eval_mean_logprobs/var": 0.0001068115234375, "eval_missing_response_penalty": -0.05666666850447655, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.1603773534297943, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": 0.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.13593891263008118, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.053634338080883026, "eval_num_completions/total": 9300, "eval_pred_value_log_probs": -0.6279247403144836, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.67578125, "eval_pred_value_log_probs/min": -1.078125, "eval_pred_value_log_probs/p25": -0.67578125, "eval_pred_value_log_probs/p75": -0.67578125, "eval_pred_value_log_probs/var": 0.02614113688468933, "eval_reward": 0.6466667056083679, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.22925306856632233, "eval_rewards/accuracy_reward": 0.6466667056083679, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.22925306856632233, "eval_rewards/answer_logprob_reward": 3.0655412673950195, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.6094379425048828, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.12503387033939362, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.059598326683044434, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.004508196841925383, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.06665706634521484, "eval_rewards/len_reward/p75": 0.245398610830307, "eval_rewards/len_reward/var": 0.05894996598362923, "eval_rewards/missing_response_penalty": -0.05666666850447655, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.053634338080883026, "eval_runtime": 165.3245, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.859175443649292, "eval_values/max": 1.9910888671875, "eval_values/median": -0.82080078125, "eval_values/min": -1.9976825714111328, "eval_values/p25": -0.82080078125, "eval_values/p75": -0.82080078125, "eval_values/var": 0.12633655965328217, "step": 31 }, { "accuracy_reward": 0.8072916865348816, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.1563863456249237, "adam_stats/lr_effective_max": 5.642576979880687e-06, "adam_stats/lr_effective_mean": 1.6730929142116935e-11, "adam_stats/lr_effective_min": -5.89284081797814e-06, "adam_stats/m_t_max": 0.0029257573187351227, "adam_stats/m_t_mean": 3.1172169880078204e-12, "adam_stats/m_t_min": -0.0021233742590993643, "adam_stats/v_t_max": 2.3117683667805977e-05, "adam_stats/v_t_mean": 1.5888793102394883e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.8072916865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.1563863456249237, "all_logprobs": -0.012487849220633507, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -7.78125, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.01879574917256832, "answer_logprob_reward": 3.049858808517456, "answer_logprob_reward/correct": 3.0904715061187744, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.0562736913561821, "answer_logprob_reward/incorrect": 2.879723310470581, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 2.079441547393799, "answer_logprob_reward/incorrect/p25": 2.4849066734313965, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.18292562663555145, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 2.079441547393799, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.08679652959108353, "clip_ratio": 0.0, "completion_length": 497.4270935058594, "completion_length/correct": 443.2064514160156, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 398.0, "completion_length/correct/min": 93.0, "completion_length/correct/p25": 293.5, "completion_length/correct/p75": 545.0, "completion_length/correct/var": 41991.97265625, "completion_length/incorrect": 724.5675659179688, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 736.0, "completion_length/incorrect/min": 292.0, "completion_length/incorrect/p25": 461.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 72786.640625, "completion_length/max": 1024.0, "completion_length/median": 431.0, "completion_length/min": 93.0, "completion_length/p25": 313.0, "completion_length/p75": 626.25, "completion_length/var": 59956.52734375, "epoch": 0.810126582278481, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 3402395.0, "grad_norm": 0.23521116375923157, "kl": 0.013671875, "learning_rate": 9.549150281252633e-07, "len_reward": 0.09783382713794708, "len_reward/correct": 0.16561655700206757, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.20136186480522156, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.004637676291167736, "len_reward/correct/p75": 0.3416718542575836, "len_reward/correct/var": 0.055385835468769073, "len_reward/incorrect": -0.1861209124326706, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.12238359451293945, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.3121117949485779, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03858597204089165, "len_reward/max": 0.5, "len_reward/median": 0.10651750862598419, "len_reward/min": -0.5, "len_reward/p25": -0.06163071095943451, "len_reward/p75": 0.2982640862464905, "len_reward/var": 0.07127738744020462, "loss": -0.8067, "mean_logprobs": -0.0120849609375, "mean_logprobs/var": 0.00015544891357421875, "missing_response_penalty": -0.0677083358168602, "missing_response_penalty/correct": -0.006451612804085016, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": -1.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.006451612338423729, "missing_response_penalty/incorrect": -0.3243243396282196, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.22522523999214172, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.06345440447330475, "non_smoothed_logprobs/min": -17.3125, "num_completions/total": 6144, "policy_loss": -0.8067445755004883, "policy_loss/max": 0.0009545485372655094, "policy_loss/median": -0.9994170665740967, "policy_loss/min": -0.9998272657394409, "policy_loss/p25": -0.9995482563972473, "policy_loss/p75": -0.9991546869277954, "policy_loss/var": 0.15637345612049103, "pred_value_log_probs": -0.6289335489273071, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.67578125, "pred_value_log_probs/min": -1.2578125, "pred_value_log_probs/p25": -0.67578125, "pred_value_log_probs/p75": -0.67578125, "pred_value_log_probs/var": 0.030305588617920876, "reward": 0.8072916865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.1563863456249237, "rewards/accuracy_reward": 0.8072916865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.1563863456249237, "rewards/answer_logprob_reward": 3.049858808517456, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 2.079441547393799, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.08679652959108353, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.09783382713794708, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.10651750862598419, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.06163071095943451, "rewards/len_reward/p75": 0.2982640862464905, "rewards/len_reward/var": 0.07127738744020462, "rewards/missing_response_penalty": -0.0677083358168602, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.06345440447330475, "smoothed_logprobs/min": -1.0, "step": 32, "value_loss": 3.078125, "values": -0.7944979071617126, "values/max": 2.0, "values/median": -0.82080078125, "values/min": -2.0, "values/p25": -0.82080078125, "values/p75": -0.82080078125, "values/var": 0.2095223218202591 }, { "adam_stats/lr_effective_max": 5.642576979880687e-06, "adam_stats/lr_effective_mean": 1.6730929142116935e-11, "adam_stats/lr_effective_min": -5.89284081797814e-06, "adam_stats/m_t_max": 0.0029257573187351227, "adam_stats/m_t_mean": 3.1172169880078204e-12, "adam_stats/m_t_min": -0.0021233742590993643, "adam_stats/v_t_max": 2.3117683667805977e-05, "adam_stats/v_t_mean": 1.5888793102394883e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.810126582278481, "eval_accuracy_reward": 0.6299999952316284, "eval_accuracy_reward/correct": 0.9999999403953552, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.23387958109378815, "eval_advantages": 0.6299999952316284, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.23387958109378815, "eval_all_logprobs": -0.013631907291710377, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -6.71875, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.02088269777595997, "eval_answer_logprob_reward": 3.025315761566162, "eval_answer_logprob_reward/correct": 3.106083869934082, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.07291579246520996, "eval_answer_logprob_reward/incorrect": 2.887791633605957, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.4271163940429688, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.2875722646713257, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.4271163940429688, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.1627872735261917, "eval_completion_length": 530.8900146484375, "eval_completion_length/correct": 459.5661315917969, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 416.0, "eval_completion_length/correct/min": 102.0, "eval_completion_length/correct/p25": 282.0, "eval_completion_length/correct/p75": 607.0, "eval_completion_length/correct/var": 49799.45703125, "eval_completion_length/incorrect": 652.3333740234375, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 600.0, "eval_completion_length/incorrect/min": 160.0, "eval_completion_length/incorrect/p25": 399.5, "eval_completion_length/incorrect/p75": 997.5, "eval_completion_length/incorrect/var": 80186.265625, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 493.0, "eval_completion_length/min": 102.0, "eval_completion_length/p25": 310.75, "eval_completion_length/p75": 692.0, "eval_completion_length/var": 69502.78125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 5353400.0, "eval_len_reward": 0.030771147459745407, "eval_len_reward/correct": 0.1357271522283554, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.17058824002742767, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": -0.005235602147877216, "eval_len_reward/correct/p75": 0.3363715410232544, "eval_len_reward/correct/var": 0.06353338062763214, "eval_len_reward/incorrect": -0.14793772995471954, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.0357557013630867, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.2836073637008667, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.031824991106987, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.1500728577375412, "eval_len_reward/p75": 0.25340428948402405, "eval_len_reward/var": 0.07047490775585175, "eval_mean_logprobs": -0.01312255859375, "eval_mean_logprobs/var": 0.000232696533203125, "eval_missing_response_penalty": -0.09333333373069763, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.2522522509098053, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -0.5, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.19033581018447876, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.084905244410038, "eval_num_completions/total": 9600, "eval_pred_value_log_probs": -0.6305449604988098, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.6796875, "eval_pred_value_log_probs/min": -1.015625, "eval_pred_value_log_probs/p25": -0.6796875, "eval_pred_value_log_probs/p75": -0.6796875, "eval_pred_value_log_probs/var": 0.026422210037708282, "eval_reward": 0.6299999952316284, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.23387958109378815, "eval_rewards/accuracy_reward": 0.6299999952316284, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.23387958109378815, "eval_rewards/answer_logprob_reward": 3.025315761566162, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.4271163940429688, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.1627872735261917, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.030771147459745407, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.1500728577375412, "eval_rewards/len_reward/p75": 0.25340428948402405, "eval_rewards/len_reward/var": 0.07047490775585175, "eval_rewards/missing_response_penalty": -0.09333333373069763, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.084905244410038, "eval_runtime": 179.956, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8578459024429321, "eval_values/max": 1.9912567138671875, "eval_values/median": -0.8193359375, "eval_values/min": -1.997598648071289, "eval_values/p25": -0.8193359375, "eval_values/p75": -0.8193359375, "eval_values/var": 0.12734489142894745, "step": 32 }, { "accuracy_reward": 0.71875, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.20320680737495422, "adam_stats/lr_effective_max": 4.1723183130670805e-06, "adam_stats/lr_effective_mean": 1.9211961882481532e-11, "adam_stats/lr_effective_min": -4.256146894476842e-06, "adam_stats/m_t_max": 0.003475466975942254, "adam_stats/m_t_mean": 3.5379600724777527e-12, "adam_stats/m_t_min": -0.002588527277112007, "adam_stats/v_t_max": 2.316551035619341e-05, "adam_stats/v_t_mean": 1.5930556570079024e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.71875, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.20320680737495422, "all_logprobs": -0.012182705104351044, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -6.0, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.016968917101621628, "answer_logprob_reward": 2.9920966625213623, "answer_logprob_reward/correct": 3.078829050064087, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.06911642849445343, "answer_logprob_reward/incorrect": 2.770446300506592, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.7917594909667969, "answer_logprob_reward/incorrect/p25": 2.079441547393799, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.3613618314266205, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.7917594909667969, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.16917377710342407, "clip_ratio": 0.0, "completion_length": 554.1458740234375, "completion_length/correct": 465.58697509765625, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 394.0, "completion_length/correct/min": 137.0, "completion_length/correct/p25": 295.0, "completion_length/correct/p75": 613.25, "completion_length/correct/var": 47973.8515625, "completion_length/incorrect": 780.4629516601562, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 768.0, "completion_length/incorrect/min": 247.0, "completion_length/incorrect/p25": 635.25, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 51827.8046875, "completion_length/max": 1024.0, "completion_length/median": 495.0, "completion_length/min": 137.0, "completion_length/p25": 341.75, "completion_length/p75": 754.75, "completion_length/var": 68939.421875, "epoch": 0.8354430379746836, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 3508791.0, "grad_norm": 0.22134709358215332, "kl": 0.0133056640625, "learning_rate": 7.077560319906696e-07, "len_reward": 0.08945988863706589, "len_reward/correct": 0.19692103564739227, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.24103586375713348, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.011464161798357964, "len_reward/correct/p75": 0.3868865966796875, "len_reward/correct/var": 0.053773920983076096, "len_reward/incorrect": -0.18516303598880768, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.17213115096092224, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.29456648230552673, "len_reward/incorrect/p75": -0.004050925839692354, "len_reward/incorrect/var": 0.0294047724455595, "len_reward/max": 0.5, "len_reward/median": 0.06662239879369736, "len_reward/min": -0.5, "len_reward/p25": -0.1273576319217682, "len_reward/p75": 0.3320743441581726, "len_reward/var": 0.07639607042074203, "loss": -0.7182, "mean_logprobs": -0.01214599609375, "mean_logprobs/var": 8.153915405273438e-05, "missing_response_penalty": -0.0989583358168602, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.35185185074806213, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.232354998588562, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.08963241428136826, "non_smoothed_logprobs/min": -17.4375, "num_completions/total": 6336, "policy_loss": -0.7182163000106812, "policy_loss/max": 0.001128749456256628, "policy_loss/median": -0.9993742108345032, "policy_loss/min": -0.999798059463501, "policy_loss/p25": -0.9995559453964233, "policy_loss/p75": 0.0002911695628426969, "policy_loss/var": 0.20319871604442596, "pred_value_log_probs": -0.6311730742454529, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.6796875, "pred_value_log_probs/min": -0.7265625, "pred_value_log_probs/p25": -0.6796875, "pred_value_log_probs/p75": -0.6796875, "pred_value_log_probs/var": 0.0247678030282259, "reward": 0.71875, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.20320680737495422, "rewards/accuracy_reward": 0.71875, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.20320680737495422, "rewards/answer_logprob_reward": 2.9920966625213623, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.7917594909667969, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.16917377710342407, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.08945988863706589, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.06662239879369736, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.1273576319217682, "rewards/len_reward/p75": 0.3320743441581726, "rewards/len_reward/var": 0.07639607042074203, "rewards/missing_response_penalty": -0.0989583358168602, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.08963241428136826, "smoothed_logprobs/min": -1.0, "step": 33, "value_loss": 2.8046875, "values": -0.8610403537750244, "values/max": 1.9999579191207886, "values/median": -0.8193359375, "values/min": -2.0, "values/p25": -0.8193359375, "values/p75": -0.8193359375, "values/var": 0.1409376859664917 }, { "adam_stats/lr_effective_max": 4.1723183130670805e-06, "adam_stats/lr_effective_mean": 1.9211961882481532e-11, "adam_stats/lr_effective_min": -4.256146894476842e-06, "adam_stats/m_t_max": 0.003475466975942254, "adam_stats/m_t_mean": 3.5379600724777527e-12, "adam_stats/m_t_min": -0.002588527277112007, "adam_stats/v_t_max": 2.316551035619341e-05, "adam_stats/v_t_mean": 1.5930556570079024e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.8354430379746836, "eval_accuracy_reward": 0.6133333444595337, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.2379487156867981, "eval_advantages": 0.6133333444595337, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.2379487156867981, "eval_all_logprobs": -0.014189048670232296, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -7.25, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.023575305938720703, "eval_answer_logprob_reward": 3.053041696548462, "eval_answer_logprob_reward/correct": 3.1240580081939697, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.06191055476665497, "eval_answer_logprob_reward/incorrect": 2.940394878387451, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.8325815200805664, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.20025868713855743, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.8325815200805664, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.12294083833694458, "eval_completion_length": 526.8699951171875, "eval_completion_length/correct": 460.2880554199219, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 415.0, "eval_completion_length/correct/min": 123.0, "eval_completion_length/correct/p25": 288.0, "eval_completion_length/correct/p75": 585.5, "eval_completion_length/correct/var": 46318.1328125, "eval_completion_length/incorrect": 632.4827270507812, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 580.0, "eval_completion_length/incorrect/min": 161.0, "eval_completion_length/incorrect/p25": 385.25, "eval_completion_length/incorrect/p75": 1000.75, "eval_completion_length/incorrect/var": 84424.96875, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 473.0, "eval_completion_length/min": 123.0, "eval_completion_length/p25": 311.0, "eval_completion_length/p75": 680.25, "eval_completion_length/var": 67875.1171875, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 5511461.0, "eval_len_reward": 0.048468347638845444, "eval_len_reward/correct": 0.16129276156425476, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.18827585875988007, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.043197937309741974, "eval_len_reward/correct/p75": 0.3443061411380768, "eval_len_reward/correct/var": 0.053549688309431076, "eval_len_reward/incorrect": -0.13049453496932983, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.011725585907697678, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.260113000869751, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03062245063483715, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.08182896673679352, "eval_len_reward/p75": 0.23203158378601074, "eval_len_reward/var": 0.0648113414645195, "eval_mean_logprobs": -0.0137939453125, "eval_mean_logprobs/var": 0.00011444091796875, "eval_missing_response_penalty": -0.08666666597127914, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.22413793206214905, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": 0.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.17541228234767914, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.0794202908873558, "eval_num_completions/total": 9900, "eval_pred_value_log_probs": -0.6236971020698547, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.671875, "eval_pred_value_log_probs/min": -0.95703125, "eval_pred_value_log_probs/p25": -0.671875, "eval_pred_value_log_probs/p75": -0.671875, "eval_pred_value_log_probs/var": 0.02542439103126526, "eval_reward": 0.6133333444595337, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.2379487156867981, "eval_rewards/accuracy_reward": 0.6133333444595337, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.2379487156867981, "eval_rewards/answer_logprob_reward": 3.053041696548462, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.8325815200805664, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.12294083833694458, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.048468347638845444, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.08182896673679352, "eval_rewards/len_reward/p75": 0.23203158378601074, "eval_rewards/len_reward/var": 0.0648113414645195, "eval_rewards/missing_response_penalty": -0.08666666597127914, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.0794202908873558, "eval_runtime": 167.6439, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8649055361747742, "eval_values/max": 1.9913864135742188, "eval_values/median": -0.82763671875, "eval_values/min": -1.9897403717041016, "eval_values/p25": -0.82763671875, "eval_values/p75": -0.82763671875, "eval_values/var": 0.12723535299301147, "step": 33 }, { "accuracy_reward": 0.7552083730697632, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.18583661317825317, "adam_stats/lr_effective_max": 2.912940090027405e-06, "adam_stats/lr_effective_mean": 1.2614843197811254e-11, "adam_stats/lr_effective_min": -2.9891004942328436e-06, "adam_stats/m_t_max": 0.0034666655119508505, "adam_stats/m_t_mean": 8.689088511204535e-12, "adam_stats/m_t_min": -0.0025768668856471777, "adam_stats/v_t_max": 2.315381971129682e-05, "adam_stats/v_t_mean": 1.5974886344305439e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7552083730697632, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.18583661317825317, "all_logprobs": -0.013020492158830166, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -7.15625, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.021044790744781494, "answer_logprob_reward": 3.0814647674560547, "answer_logprob_reward/correct": 3.120690107345581, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.4849066734313965, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.036724288016557693, "answer_logprob_reward/incorrect": 2.960451364517212, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 2.079441547393799, "answer_logprob_reward/incorrect/p25": 2.4849066734313965, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.13478022813796997, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 2.079441547393799, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.06491918861865997, "clip_ratio": 0.0, "completion_length": 488.953125, "completion_length/correct": 431.220703125, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 385.0, "completion_length/correct/min": 154.0, "completion_length/correct/p25": 284.0, "completion_length/correct/p75": 543.0, "completion_length/correct/var": 36086.546875, "completion_length/incorrect": 667.0637817382812, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 603.0, "completion_length/incorrect/min": 223.0, "completion_length/incorrect/p25": 470.5, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 70686.1015625, "completion_length/max": 1024.0, "completion_length/median": 434.0, "completion_length/min": 154.0, "completion_length/p25": 305.75, "completion_length/p75": 610.0, "completion_length/var": 54567.08984375, "epoch": 0.8607594936708861, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 3602670.0, "grad_norm": 0.22213982045650482, "kl": 0.013824462890625, "learning_rate": 4.951556604879049e-07, "len_reward": 0.1257983297109604, "len_reward/correct": 0.2184109091758728, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.26282352209091187, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.08117646723985672, "len_reward/correct/p75": 0.38985350728034973, "len_reward/correct/var": 0.04815307632088661, "len_reward/incorrect": -0.1599213033914566, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.004919730592519045, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.37439972162246704, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.04051285609602928, "len_reward/max": 0.5, "len_reward/median": 0.15333938598632812, "len_reward/min": -0.5, "len_reward/p25": -0.002818167908117175, "len_reward/p75": 0.349761962890625, "len_reward/var": 0.07266069203615189, "loss": -0.7547, "mean_logprobs": -0.012451171875, "mean_logprobs/var": 0.0001010894775390625, "missing_response_penalty": -0.0625, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.25531914830207825, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -0.5, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.19426457583904266, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.05890052020549774, "non_smoothed_logprobs/min": -17.25, "num_completions/total": 6528, "policy_loss": -0.7546552419662476, "policy_loss/max": 0.0010960586369037628, "policy_loss/median": -0.9994020462036133, "policy_loss/min": -0.9998460412025452, "policy_loss/p25": -0.9995385408401489, "policy_loss/p75": -0.9988569617271423, "policy_loss/var": 0.18583497405052185, "pred_value_log_probs": -0.6407127380371094, "pred_value_log_probs/max": -0.03466796875, "pred_value_log_probs/median": -0.671875, "pred_value_log_probs/min": -1.3046875, "pred_value_log_probs/p25": -0.671875, "pred_value_log_probs/p75": -0.671875, "pred_value_log_probs/var": 0.025662986561655998, "reward": 0.7552083730697632, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.18583661317825317, "rewards/accuracy_reward": 0.7552083730697632, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.18583661317825317, "rewards/answer_logprob_reward": 3.0814647674560547, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 2.079441547393799, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.06491918861865997, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.1257983297109604, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.15333938598632812, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.002818167908117175, "rewards/len_reward/p75": 0.349761962890625, "rewards/len_reward/var": 0.07266069203615189, "rewards/missing_response_penalty": -0.0625, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.05890052020549774, "smoothed_logprobs/min": -1.0, "step": 34, "value_loss": 2.6796875, "values": -0.8266692161560059, "values/max": 1.817474365234375, "values/median": -0.82763671875, "values/min": -1.901123046875, "values/p25": -0.82763671875, "values/p75": -0.82763671875, "values/var": 0.16586531698703766 }, { "adam_stats/lr_effective_max": 2.912940090027405e-06, "adam_stats/lr_effective_mean": 1.2614843197811254e-11, "adam_stats/lr_effective_min": -2.9891004942328436e-06, "adam_stats/m_t_max": 0.0034666655119508505, "adam_stats/m_t_mean": 8.689088511204535e-12, "adam_stats/m_t_min": -0.0025768668856471777, "adam_stats/v_t_max": 2.315381971129682e-05, "adam_stats/v_t_mean": 1.5974886344305439e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.8607594936708861, "eval_accuracy_reward": 0.6066666841506958, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.23942029476165771, "eval_advantages": 0.6066666841506958, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.23942029476165771, "eval_all_logprobs": -0.013087671250104904, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -7.25, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.019924871623516083, "eval_answer_logprob_reward": 3.0365922451019287, "eval_answer_logprob_reward/correct": 3.110943555831909, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.0733567550778389, "eval_answer_logprob_reward/incorrect": 2.921915292739868, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.6094379425048828, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.24640268087387085, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.6094379425048828, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.1493799239397049, "eval_completion_length": 528.7733764648438, "eval_completion_length/correct": 449.76373291015625, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 423.0, "eval_completion_length/correct/min": 102.0, "eval_completion_length/correct/p25": 292.75, "eval_completion_length/correct/p75": 575.5, "eval_completion_length/correct/var": 39934.58984375, "eval_completion_length/incorrect": 650.6356201171875, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 605.0, "eval_completion_length/incorrect/min": 161.0, "eval_completion_length/incorrect/p25": 383.0, "eval_completion_length/incorrect/p75": 1011.75, "eval_completion_length/incorrect/var": 88653.1640625, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 458.0, "eval_completion_length/min": 102.0, "eval_completion_length/p25": 317.75, "eval_completion_length/p75": 680.5, "eval_completion_length/var": 68525.3046875, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 5670093.0, "eval_len_reward": 0.033590104430913925, "eval_len_reward/correct": 0.14972682297229767, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.164000004529953, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.011975601315498352, "eval_len_reward/correct/p75": 0.31185182929039, "eval_len_reward/correct/var": 0.05152174457907677, "eval_len_reward/incorrect": -0.14553600549697876, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.03591160103678703, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.30224093794822693, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03321918472647667, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.09572094678878784, "eval_len_reward/p75": 0.22523236274719238, "eval_len_reward/var": 0.06506025791168213, "eval_mean_logprobs": -0.012451171875, "eval_mean_logprobs/var": 8.058547973632812e-05, "eval_missing_response_penalty": -0.09333333373069763, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.23728813230991364, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": 0.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.18252931535243988, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.084905244410038, "eval_num_completions/total": 10200, "eval_pred_value_log_probs": -0.6302194595336914, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.6796875, "eval_pred_value_log_probs/min": -0.97265625, "eval_pred_value_log_probs/p25": -0.6796875, "eval_pred_value_log_probs/p75": -0.6796875, "eval_pred_value_log_probs/var": 0.026019588112831116, "eval_reward": 0.6066666841506958, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.23942029476165771, "eval_rewards/accuracy_reward": 0.6066666841506958, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.23942029476165771, "eval_rewards/answer_logprob_reward": 3.0365922451019287, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.6094379425048828, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.1493799239397049, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.033590104430913925, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.09572094678878784, "eval_rewards/len_reward/p75": 0.22523236274719238, "eval_rewards/len_reward/var": 0.06506025791168213, "eval_rewards/missing_response_penalty": -0.09333333373069763, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.084905244410038, "eval_runtime": 172.9978, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8618219494819641, "eval_values/max": 1.9911537170410156, "eval_values/median": -0.8232421875, "eval_values/min": -1.9976177215576172, "eval_values/p25": -0.8232421875, "eval_values/p75": -0.8232421875, "eval_values/var": 0.12568846344947815, "step": 34 }, { "accuracy_reward": 0.7395833730697632, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.19360822439193726, "adam_stats/lr_effective_max": 1.8681564597500255e-06, "adam_stats/lr_effective_mean": 3.9044540170452e-12, "adam_stats/lr_effective_min": -1.9609640276030404e-06, "adam_stats/m_t_max": 0.003876834874972701, "adam_stats/m_t_mean": -3.680513793041795e-12, "adam_stats/m_t_min": -0.0027922026347368956, "adam_stats/v_t_max": 2.3187945771496743e-05, "adam_stats/v_t_mean": 1.6035186416532565e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7395833730697632, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.19360822439193726, "all_logprobs": -0.011429689824581146, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -5.875, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.016206832602620125, "answer_logprob_reward": 3.0354182720184326, "answer_logprob_reward/correct": 3.082453489303589, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.06069010868668556, "answer_logprob_reward/incorrect": 2.901837110519409, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 2.079441547393799, "answer_logprob_reward/incorrect/p25": 2.4849066734313965, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.16296735405921936, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 2.079441547393799, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.09292694181203842, "clip_ratio": 0.0, "completion_length": 535.640625, "completion_length/correct": 487.45068359375, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 457.0, "completion_length/correct/min": 95.0, "completion_length/correct/p25": 341.5, "completion_length/correct/p75": 614.5, "completion_length/correct/var": 44771.44140625, "completion_length/incorrect": 672.5, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 637.0, "completion_length/incorrect/min": 205.0, "completion_length/incorrect/p25": 426.25, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 80784.578125, "completion_length/max": 1024.0, "completion_length/median": 470.0, "completion_length/min": 95.0, "completion_length/p25": 363.0, "completion_length/p75": 668.5, "completion_length/var": 60405.78515625, "epoch": 0.8860759493670886, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 3705513.0, "grad_norm": 0.23321214318275452, "kl": 0.013275146484375, "learning_rate": 3.18825646801314e-07, "len_reward": 0.041230458766222, "len_reward/correct": 0.11422709375619888, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.10663507133722305, "len_reward/correct/min": -0.5, "len_reward/correct/p25": -0.045150503516197205, "len_reward/correct/p75": 0.291431725025177, "len_reward/correct/var": 0.05920525640249252, "len_reward/incorrect": -0.16608001291751862, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.14477798342704773, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.2727471590042114, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03390161693096161, "len_reward/max": 0.5, "len_reward/median": 0.0039627039805054665, "len_reward/min": -0.5, "len_reward/p25": -0.1009591817855835, "len_reward/p75": 0.21456876397132874, "len_reward/var": 0.06761597096920013, "loss": -0.7391, "mean_logprobs": -0.011962890625, "mean_logprobs/var": 8.392333984375e-05, "missing_response_penalty": -0.0677083358168602, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.25999999046325684, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -0.75, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.19632652401924133, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.06345440447330475, "non_smoothed_logprobs/min": -17.125, "num_completions/total": 6720, "policy_loss": -0.7390528917312622, "policy_loss/max": 0.0013184287818148732, "policy_loss/median": -0.9994184970855713, "policy_loss/min": -0.999813973903656, "policy_loss/p25": -0.9995690584182739, "policy_loss/p75": 0.0002589412033557892, "policy_loss/var": 0.19361191987991333, "pred_value_log_probs": -0.6269299983978271, "pred_value_log_probs/max": -1.0728836059570312e-06, "pred_value_log_probs/median": -0.6796875, "pred_value_log_probs/min": -1.203125, "pred_value_log_probs/p25": -0.6796875, "pred_value_log_probs/p75": -0.6796875, "pred_value_log_probs/var": 0.03591961786150932, "reward": 0.7395833730697632, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.19360822439193726, "rewards/accuracy_reward": 0.7395833730697632, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.19360822439193726, "rewards/answer_logprob_reward": 3.0354182720184326, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 2.079441547393799, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.09292694181203842, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.041230458766222, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.0039627039805054665, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.1009591817855835, "rewards/len_reward/p75": 0.21456876397132874, "rewards/len_reward/var": 0.06761597096920013, "rewards/missing_response_penalty": -0.0677083358168602, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.06345440447330475, "smoothed_logprobs/min": -1.0, "step": 35, "value_loss": 2.78125, "values": -0.8612565994262695, "values/max": 1.55291748046875, "values/median": -0.8232421875, "values/min": -1.9980392456054688, "values/p25": -0.8232421875, "values/p75": -0.8232421875, "values/var": 0.179811492562294 }, { "adam_stats/lr_effective_max": 1.8681564597500255e-06, "adam_stats/lr_effective_mean": 3.9044540170452e-12, "adam_stats/lr_effective_min": -1.9609640276030404e-06, "adam_stats/m_t_max": 0.003876834874972701, "adam_stats/m_t_mean": -3.680513793041795e-12, "adam_stats/m_t_min": -0.0027922026347368956, "adam_stats/v_t_max": 2.3187945771496743e-05, "adam_stats/v_t_mean": 1.6035186416532565e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.8860759493670886, "eval_accuracy_reward": 0.6366666555404663, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.2320958822965622, "eval_advantages": 0.6366666555404663, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.2320958822965622, "eval_all_logprobs": -0.01600293070077896, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -7.0, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.02648807317018509, "eval_answer_logprob_reward": 3.058274030685425, "eval_answer_logprob_reward/correct": 3.107264995574951, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.07228077203035355, "eval_answer_logprob_reward/incorrect": 2.9724276065826416, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.8325815200805664, "eval_answer_logprob_reward/incorrect/p25": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.230248361825943, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.8325815200805664, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.13331730663776398, "eval_completion_length": 524.0466918945312, "eval_completion_length/correct": 463.9371643066406, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 427.0, "eval_completion_length/correct/min": 113.0, "eval_completion_length/correct/p25": 296.0, "eval_completion_length/correct/p75": 610.5, "eval_completion_length/correct/var": 44254.5, "eval_completion_length/incorrect": 629.3760986328125, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 588.0, "eval_completion_length/incorrect/min": 160.0, "eval_completion_length/incorrect/p25": 360.0, "eval_completion_length/incorrect/p75": 975.0, "eval_completion_length/incorrect/var": 86309.28125, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 466.0, "eval_completion_length/min": 113.0, "eval_completion_length/p25": 311.75, "eval_completion_length/p75": 679.75, "eval_completion_length/var": 65649.328125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 5827307.0, "eval_len_reward": 0.04057655483484268, "eval_len_reward/correct": 0.14831168949604034, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.18240000307559967, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.0, "eval_len_reward/correct/p75": 0.310127854347229, "eval_len_reward/correct/var": 0.05233708769083023, "eval_len_reward/incorrect": -0.14820705354213715, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.010466397739946842, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.30244675278663635, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.035756561905145645, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.10545950382947922, "eval_len_reward/p75": 0.23584264516830444, "eval_len_reward/var": 0.06657975167036057, "eval_mean_logprobs": -0.0145263671875, "eval_mean_logprobs/var": 0.00049591064453125, "eval_missing_response_penalty": -0.07999999821186066, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.22018347680568695, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": 0.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.17329256236553192, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.07384615391492844, "eval_num_completions/total": 10500, "eval_pred_value_log_probs": -0.6315984129905701, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.6796875, "eval_pred_value_log_probs/min": -1.0703125, "eval_pred_value_log_probs/p25": -0.6796875, "eval_pred_value_log_probs/p75": -0.6796875, "eval_pred_value_log_probs/var": 0.025896800681948662, "eval_reward": 0.6366666555404663, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.2320958822965622, "eval_rewards/accuracy_reward": 0.6366666555404663, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.2320958822965622, "eval_rewards/answer_logprob_reward": 3.058274030685425, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.8325815200805664, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.13331730663776398, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.04057655483484268, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.10545950382947922, "eval_rewards/len_reward/p75": 0.23584264516830444, "eval_rewards/len_reward/var": 0.06657975167036057, "eval_rewards/missing_response_penalty": -0.07999999821186066, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.07384615391492844, "eval_runtime": 166.2341, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8527399301528931, "eval_values/max": 1.9915237426757812, "eval_values/median": -0.8154296875, "eval_values/min": -1.9976787567138672, "eval_values/p25": -0.8154296875, "eval_values/p75": -0.8154296875, "eval_values/var": 0.12768949568271637, "step": 35 }, { "accuracy_reward": 0.7864583730697632, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.16882088780403137, "adam_stats/lr_effective_max": 1.0349542662879685e-06, "adam_stats/lr_effective_mean": -5.445842344783958e-13, "adam_stats/lr_effective_min": -1.0831272447830997e-06, "adam_stats/m_t_max": 0.003202286083251238, "adam_stats/m_t_mean": -3.4583507932395285e-12, "adam_stats/m_t_min": -0.0023229410871863365, "adam_stats/v_t_max": 2.3172986402641982e-05, "adam_stats/v_t_mean": 1.6057590370224806e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7864583730697632, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.16882088780403137, "all_logprobs": -0.01194229256361723, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -6.96875, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.018675291910767555, "answer_logprob_reward": 2.993354558944702, "answer_logprob_reward/correct": 3.0919620990753174, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.06152000278234482, "answer_logprob_reward/incorrect": 2.630190849304199, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.2321436405181885, "answer_logprob_reward/incorrect/p25": 1.7917594909667969, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.6376937627792358, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.2321436405181885, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.21786069869995117, "clip_ratio": 0.0, "completion_length": 525.8958740234375, "completion_length/correct": 454.3907165527344, "completion_length/correct/max": 1024.0, "completion_length/correct/median": 421.0, "completion_length/correct/min": 140.0, "completion_length/correct/p25": 311.5, "completion_length/correct/p75": 542.5, "completion_length/correct/var": 38425.33203125, "completion_length/incorrect": 789.243896484375, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 932.0, "completion_length/incorrect/min": 251.0, "completion_length/incorrect/p25": 568.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 72143.4921875, "completion_length/max": 1024.0, "completion_length/median": 466.0, "completion_length/min": 140.0, "completion_length/p25": 331.75, "completion_length/p75": 669.75, "completion_length/var": 64214.8671875, "epoch": 0.9113924050632911, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 3806485.0, "grad_norm": 0.22095003724098206, "kl": 0.013275146484375, "learning_rate": 1.801856965207338e-07, "len_reward": 0.09725229442119598, "len_reward/correct": 0.17949548363685608, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.2398190051317215, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.04010206088423729, "len_reward/correct/p75": 0.3279731869697571, "len_reward/correct/var": 0.0479915477335453, "len_reward/incorrect": -0.205643430352211, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.1445556879043579, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.35064664483070374, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.037356629967689514, "len_reward/max": 0.5, "len_reward/median": 0.10348469018936157, "len_reward/min": -0.5, "len_reward/p25": -0.04575807601213455, "len_reward/p75": 0.31137803196907043, "len_reward/var": 0.07055462151765823, "loss": -0.7859, "mean_logprobs": -0.0115966796875, "mean_logprobs/var": 7.343292236328125e-05, "missing_response_penalty": -0.0885416716337204, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.4146341383457184, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.2487805187702179, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.08112456649541855, "non_smoothed_logprobs/min": -17.25, "num_completions/total": 6912, "policy_loss": -0.7859267592430115, "policy_loss/max": 0.0006974264979362488, "policy_loss/median": -0.9994094371795654, "policy_loss/min": -0.9998270869255066, "policy_loss/p25": -0.9995585680007935, "policy_loss/p75": -0.9990115165710449, "policy_loss/var": 0.16880235075950623, "pred_value_log_probs": -0.6317064166069031, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.6796875, "pred_value_log_probs/min": -1.0703125, "pred_value_log_probs/p25": -0.6796875, "pred_value_log_probs/p75": -0.6796875, "pred_value_log_probs/var": 0.031910497695207596, "reward": 0.7864583730697632, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.16882088780403137, "rewards/accuracy_reward": 0.7864583730697632, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.16882088780403137, "rewards/answer_logprob_reward": 2.993354558944702, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.2321436405181885, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.21786069869995117, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.09725229442119598, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.10348469018936157, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.04575807601213455, "rewards/len_reward/p75": 0.31137803196907043, "rewards/len_reward/var": 0.07055462151765823, "rewards/missing_response_penalty": -0.0885416716337204, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.08112456649541855, "smoothed_logprobs/min": -1.0, "step": 36, "value_loss": 3.0, "values": -0.8033027648925781, "values/max": 1.9813604354858398, "values/median": -0.8154296875, "values/min": -1.9923381805419922, "values/p25": -0.8154296875, "values/p75": -0.8154296875, "values/var": 0.20548765361309052 }, { "adam_stats/lr_effective_max": 1.0349542662879685e-06, "adam_stats/lr_effective_mean": -5.445842344783958e-13, "adam_stats/lr_effective_min": -1.0831272447830997e-06, "adam_stats/m_t_max": 0.003202286083251238, "adam_stats/m_t_mean": -3.4583507932395285e-12, "adam_stats/m_t_min": -0.0023229410871863365, "adam_stats/v_t_max": 2.3172986402641982e-05, "adam_stats/v_t_mean": 1.6057590370224806e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.9113924050632911, "eval_accuracy_reward": 0.6033333539962769, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.24012264609336853, "eval_advantages": 0.6033333539962769, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.24012264609336853, "eval_all_logprobs": -0.013793468475341797, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -6.9375, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.02206967957317829, "eval_answer_logprob_reward": 3.00046706199646, "eval_answer_logprob_reward/correct": 3.1109981536865234, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.06843782961368561, "eval_answer_logprob_reward/incorrect": 2.832347869873047, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.4271163940429688, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.35508111119270325, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.4271163940429688, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.1999768763780594, "eval_completion_length": 539.5366821289062, "eval_completion_length/correct": 459.8840026855469, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 405.0, "eval_completion_length/correct/min": 123.0, "eval_completion_length/correct/p25": 284.0, "eval_completion_length/correct/p75": 585.0, "eval_completion_length/correct/var": 46491.90234375, "eval_completion_length/incorrect": 660.6890869140625, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 639.0, "eval_completion_length/incorrect/min": 161.0, "eval_completion_length/incorrect/p25": 404.0, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 86875.921875, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 476.0, "eval_completion_length/min": 123.0, "eval_completion_length/p25": 311.75, "eval_completion_length/p75": 729.75, "eval_completion_length/var": 71956.3125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 5989168.0, "eval_len_reward": 0.06377188116312027, "eval_len_reward/correct": 0.18652506172657013, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.2197403609752655, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.040242165327072144, "eval_len_reward/correct/p75": 0.3611909747123718, "eval_len_reward/correct/var": 0.052537575364112854, "eval_len_reward/incorrect": -0.12293671816587448, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": 0.0, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.24470418691635132, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.029403019696474075, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.06674864888191223, "eval_len_reward/p75": 0.26565515995025635, "eval_len_reward/var": 0.06622755527496338, "eval_mean_logprobs": -0.01300048828125, "eval_mean_logprobs/var": 0.00014495849609375, "eval_missing_response_penalty": -0.1133333370089531, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.2857142984867096, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.20581114292144775, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.10082497447729111, "eval_num_completions/total": 10800, "eval_pred_value_log_probs": -0.6211370229721069, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.66796875, "eval_pred_value_log_probs/min": -1.015625, "eval_pred_value_log_probs/p25": -0.66796875, "eval_pred_value_log_probs/p75": -0.66796875, "eval_pred_value_log_probs/var": 0.025079844519495964, "eval_reward": 0.6033333539962769, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.24012264609336853, "eval_rewards/accuracy_reward": 0.6033333539962769, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.24012264609336853, "eval_rewards/answer_logprob_reward": 3.00046706199646, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.4271163940429688, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.1999768763780594, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.06377188116312027, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.06674864888191223, "eval_rewards/len_reward/p75": 0.26565515995025635, "eval_rewards/len_reward/var": 0.06622755527496338, "eval_rewards/missing_response_penalty": -0.1133333370089531, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.10082497447729111, "eval_runtime": 166.1014, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8637380003929138, "eval_values/max": 1.991241455078125, "eval_values/median": -0.82666015625, "eval_values/min": -1.9975910186767578, "eval_values/p25": -0.82666015625, "eval_values/p75": -0.82666015625, "eval_values/var": 0.12652793526649475, "step": 36 }, { "accuracy_reward": 0.7291666865348816, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.19851656258106232, "adam_stats/lr_effective_max": 4.674313629493554e-07, "adam_stats/lr_effective_mean": -8.888273147003578e-13, "adam_stats/lr_effective_min": -4.590939965964935e-07, "adam_stats/m_t_max": 0.0032116472721099854, "adam_stats/m_t_mean": -1.6246075665304893e-11, "adam_stats/m_t_min": -0.0027345679700374603, "adam_stats/v_t_max": 2.3160677301348187e-05, "adam_stats/v_t_mean": 1.6107731468095743e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7291666865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.19851656258106232, "all_logprobs": -0.014034952968358994, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -8.0, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.021980132907629013, "answer_logprob_reward": 2.9622597694396973, "answer_logprob_reward/correct": 3.0406382083892822, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.0865296870470047, "answer_logprob_reward/incorrect": 2.751241445541382, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.7917594909667969, "answer_logprob_reward/incorrect/p25": 2.079441547393799, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.2960805594921112, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.7917594909667969, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.15865588188171387, "clip_ratio": 0.0, "completion_length": 531.421875, "completion_length/correct": 444.3071594238281, "completion_length/correct/max": 1002.0, "completion_length/correct/median": 411.0, "completion_length/correct/min": 170.0, "completion_length/correct/p25": 293.75, "completion_length/correct/p75": 534.0, "completion_length/correct/var": 30305.453125, "completion_length/incorrect": 765.9615478515625, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 805.0, "completion_length/incorrect/min": 101.0, "completion_length/incorrect/p25": 561.75, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 81878.4296875, "completion_length/max": 1024.0, "completion_length/median": 463.0, "completion_length/min": 101.0, "completion_length/p25": 340.75, "completion_length/p75": 668.75, "completion_length/var": 64456.4140625, "epoch": 0.9367088607594937, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 3908518.0, "grad_norm": 0.22948013246059418, "kl": 0.0140380859375, "learning_rate": 8.035205700685167e-08, "len_reward": 0.10592197626829147, "len_reward/correct": 0.2098534256219864, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.24943362176418304, "len_reward/correct/min": -0.5, "len_reward/correct/p25": 0.08449641615152359, "len_reward/correct/p75": 0.35881540179252625, "len_reward/correct/var": 0.043855007737874985, "len_reward/incorrect": -0.1738934963941574, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.1606655716896057, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.2525708079338074, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.032235048711299896, "len_reward/max": 0.5, "len_reward/median": 0.14386044442653656, "len_reward/min": -0.5, "len_reward/p25": -0.042322441935539246, "len_reward/p75": 0.31434303522109985, "len_reward/var": 0.06975657492876053, "loss": -0.7286, "mean_logprobs": -0.013916015625, "mean_logprobs/var": 0.000415802001953125, "missing_response_penalty": -0.1041666716337204, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.38461539149284363, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.24132730066776276, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.09380454570055008, "non_smoothed_logprobs/min": -17.4375, "num_completions/total": 7104, "policy_loss": -0.7286052107810974, "policy_loss/max": 0.0022027799859642982, "policy_loss/median": -0.9993489980697632, "policy_loss/min": -0.9998674988746643, "policy_loss/p25": -0.9995580315589905, "policy_loss/p75": 0.0002452961343806237, "policy_loss/var": 0.19849976897239685, "pred_value_log_probs": -0.607271134853363, "pred_value_log_probs/max": -0.0019378662109375, "pred_value_log_probs/median": -0.66796875, "pred_value_log_probs/min": -1.2734375, "pred_value_log_probs/p25": -0.66796875, "pred_value_log_probs/p75": -0.66796875, "pred_value_log_probs/var": 0.04597088694572449, "reward": 0.7291666865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.19851656258106232, "rewards/accuracy_reward": 0.7291666865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.19851656258106232, "rewards/answer_logprob_reward": 2.9622597694396973, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.7917594909667969, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.15865588188171387, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.10592197626829147, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.14386044442653656, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.042322441935539246, "rewards/len_reward/p75": 0.31434303522109985, "rewards/len_reward/var": 0.06975657492876053, "rewards/missing_response_penalty": -0.1041666716337204, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.09380454570055008, "smoothed_logprobs/min": -1.0, "step": 37, "value_loss": 2.796875, "values": -0.8294498324394226, "values/max": 1.11810302734375, "values/median": -0.82666015625, "values/min": -1.9980120658874512, "values/p25": -0.82666015625, "values/p75": -0.82666015625, "values/var": 0.19865192472934723 }, { "adam_stats/lr_effective_max": 4.674313629493554e-07, "adam_stats/lr_effective_mean": -8.888273147003578e-13, "adam_stats/lr_effective_min": -4.590939965964935e-07, "adam_stats/m_t_max": 0.0032116472721099854, "adam_stats/m_t_mean": -1.6246075665304893e-11, "adam_stats/m_t_min": -0.0027345679700374603, "adam_stats/v_t_max": 2.3160677301348187e-05, "adam_stats/v_t_mean": 1.6107731468095743e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.9367088607594937, "eval_accuracy_reward": 0.6266666650772095, "eval_accuracy_reward/correct": 0.9999999403953552, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.23473802208900452, "eval_advantages": 0.6266666650772095, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.23473802208900452, "eval_all_logprobs": -0.013606792315840721, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -10.0, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.02223222889006138, "eval_answer_logprob_reward": 3.021596670150757, "eval_answer_logprob_reward/correct": 3.1150147914886475, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.06629540771245956, "eval_answer_logprob_reward/incorrect": 2.864788293838501, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.6094379425048828, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.2743638753890991, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.6094379425048828, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.1580141931772232, "eval_completion_length": 531.4367065429688, "eval_completion_length/correct": 460.1861572265625, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 438.0, "eval_completion_length/correct/min": 123.0, "eval_completion_length/correct/p25": 289.75, "eval_completion_length/correct/p75": 577.75, "eval_completion_length/correct/var": 42336.1328125, "eval_completion_length/incorrect": 651.0357666015625, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 603.0, "eval_completion_length/incorrect/min": 160.0, "eval_completion_length/incorrect/p25": 393.75, "eval_completion_length/incorrect/p75": 1024.0, "eval_completion_length/incorrect/var": 83656.65625, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 482.0, "eval_completion_length/min": 123.0, "eval_completion_length/p25": 312.75, "eval_completion_length/p75": 694.0, "eval_completion_length/var": 66084.2578125, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 6148599.0, "eval_len_reward": 0.03777023404836655, "eval_len_reward/correct": 0.14650936424732208, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.16897918283939362, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.005582460202276707, "eval_len_reward/correct/p75": 0.31155991554260254, "eval_len_reward/correct/var": 0.05335502699017525, "eval_len_reward/incorrect": -0.14475619792938232, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": -0.023785926401615143, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.2747696042060852, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.032570742070674896, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.08813288807868958, "eval_len_reward/p75": 0.226766437292099, "eval_len_reward/var": 0.06537482887506485, "eval_mean_logprobs": -0.01318359375, "eval_mean_logprobs/var": 0.00010204315185546875, "eval_missing_response_penalty": -0.09666667133569717, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.2589285969734192, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": -1.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.19361327588558197, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.0876142680644989, "eval_num_completions/total": 11100, "eval_pred_value_log_probs": -0.6239250302314758, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.671875, "eval_pred_value_log_probs/min": -1.0234375, "eval_pred_value_log_probs/p25": -0.671875, "eval_pred_value_log_probs/p75": -0.671875, "eval_pred_value_log_probs/var": 0.025610415264964104, "eval_reward": 0.6266666650772095, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.23473802208900452, "eval_rewards/accuracy_reward": 0.6266666650772095, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.23473802208900452, "eval_rewards/answer_logprob_reward": 3.021596670150757, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.6094379425048828, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.1580141931772232, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.03777023404836655, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.08813288807868958, "eval_rewards/len_reward/p75": 0.226766437292099, "eval_rewards/len_reward/var": 0.06537482887506485, "eval_rewards/missing_response_penalty": -0.09666667133569717, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.0876142680644989, "eval_runtime": 173.5451, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8668287396430969, "eval_values/max": 1.9913330078125, "eval_values/median": -0.82958984375, "eval_values/min": -1.998321533203125, "eval_values/p25": -0.82958984375, "eval_values/p75": -0.82958984375, "eval_values/var": 0.12681037187576294, "step": 37 }, { "accuracy_reward": 0.7916666865348816, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 1.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.16579407453536987, "adam_stats/lr_effective_max": 1.1624864271198021e-07, "adam_stats/lr_effective_mean": -4.2340298476474525e-13, "adam_stats/lr_effective_min": -1.1628848994860164e-07, "adam_stats/m_t_max": 0.003072062274441123, "adam_stats/m_t_mean": -1.9277281160245963e-11, "adam_stats/m_t_min": -0.0031752227805554867, "adam_stats/v_t_max": 2.3140813937061466e-05, "adam_stats/v_t_mean": 1.61747449201749e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.7916666865348816, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 1.0, "advantages/p75": 1.0, "advantages/var": 0.16579407453536987, "all_logprobs": -0.012521984055638313, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -8.25, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.018667710945010185, "answer_logprob_reward": 3.0625293254852295, "answer_logprob_reward/correct": 3.1005308628082275, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.4849066734313965, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.04804110899567604, "answer_logprob_reward/incorrect": 2.918123722076416, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.7917594909667969, "answer_logprob_reward/incorrect/p25": 2.4849066734313965, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.21404796838760376, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.7917594909667969, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.08720266073942184, "clip_ratio": 0.0, "completion_length": 510.26043701171875, "completion_length/correct": 448.27630615234375, "completion_length/correct/max": 1013.0, "completion_length/correct/median": 410.0, "completion_length/correct/min": 125.0, "completion_length/correct/p25": 301.5, "completion_length/correct/p75": 585.25, "completion_length/correct/var": 36263.9921875, "completion_length/incorrect": 745.7999877929688, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 725.0, "completion_length/incorrect/min": 271.0, "completion_length/incorrect/p25": 522.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 61569.08203125, "completion_length/max": 1024.0, "completion_length/median": 442.0, "completion_length/min": 125.0, "completion_length/p25": 326.75, "completion_length/p75": 661.5, "completion_length/var": 55917.28125, "epoch": 0.9620253164556962, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 4006488.0, "grad_norm": 0.24734780192375183, "kl": 0.013702392578125, "learning_rate": 2.012853002380466e-08, "len_reward": 0.10989409685134888, "len_reward/correct": 0.18586990237236023, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.2170129120349884, "len_reward/correct/min": -0.40889251232147217, "len_reward/correct/p25": 0.05338798090815544, "len_reward/correct/p75": 0.3427121341228485, "len_reward/correct/var": 0.04710303992033005, "len_reward/incorrect": -0.17881394922733307, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.12961409986019135, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.2924572229385376, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03831915184855461, "len_reward/max": 0.5, "len_reward/median": 0.1652684509754181, "len_reward/min": -0.5, "len_reward/p25": -0.05153507739305496, "len_reward/p75": 0.30100107192993164, "len_reward/var": 0.0671125203371048, "loss": -0.7911, "mean_logprobs": -0.01220703125, "mean_logprobs/var": 0.00010061264038085938, "missing_response_penalty": -0.0572916679084301, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.2750000059604645, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.2044871747493744, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.05429210141301155, "non_smoothed_logprobs/min": -17.0625, "num_completions/total": 7296, "policy_loss": -0.7911183834075928, "policy_loss/max": 0.000813791120890528, "policy_loss/median": -0.9994096755981445, "policy_loss/min": -0.9997642040252686, "policy_loss/p25": -0.9995551109313965, "policy_loss/p75": -0.9991562962532043, "policy_loss/var": 0.16578398644924164, "pred_value_log_probs": -0.6019009351730347, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.671875, "pred_value_log_probs/min": -1.1875, "pred_value_log_probs/p25": -0.671875, "pred_value_log_probs/p75": -0.671875, "pred_value_log_probs/var": 0.037637632340192795, "reward": 0.7916666865348816, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 1.0, "reward/p75": 1.0, "reward/var": 0.16579407453536987, "rewards/accuracy_reward": 0.7916666865348816, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 1.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.16579407453536987, "rewards/answer_logprob_reward": 3.0625293254852295, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.7917594909667969, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.08720266073942184, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.10989409685134888, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.1652684509754181, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.05153507739305496, "rewards/len_reward/p75": 0.30100107192993164, "rewards/len_reward/var": 0.0671125203371048, "rewards/missing_response_penalty": -0.0572916679084301, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.05429210141301155, "smoothed_logprobs/min": -1.0, "step": 38, "value_loss": 3.0, "values": -0.8338667750358582, "values/max": 1.6683349609375, "values/median": -0.82958984375, "values/min": -2.0, "values/p25": -0.82958984375, "values/p75": -0.82958984375, "values/var": 0.21893833577632904 }, { "adam_stats/lr_effective_max": 1.1624864271198021e-07, "adam_stats/lr_effective_mean": -4.2340298476474525e-13, "adam_stats/lr_effective_min": -1.1628848994860164e-07, "adam_stats/m_t_max": 0.003072062274441123, "adam_stats/m_t_mean": -1.9277281160245963e-11, "adam_stats/m_t_min": -0.0031752227805554867, "adam_stats/v_t_max": 2.3140813937061466e-05, "adam_stats/v_t_mean": 1.61747449201749e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.9620253164556962, "eval_accuracy_reward": 0.6033333539962769, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.24012264609336853, "eval_advantages": 0.6033333539962769, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.24012264609336853, "eval_all_logprobs": -0.012280795723199844, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -4.625, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.01731371320784092, "eval_answer_logprob_reward": 3.0406923294067383, "eval_answer_logprob_reward/correct": 3.1116490364074707, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.525728702545166, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.06317559629678726, "eval_answer_logprob_reward/incorrect": 2.932767152786255, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.4271163940429688, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.2800227105617523, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.4271163940429688, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.15622638165950775, "eval_completion_length": 512.9666748046875, "eval_completion_length/correct": 441.5027770996094, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 394.0, "eval_completion_length/correct/min": 123.0, "eval_completion_length/correct/p25": 289.0, "eval_completion_length/correct/p75": 560.0, "eval_completion_length/correct/var": 40204.70703125, "eval_completion_length/incorrect": 621.6638793945312, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 556.0, "eval_completion_length/incorrect/min": 155.0, "eval_completion_length/incorrect/p25": 384.0, "eval_completion_length/incorrect/p75": 923.0, "eval_completion_length/incorrect/var": 79311.71875, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 456.0, "eval_completion_length/min": 123.0, "eval_completion_length/p25": 313.0, "eval_completion_length/p75": 660.25, "eval_completion_length/var": 63297.6796875, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 6302489.0, "eval_len_reward": 0.0595286525785923, "eval_len_reward/correct": 0.17562632262706757, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.18323656916618347, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": 0.05504395067691803, "eval_len_reward/correct/p75": 0.32739973068237305, "eval_len_reward/correct/var": 0.04462508112192154, "eval_len_reward/incorrect": -0.11705687642097473, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": 0.0, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.20975428819656372, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.02766287699341774, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.03335507586598396, "eval_len_reward/p75": 0.2350156605243683, "eval_len_reward/var": 0.05835144594311714, "eval_mean_logprobs": -0.0123291015625, "eval_mean_logprobs/var": 7.43865966796875e-05, "eval_missing_response_penalty": -0.09000000357627869, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.22689077258110046, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": 0.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.17689786851406097, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.0821739062666893, "eval_num_completions/total": 11400, "eval_pred_value_log_probs": -0.6375287771224976, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.6875, "eval_pred_value_log_probs/min": -1.0234375, "eval_pred_value_log_probs/p25": -0.6875, "eval_pred_value_log_probs/p75": -0.6875, "eval_pred_value_log_probs/var": 0.02683435007929802, "eval_reward": 0.6033333539962769, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.24012264609336853, "eval_rewards/accuracy_reward": 0.6033333539962769, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.24012264609336853, "eval_rewards/answer_logprob_reward": 3.0406923294067383, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.4271163940429688, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.15622638165950775, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.0595286525785923, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.03335507586598396, "eval_rewards/len_reward/p75": 0.2350156605243683, "eval_rewards/len_reward/var": 0.05835144594311714, "eval_rewards/missing_response_penalty": -0.09000000357627869, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.0821739062666893, "eval_runtime": 177.0796, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8462945818901062, "eval_values/max": 1.9913215637207031, "eval_values/median": -0.8076171875, "eval_values/min": -1.9895877838134766, "eval_values/p25": -0.8076171875, "eval_values/p75": -0.8076171875, "eval_values/var": 0.1265174001455307, "step": 38 }, { "accuracy_reward": 0.71875, "accuracy_reward/correct": 1.0, "accuracy_reward/correct/max": 1.0, "accuracy_reward/correct/median": 1.0, "accuracy_reward/correct/min": 1.0, "accuracy_reward/correct/p25": 1.0, "accuracy_reward/correct/p75": 1.0, "accuracy_reward/correct/var": 0.0, "accuracy_reward/incorrect": 0.0, "accuracy_reward/incorrect/max": 0.0, "accuracy_reward/incorrect/median": 0.0, "accuracy_reward/incorrect/min": 0.0, "accuracy_reward/incorrect/p25": 0.0, "accuracy_reward/incorrect/p75": 0.0, "accuracy_reward/incorrect/var": 0.0, "accuracy_reward/max": 1.0, "accuracy_reward/median": 1.0, "accuracy_reward/min": 0.0, "accuracy_reward/p25": 0.0, "accuracy_reward/p75": 1.0, "accuracy_reward/var": 0.20320680737495422, "adam_stats/lr_effective_max": -0.0, "adam_stats/lr_effective_mean": 0.0, "adam_stats/lr_effective_min": -0.0, "adam_stats/m_t_max": 0.003430139273405075, "adam_stats/m_t_mean": -2.669468639193706e-11, "adam_stats/m_t_min": -0.0036267435643821955, "adam_stats/v_t_max": 2.3161932404036634e-05, "adam_stats/v_t_mean": 1.6231577713854417e-12, "adam_stats/v_t_min": 0.0, "advantages": 0.71875, "advantages/max": 1.0, "advantages/median": 1.0, "advantages/min": 0.0, "advantages/p25": 0.0, "advantages/p75": 1.0, "advantages/var": 0.20320680737495422, "all_logprobs": -0.011965964920818806, "all_logprobs/max": 0.0, "all_logprobs/median": 0.0, "all_logprobs/min": -5.28125, "all_logprobs/p25": 0.0, "all_logprobs/p75": 0.0, "all_logprobs/var": 0.01780596747994423, "answer_logprob_reward": 2.9355921745300293, "answer_logprob_reward/correct": 3.0717215538024902, "answer_logprob_reward/correct/max": 3.178053855895996, "answer_logprob_reward/correct/median": 3.178053855895996, "answer_logprob_reward/correct/min": 2.079441547393799, "answer_logprob_reward/correct/p25": 3.178053855895996, "answer_logprob_reward/correct/p75": 3.178053855895996, "answer_logprob_reward/correct/var": 0.0693558007478714, "answer_logprob_reward/incorrect": 2.5877063274383545, "answer_logprob_reward/incorrect/max": 3.178053855895996, "answer_logprob_reward/incorrect/median": 3.178053855895996, "answer_logprob_reward/incorrect/min": 1.2321436405181885, "answer_logprob_reward/incorrect/p25": 1.7917594909667969, "answer_logprob_reward/incorrect/p75": 3.178053855895996, "answer_logprob_reward/incorrect/var": 0.5624627470970154, "answer_logprob_reward/max": 3.178053855895996, "answer_logprob_reward/median": 3.178053855895996, "answer_logprob_reward/min": 1.2321436405181885, "answer_logprob_reward/p25": 3.178053855895996, "answer_logprob_reward/p75": 3.178053855895996, "answer_logprob_reward/var": 0.25342875719070435, "clip_ratio": 0.0, "completion_length": 532.90625, "completion_length/correct": 436.1449279785156, "completion_length/correct/max": 972.0, "completion_length/correct/median": 403.0, "completion_length/correct/min": 120.0, "completion_length/correct/p25": 312.5, "completion_length/correct/p75": 533.5, "completion_length/correct/var": 33634.80078125, "completion_length/incorrect": 780.1851806640625, "completion_length/incorrect/max": 1024.0, "completion_length/incorrect/median": 880.0, "completion_length/incorrect/min": 165.0, "completion_length/incorrect/p25": 554.0, "completion_length/incorrect/p75": 1024.0, "completion_length/incorrect/var": 75946.609375, "completion_length/max": 1024.0, "completion_length/median": 447.0, "completion_length/min": 120.0, "completion_length/p25": 337.5, "completion_length/p75": 683.5, "completion_length/var": 69251.984375, "epoch": 0.9873417721518988, "format_reward": 0.0, "format_reward/correct": 0.0, "format_reward/correct/max": 0.0, "format_reward/correct/median": 0.0, "format_reward/correct/min": 0.0, "format_reward/correct/p25": 0.0, "format_reward/correct/p75": 0.0, "format_reward/correct/var": 0.0, "format_reward/incorrect": 0.0, "format_reward/incorrect/max": 0.0, "format_reward/incorrect/median": 0.0, "format_reward/incorrect/min": 0.0, "format_reward/incorrect/p25": 0.0, "format_reward/incorrect/p75": 0.0, "format_reward/incorrect/var": 0.0, "format_reward/max": 0.0, "format_reward/median": 0.0, "format_reward/min": 0.0, "format_reward/p25": 0.0, "format_reward/p75": 0.0, "format_reward/var": 0.0, "generated_tokens/total": 4108806.0, "grad_norm": 0.2184079885482788, "kl": 0.013671875, "learning_rate": 0.0, "len_reward": 0.09715957939624786, "len_reward/correct": 0.2102111130952835, "len_reward/correct/max": 0.5, "len_reward/correct/median": 0.2156357318162918, "len_reward/correct/min": -0.4198760390281677, "len_reward/correct/p25": 0.10082200914621353, "len_reward/correct/p75": 0.3562450110912323, "len_reward/correct/var": 0.03685981035232544, "len_reward/incorrect": -0.1917499303817749, "len_reward/incorrect/max": 0.0, "len_reward/incorrect/median": -0.1574459969997406, "len_reward/incorrect/min": -0.5, "len_reward/incorrect/p25": -0.3096248507499695, "len_reward/incorrect/p75": 0.0, "len_reward/incorrect/var": 0.03601037338376045, "len_reward/max": 0.5, "len_reward/median": 0.11759082227945328, "len_reward/min": -0.5, "len_reward/p25": -0.03279658034443855, "len_reward/p75": 0.2950369417667389, "len_reward/var": 0.06926378607749939, "loss": -0.7182, "mean_logprobs": -0.0118408203125, "mean_logprobs/var": 6.771087646484375e-05, "missing_response_penalty": -0.125, "missing_response_penalty/correct": 0.0, "missing_response_penalty/correct/max": 0.0, "missing_response_penalty/correct/median": 0.0, "missing_response_penalty/correct/min": 0.0, "missing_response_penalty/correct/p25": 0.0, "missing_response_penalty/correct/p75": 0.0, "missing_response_penalty/correct/var": 0.0, "missing_response_penalty/incorrect": -0.4444444477558136, "missing_response_penalty/incorrect/max": 0.0, "missing_response_penalty/incorrect/median": 0.0, "missing_response_penalty/incorrect/min": -1.0, "missing_response_penalty/incorrect/p25": -1.0, "missing_response_penalty/incorrect/p75": 0.0, "missing_response_penalty/incorrect/var": 0.2515723407268524, "missing_response_penalty/max": 0.0, "missing_response_penalty/median": 0.0, "missing_response_penalty/min": -1.0, "missing_response_penalty/p25": 0.0, "missing_response_penalty/p75": 0.0, "missing_response_penalty/var": 0.10994765162467957, "non_smoothed_logprobs/min": -17.625, "num_completions/total": 7488, "policy_loss": -0.6661717891693115, "policy_loss/max": 0.000675500079523772, "policy_loss/median": -0.9994784593582153, "policy_loss/min": -0.9996159672737122, "policy_loss/p25": -0.9995781183242798, "policy_loss/p75": 0.0003850152716040611, "policy_loss/var": 0.23185908794403076, "pred_value_log_probs": -0.644974410533905, "pred_value_log_probs/max": 0.0, "pred_value_log_probs/median": -0.6875, "pred_value_log_probs/min": -1.21875, "pred_value_log_probs/p25": -0.6875, "pred_value_log_probs/p75": -0.6875, "pred_value_log_probs/var": 0.04008783400058746, "reward": 0.71875, "reward/max": 1.0, "reward/median": 1.0, "reward/min": 0.0, "reward/p25": 0.0, "reward/p75": 1.0, "reward/var": 0.20320680737495422, "rewards/accuracy_reward": 0.71875, "rewards/accuracy_reward/max": 1.0, "rewards/accuracy_reward/median": 1.0, "rewards/accuracy_reward/min": 0.0, "rewards/accuracy_reward/p25": 0.0, "rewards/accuracy_reward/p75": 1.0, "rewards/accuracy_reward/var": 0.20320680737495422, "rewards/answer_logprob_reward": 2.9355921745300293, "rewards/answer_logprob_reward/max": 3.178053855895996, "rewards/answer_logprob_reward/median": 3.178053855895996, "rewards/answer_logprob_reward/min": 1.2321436405181885, "rewards/answer_logprob_reward/p25": 3.178053855895996, "rewards/answer_logprob_reward/p75": 3.178053855895996, "rewards/answer_logprob_reward/var": 0.25342875719070435, "rewards/format_reward": 0.0, "rewards/format_reward/max": 0.0, "rewards/format_reward/median": 0.0, "rewards/format_reward/min": 0.0, "rewards/format_reward/p25": 0.0, "rewards/format_reward/p75": 0.0, "rewards/format_reward/var": 0.0, "rewards/len_reward": 0.09715957939624786, "rewards/len_reward/max": 0.5, "rewards/len_reward/median": 0.11759082227945328, "rewards/len_reward/min": -0.5, "rewards/len_reward/p25": -0.03279658034443855, "rewards/len_reward/p75": 0.2950369417667389, "rewards/len_reward/var": 0.06926378607749939, "rewards/missing_response_penalty": -0.125, "rewards/missing_response_penalty/max": 0.0, "rewards/missing_response_penalty/median": 0.0, "rewards/missing_response_penalty/min": -1.0, "rewards/missing_response_penalty/p25": 0.0, "rewards/missing_response_penalty/p75": 0.0, "rewards/missing_response_penalty/var": 0.10994765162467957, "smoothed_logprobs/min": -1.0, "step": 39, "value_loss": 2.859375, "values": -0.7773435115814209, "values/max": 1.5750732421875, "values/median": -0.8076171875, "values/min": -2.0, "values/p25": -0.8076171875, "values/p75": -0.8076171875, "values/var": 0.21612520515918732 }, { "adam_stats/lr_effective_max": -0.0, "adam_stats/lr_effective_mean": 0.0, "adam_stats/lr_effective_min": -0.0, "adam_stats/m_t_max": 0.003430139273405075, "adam_stats/m_t_mean": -2.669468639193706e-11, "adam_stats/m_t_min": -0.0036267435643821955, "adam_stats/v_t_max": 2.3161932404036634e-05, "adam_stats/v_t_mean": 1.6231577713854417e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.9873417721518988, "eval_accuracy_reward": 0.6399999856948853, "eval_accuracy_reward/correct": 1.0, "eval_accuracy_reward/correct/max": 1.0, "eval_accuracy_reward/correct/median": 1.0, "eval_accuracy_reward/correct/min": 1.0, "eval_accuracy_reward/correct/p25": 1.0, "eval_accuracy_reward/correct/p75": 1.0, "eval_accuracy_reward/correct/var": 0.0, "eval_accuracy_reward/incorrect": 0.0, "eval_accuracy_reward/incorrect/max": 0.0, "eval_accuracy_reward/incorrect/median": 0.0, "eval_accuracy_reward/incorrect/min": 0.0, "eval_accuracy_reward/incorrect/p25": 0.0, "eval_accuracy_reward/incorrect/p75": 0.0, "eval_accuracy_reward/incorrect/var": 0.0, "eval_accuracy_reward/max": 1.0, "eval_accuracy_reward/median": 1.0, "eval_accuracy_reward/min": 0.0, "eval_accuracy_reward/p25": 0.0, "eval_accuracy_reward/p75": 1.0, "eval_accuracy_reward/var": 0.23117057979106903, "eval_advantages": 0.6399999856948853, "eval_advantages/max": 1.0, "eval_advantages/median": 1.0, "eval_advantages/min": 0.0, "eval_advantages/p25": 0.0, "eval_advantages/p75": 1.0, "eval_advantages/var": 0.23117057979106903, "eval_all_logprobs": -0.012546074576675892, "eval_all_logprobs/max": 0.0, "eval_all_logprobs/median": 0.0, "eval_all_logprobs/min": -5.84375, "eval_all_logprobs/p25": 0.0, "eval_all_logprobs/p75": 0.0, "eval_all_logprobs/var": 0.018090752884745598, "eval_answer_logprob_reward": 3.054785966873169, "eval_answer_logprob_reward/correct": 3.12078857421875, "eval_answer_logprob_reward/correct/max": 3.2188758850097656, "eval_answer_logprob_reward/correct/median": 3.2188758850097656, "eval_answer_logprob_reward/correct/min": 2.1202635765075684, "eval_answer_logprob_reward/correct/p25": 3.2188758850097656, "eval_answer_logprob_reward/correct/p75": 3.2188758850097656, "eval_answer_logprob_reward/correct/var": 0.0633377954363823, "eval_answer_logprob_reward/incorrect": 2.9374477863311768, "eval_answer_logprob_reward/incorrect/max": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/median": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/min": 1.8325815200805664, "eval_answer_logprob_reward/incorrect/p25": 2.525728702545166, "eval_answer_logprob_reward/incorrect/p75": 3.2188758850097656, "eval_answer_logprob_reward/incorrect/var": 0.23042640089988708, "eval_answer_logprob_reward/max": 3.2188758850097656, "eval_answer_logprob_reward/median": 3.2188758850097656, "eval_answer_logprob_reward/min": 1.8325815200805664, "eval_answer_logprob_reward/p25": 3.2188758850097656, "eval_answer_logprob_reward/p75": 3.2188758850097656, "eval_answer_logprob_reward/var": 0.13069075345993042, "eval_completion_length": 522.9367065429688, "eval_completion_length/correct": 466.28125, "eval_completion_length/correct/max": 1024.0, "eval_completion_length/correct/median": 427.0, "eval_completion_length/correct/min": 123.0, "eval_completion_length/correct/p25": 282.5, "eval_completion_length/correct/p75": 624.0, "eval_completion_length/correct/var": 45490.38671875, "eval_completion_length/incorrect": 623.6574096679688, "eval_completion_length/incorrect/max": 1024.0, "eval_completion_length/incorrect/median": 563.0, "eval_completion_length/incorrect/min": 161.0, "eval_completion_length/incorrect/p25": 354.25, "eval_completion_length/incorrect/p75": 982.0, "eval_completion_length/incorrect/var": 91662.375, "eval_completion_length/max": 1024.0, "eval_completion_length/median": 462.0, "eval_completion_length/min": 123.0, "eval_completion_length/p25": 304.75, "eval_completion_length/p75": 716.75, "eval_completion_length/var": 67586.8046875, "eval_format_reward": 0.0, "eval_format_reward/correct": 0.0, "eval_format_reward/correct/max": 0.0, "eval_format_reward/correct/median": 0.0, "eval_format_reward/correct/min": 0.0, "eval_format_reward/correct/p25": 0.0, "eval_format_reward/correct/p75": 0.0, "eval_format_reward/correct/var": 0.0, "eval_format_reward/incorrect": 0.0, "eval_format_reward/incorrect/max": 0.0, "eval_format_reward/incorrect/median": 0.0, "eval_format_reward/incorrect/min": 0.0, "eval_format_reward/incorrect/p25": 0.0, "eval_format_reward/incorrect/p75": 0.0, "eval_format_reward/incorrect/var": 0.0, "eval_format_reward/max": 0.0, "eval_format_reward/median": 0.0, "eval_format_reward/min": 0.0, "eval_format_reward/p25": 0.0, "eval_format_reward/p75": 0.0, "eval_format_reward/var": 0.0, "eval_generated_tokens/total": 6459370.0, "eval_len_reward": 0.021767830476164818, "eval_len_reward/correct": 0.12273482978343964, "eval_len_reward/correct/max": 0.5, "eval_len_reward/correct/median": 0.13476347923278809, "eval_len_reward/correct/min": -0.5, "eval_len_reward/correct/p25": -0.04037480056285858, "eval_len_reward/correct/p75": 0.302950918674469, "eval_len_reward/correct/var": 0.061008352786302567, "eval_len_reward/incorrect": -0.15772904455661774, "eval_len_reward/incorrect/max": 0.0, "eval_len_reward/incorrect/median": 0.0, "eval_len_reward/incorrect/min": -0.5, "eval_len_reward/incorrect/p25": -0.3225032091140747, "eval_len_reward/incorrect/p75": 0.0, "eval_len_reward/incorrect/var": 0.03720378503203392, "eval_len_reward/max": 0.5, "eval_len_reward/median": 0.0, "eval_len_reward/min": -0.5, "eval_len_reward/p25": -0.13391777873039246, "eval_len_reward/p75": 0.21725769340991974, "eval_len_reward/var": 0.07046949863433838, "eval_mean_logprobs": -0.0128173828125, "eval_mean_logprobs/var": 0.00011205673217773438, "eval_missing_response_penalty": -0.0833333358168602, "eval_missing_response_penalty/correct": 0.0, "eval_missing_response_penalty/correct/max": 0.0, "eval_missing_response_penalty/correct/median": 0.0, "eval_missing_response_penalty/correct/min": 0.0, "eval_missing_response_penalty/correct/p25": 0.0, "eval_missing_response_penalty/correct/p75": 0.0, "eval_missing_response_penalty/correct/var": 0.0, "eval_missing_response_penalty/incorrect": -0.23148147761821747, "eval_missing_response_penalty/incorrect/max": 0.0, "eval_missing_response_penalty/incorrect/median": 0.0, "eval_missing_response_penalty/incorrect/min": -1.0, "eval_missing_response_penalty/incorrect/p25": 0.0, "eval_missing_response_penalty/incorrect/p75": 0.0, "eval_missing_response_penalty/incorrect/var": 0.17956037819385529, "eval_missing_response_penalty/max": 0.0, "eval_missing_response_penalty/median": 0.0, "eval_missing_response_penalty/min": -1.0, "eval_missing_response_penalty/p25": 0.0, "eval_missing_response_penalty/p75": 0.0, "eval_missing_response_penalty/var": 0.07664436846971512, "eval_num_completions/total": 11700, "eval_pred_value_log_probs": -0.635452389717102, "eval_pred_value_log_probs/max": 0.0, "eval_pred_value_log_probs/median": -0.68359375, "eval_pred_value_log_probs/min": -1.078125, "eval_pred_value_log_probs/p25": -0.68359375, "eval_pred_value_log_probs/p75": -0.68359375, "eval_pred_value_log_probs/var": 0.026864545419812202, "eval_reward": 0.6399999856948853, "eval_reward/max": 1.0, "eval_reward/median": 1.0, "eval_reward/min": 0.0, "eval_reward/p25": 0.0, "eval_reward/p75": 1.0, "eval_reward/var": 0.23117057979106903, "eval_rewards/accuracy_reward": 0.6399999856948853, "eval_rewards/accuracy_reward/max": 1.0, "eval_rewards/accuracy_reward/median": 1.0, "eval_rewards/accuracy_reward/min": 0.0, "eval_rewards/accuracy_reward/p25": 0.0, "eval_rewards/accuracy_reward/p75": 1.0, "eval_rewards/accuracy_reward/var": 0.23117057979106903, "eval_rewards/answer_logprob_reward": 3.054785966873169, "eval_rewards/answer_logprob_reward/max": 3.2188758850097656, "eval_rewards/answer_logprob_reward/median": 3.2188758850097656, "eval_rewards/answer_logprob_reward/min": 1.8325815200805664, "eval_rewards/answer_logprob_reward/p25": 3.2188758850097656, "eval_rewards/answer_logprob_reward/p75": 3.2188758850097656, "eval_rewards/answer_logprob_reward/var": 0.13069075345993042, "eval_rewards/format_reward": 0.0, "eval_rewards/format_reward/max": 0.0, "eval_rewards/format_reward/median": 0.0, "eval_rewards/format_reward/min": 0.0, "eval_rewards/format_reward/p25": 0.0, "eval_rewards/format_reward/p75": 0.0, "eval_rewards/format_reward/var": 0.0, "eval_rewards/len_reward": 0.021767830476164818, "eval_rewards/len_reward/max": 0.5, "eval_rewards/len_reward/median": 0.0, "eval_rewards/len_reward/min": -0.5, "eval_rewards/len_reward/p25": -0.13391777873039246, "eval_rewards/len_reward/p75": 0.21725769340991974, "eval_rewards/len_reward/var": 0.07046949863433838, "eval_rewards/missing_response_penalty": -0.0833333358168602, "eval_rewards/missing_response_penalty/max": 0.0, "eval_rewards/missing_response_penalty/median": 0.0, "eval_rewards/missing_response_penalty/min": -1.0, "eval_rewards/missing_response_penalty/p25": 0.0, "eval_rewards/missing_response_penalty/p75": 0.0, "eval_rewards/missing_response_penalty/var": 0.07664436846971512, "eval_runtime": 175.7226, "eval_samples_per_second": 0.0, "eval_steps_per_second": 0.0, "eval_values": -0.8476092219352722, "eval_values/max": 1.9911956787109375, "eval_values/median": -0.80859375, "eval_values/min": -1.9981098175048828, "eval_values/p25": -0.80859375, "eval_values/p75": -0.80859375, "eval_values/var": 0.12626203894615173, "step": 39 }, { "adam_stats/lr_effective_max": -0.0, "adam_stats/lr_effective_mean": 0.0, "adam_stats/lr_effective_min": -0.0, "adam_stats/m_t_max": 0.003430139273405075, "adam_stats/m_t_mean": -2.669468639193706e-11, "adam_stats/m_t_min": -0.0036267435643821955, "adam_stats/v_t_max": 2.3161932404036634e-05, "adam_stats/v_t_mean": 1.6231577713854417e-12, "adam_stats/v_t_min": 0.0, "epoch": 0.9873417721518988, "step": 39, "total_flos": 0.0, "train_loss": -0.7422515612382156, "train_runtime": 31151.0975, "train_samples_per_second": 0.241, "train_steps_per_second": 0.001 } ], "logging_steps": 1, "max_steps": 39, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 1, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 24, "trial_name": null, "trial_params": null }