{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9984, "eval_steps": 50, "global_step": 312, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.1029296875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1533.0, "completions/mean_length": 463.92587890625, "completions/mean_terminated_length": 340.91392822265624, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.016, "grad_norm": 0.003800533711910248, "learning_rate": 3.1249999999999997e-07, "loss": 0.0456, "num_tokens": 19594633.0, "reward": 0.21571238338947296, "reward_std": 0.3878848910331726, "rewards/accuracy_reward": 0.06201171875, "rewards/brier_reward": 0.10856110006570815, "rewards/confidence_one_or_zero": 0.413671875, "rewards/format_reward": 0.261328125, "rewards/mean_confidence_reward": 0.5398741960525513, "rewards/smce": -0.4780743718147278, "step": 5 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.10458984375, "completions/max_length": 1536.0, "completions/max_terminated_length": 1534.0, "completions/mean_length": 445.432421875, "completions/mean_terminated_length": 318.1658874511719, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.032, "grad_norm": 0.004327881615608931, "learning_rate": 6.249999999999999e-07, "loss": 0.052, "num_tokens": 39256213.0, "reward": 0.22842271625995636, "reward_std": 0.3874061942100525, "rewards/accuracy_reward": 0.0625, "rewards/brier_reward": 0.1117446094751358, "rewards/confidence_one_or_zero": 0.432421875, "rewards/format_reward": 0.28310546875, "rewards/mean_confidence_reward": 0.5687525153160096, "rewards/smce": -0.506620317697525, "step": 10 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.050390625, "completions/max_length": 1536.0, "completions/max_terminated_length": 1472.8, "completions/mean_length": 305.22607421875, "completions/mean_terminated_length": 240.37628784179688, "completions/min_length": 1.4, "completions/min_terminated_length": 1.4, "epoch": 0.048, "grad_norm": 0.004292675759643316, "learning_rate": 9.374999999999999e-07, "loss": 0.0542, "num_tokens": 57430464.0, "reward": 0.4086169540882111, "reward_std": 0.43530423045158384, "rewards/accuracy_reward": 0.1064453125, "rewards/brier_reward": 0.20014296174049379, "rewards/confidence_one_or_zero": 0.537109375, "rewards/format_reward": 0.51123046875, "rewards/mean_confidence_reward": 0.6928913235664368, "rewards/smce": -0.5873086810111999, "step": 15 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.01923828125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1438.8, "completions/mean_length": 205.31982421875, "completions/mean_terminated_length": 179.24837951660157, "completions/min_length": 10.8, "completions/min_terminated_length": 10.8, "epoch": 0.064, "grad_norm": 0.0027189909014850855, "learning_rate": 1e-06, "loss": 0.0359, "num_tokens": 74451339.0, "reward": 0.6751530885696411, "reward_std": 0.39800527691841125, "rewards/accuracy_reward": 0.18740234375, "rewards/brier_reward": 0.3311912059783936, "rewards/confidence_one_or_zero": 0.5701171875, "rewards/format_reward": 0.83232421875, "rewards/mean_confidence_reward": 0.7987796306610108, "rewards/smce": -0.614277696609497, "step": 20 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.01142578125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1298.4, "completions/mean_length": 157.06796875, "completions/mean_terminated_length": 141.12780151367187, "completions/min_length": 23.0, "completions/min_terminated_length": 23.0, "epoch": 0.08, "grad_norm": 0.0027302594389766455, "learning_rate": 1e-06, "loss": 0.0167, "num_tokens": 90992867.0, "reward": 0.7987664818763733, "reward_std": 0.3233607649803162, "rewards/accuracy_reward": 0.2138671875, "rewards/brier_reward": 0.41909300684928896, "rewards/confidence_one_or_zero": 0.49765625, "rewards/format_reward": 0.96513671875, "rewards/mean_confidence_reward": 0.7765803933143616, "rewards/smce": -0.5664987325668335, "step": 25 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00634765625, "completions/max_length": 1536.0, "completions/max_terminated_length": 1184.8, "completions/mean_length": 135.8953125, "completions/mean_terminated_length": 126.9502456665039, "completions/min_length": 27.8, "completions/min_terminated_length": 27.8, "epoch": 0.096, "grad_norm": 0.002199189504608512, "learning_rate": 1e-06, "loss": 0.0104, "num_tokens": 107429043.0, "reward": 0.8861237287521362, "reward_std": 0.2804286539554596, "rewards/accuracy_reward": 0.23486328125, "rewards/brier_reward": 0.5540830910205841, "rewards/confidence_one_or_zero": 0.3076171875, "rewards/format_reward": 0.98369140625, "rewards/mean_confidence_reward": 0.6132706642150879, "rewards/smce": -0.39209535121917727, "step": 30 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.003515625, "completions/max_length": 1536.0, "completions/max_terminated_length": 1061.6, "completions/mean_length": 121.37578125, "completions/mean_terminated_length": 116.38601531982422, "completions/min_length": 36.2, "completions/min_terminated_length": 36.2, "epoch": 0.112, "grad_norm": 0.0014616711996495724, "learning_rate": 1e-06, "loss": 0.0069, "num_tokens": 123781403.0, "reward": 0.989423644542694, "reward_std": 0.19730665683746337, "rewards/accuracy_reward": 0.26953125, "rewards/brier_reward": 0.7158191919326782, "rewards/confidence_one_or_zero": 0.18125, "rewards/format_reward": 0.99365234375, "rewards/mean_confidence_reward": 0.38006125688552855, "rewards/smce": -0.15658431351184846, "step": 35 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0021484375, "completions/max_length": 1536.0, "completions/max_terminated_length": 876.8, "completions/mean_length": 115.8734375, "completions/mean_terminated_length": 112.8156021118164, "completions/min_length": 37.0, "completions/min_terminated_length": 37.0, "epoch": 0.128, "grad_norm": 0.0011532431235536933, "learning_rate": 1e-06, "loss": 0.0045, "num_tokens": 139884619.0, "reward": 1.0200982809066772, "reward_std": 0.13898254334926605, "rewards/accuracy_reward": 0.27294921875, "rewards/brier_reward": 0.7711090922355652, "rewards/confidence_one_or_zero": 0.1431640625, "rewards/format_reward": 0.99619140625, "rewards/mean_confidence_reward": 0.263618403673172, "rewards/smce": -0.053969383239746094, "step": 40 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00078125, "completions/max_length": 1296.2, "completions/max_terminated_length": 526.0, "completions/mean_length": 111.2216796875, "completions/mean_terminated_length": 110.10794219970703, "completions/min_length": 45.0, "completions/min_terminated_length": 45.0, "epoch": 0.144, "grad_norm": 0.0008673432166688144, "learning_rate": 1e-06, "loss": 0.002, "num_tokens": 155973961.0, "reward": 1.0399757146835327, "reward_std": 0.1108901247382164, "rewards/accuracy_reward": 0.32666015625, "rewards/brier_reward": 0.7554583191871643, "rewards/confidence_one_or_zero": 0.09560546875, "rewards/format_reward": 0.99794921875, "rewards/mean_confidence_reward": 0.21799394190311433, "rewards/smce": -0.11687580943107605, "step": 45 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0005859375, "completions/max_length": 1087.6, "completions/max_terminated_length": 440.8, "completions/mean_length": 112.95947265625, "completions/mean_terminated_length": 112.12568817138671, "completions/min_length": 50.6, "completions/min_terminated_length": 50.6, "epoch": 0.16, "grad_norm": 0.0007697119144722819, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 172151594.0, "reward": 1.041567873954773, "reward_std": 0.09320179969072342, "rewards/accuracy_reward": 0.32041015625, "rewards/brier_reward": 0.764100170135498, "rewards/confidence_one_or_zero": 0.05087890625, "rewards/format_reward": 0.99873046875, "rewards/mean_confidence_reward": 0.21918963193893432, "rewards/smce": -0.10564200878143311, "step": 50 }, { "epoch": 0.16, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 332.5, "eval_completions/max_terminated_length": 332.5, "eval_completions/mean_length": 114.48052978515625, "eval_completions/mean_terminated_length": 114.48052978515625, "eval_completions/min_length": 60.0, "eval_completions/min_terminated_length": 60.0, "eval_loss": 0.0, "eval_num_tokens": 172151594.0, "eval_reward": 1.0331834554672241, "eval_reward_std": 0.125154048204422, "eval_rewards/accuracy_reward": 0.267578125, "eval_rewards/brier_reward": 0.7988348007202148, "eval_rewards/confidence_one_or_zero": 0.037109375, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.22960937768220901, "eval_rewards/smce": -0.04650058038532734, "eval_runtime": 12.04, "eval_samples_per_second": 41.528, "eval_steps_per_second": 0.166, "step": 50 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00068359375, "completions/max_length": 1417.4, "completions/max_terminated_length": 519.0, "completions/mean_length": 113.7537109375, "completions/mean_terminated_length": 112.78092041015626, "completions/min_length": 49.2, "completions/min_terminated_length": 49.2, "epoch": 0.176, "grad_norm": 0.0008332604193128645, "learning_rate": 1e-06, "loss": 0.0015, "num_tokens": 188553552.0, "reward": 1.0444111824035645, "reward_std": 0.093848617374897, "rewards/accuracy_reward": 0.3125, "rewards/brier_reward": 0.7778581500053405, "rewards/confidence_one_or_zero": 0.0224609375, "rewards/format_reward": 0.99853515625, "rewards/mean_confidence_reward": 0.24422344267368318, "rewards/smce": -0.0715064637362957, "step": 55 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 855.2, "completions/max_terminated_length": 396.8, "completions/mean_length": 114.6798828125, "completions/mean_terminated_length": 114.40217590332031, "completions/min_length": 53.8, "completions/min_terminated_length": 53.8, "epoch": 0.192, "grad_norm": 0.0008707884699106216, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 204542690.0, "reward": 1.062576985359192, "reward_std": 0.09943905770778656, "rewards/accuracy_reward": 0.353515625, "rewards/brier_reward": 0.7720983743667602, "rewards/confidence_one_or_zero": 0.008203125, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.29078262448310854, "rewards/smce": -0.0698842965066433, "step": 60 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00048828125, "completions/max_length": 1290.6, "completions/max_terminated_length": 299.4, "completions/mean_length": 121.6306640625, "completions/mean_terminated_length": 120.94003753662109, "completions/min_length": 47.6, "completions/min_terminated_length": 47.6, "epoch": 0.208, "grad_norm": 0.0007853489951230586, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 220820412.0, "reward": 1.0697309255599976, "reward_std": 0.11286971271038056, "rewards/accuracy_reward": 0.3732421875, "rewards/brier_reward": 0.7671383023262024, "rewards/confidence_one_or_zero": 0.00322265625, "rewards/format_reward": 0.99912109375, "rewards/mean_confidence_reward": 0.3380278408527374, "rewards/smce": -0.04038693383336067, "step": 65 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 824.0, "completions/max_terminated_length": 399.2, "completions/mean_length": 123.864453125, "completions/mean_terminated_length": 123.31305847167968, "completions/min_length": 59.2, "completions/min_terminated_length": 59.2, "epoch": 0.224, "grad_norm": 0.0009797868551686406, "learning_rate": 1e-06, "loss": 0.0016, "num_tokens": 237241968.0, "reward": 1.0696279525756835, "reward_std": 0.11271759867668152, "rewards/accuracy_reward": 0.36845703125, "rewards/brier_reward": 0.7713197946548462, "rewards/confidence_one_or_zero": 0.00244140625, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.3636253893375397, "rewards/smce": -0.033345287293195726, "step": 70 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 531.2, "completions/max_terminated_length": 279.2, "completions/mean_length": 122.9078125, "completions/mean_terminated_length": 122.76952819824218, "completions/min_length": 60.0, "completions/min_terminated_length": 60.0, "epoch": 0.24, "grad_norm": 0.0009892035741358995, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 253752224.0, "reward": 1.0851356983184814, "reward_std": 0.12177922129631043, "rewards/accuracy_reward": 0.40810546875, "rewards/brier_reward": 0.7624287366867065, "rewards/confidence_one_or_zero": 0.0005859375, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.3889783203601837, "rewards/smce": -0.06828636862337589, "step": 75 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 296.4, "completions/max_terminated_length": 296.4, "completions/mean_length": 119.69375, "completions/mean_terminated_length": 119.69375, "completions/min_length": 58.6, "completions/min_terminated_length": 58.6, "epoch": 0.256, "grad_norm": 0.0009709701989777386, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 270032704.0, "reward": 1.0745017528533936, "reward_std": 0.11845734417438507, "rewards/accuracy_reward": 0.3814453125, "rewards/brier_reward": 0.7679124712944031, "rewards/confidence_one_or_zero": 0.0009765625, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4074843883514404, "rewards/smce": -0.062050586566329, "step": 80 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 543.8, "completions/max_terminated_length": 303.4, "completions/mean_length": 121.2091796875, "completions/mean_terminated_length": 121.07110595703125, "completions/min_length": 58.6, "completions/min_terminated_length": 58.6, "epoch": 0.272, "grad_norm": 0.0009048814536072314, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 286239582.0, "reward": 1.0838090896606445, "reward_std": 0.11522045135498046, "rewards/accuracy_reward": 0.40185546875, "rewards/brier_reward": 0.7661949157714844, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.4112128973007202, "rewards/smce": -0.04232981689274311, "step": 85 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 314.4, "completions/max_terminated_length": 314.4, "completions/mean_length": 119.34541015625, "completions/mean_terminated_length": 119.34541015625, "completions/min_length": 58.4, "completions/min_terminated_length": 58.4, "epoch": 0.288, "grad_norm": 0.0011435942724347115, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 302419855.0, "reward": 1.0826804637908936, "reward_std": 0.114729243516922, "rewards/accuracy_reward": 0.39580078125, "rewards/brier_reward": 0.7699063777923584, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4135368287563324, "rewards/smce": -0.05406544618308544, "step": 90 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 628.0, "completions/max_terminated_length": 370.6, "completions/mean_length": 119.10458984375, "completions/mean_terminated_length": 118.96620635986328, "completions/min_length": 58.6, "completions/min_terminated_length": 58.6, "epoch": 0.304, "grad_norm": 0.0010342359310016036, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 318569438.0, "reward": 1.0892314672470094, "reward_std": 0.11390392035245896, "rewards/accuracy_reward": 0.41328125, "rewards/brier_reward": 0.7654102206230163, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.407882821559906, "rewards/smce": -0.03404633477330208, "step": 95 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 535.6, "completions/max_terminated_length": 319.8, "completions/mean_length": 120.22197265625, "completions/mean_terminated_length": 119.94470672607422, "completions/min_length": 59.6, "completions/min_terminated_length": 59.6, "epoch": 0.32, "grad_norm": 0.0009286344284191728, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 334889215.0, "reward": 1.0949872255325317, "reward_std": 0.10952811241149903, "rewards/accuracy_reward": 0.419921875, "rewards/brier_reward": 0.7704041719436645, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.41058760285377505, "rewards/smce": -0.05937949083745479, "step": 100 }, { "epoch": 0.32, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 223.0, "eval_completions/max_terminated_length": 223.0, "eval_completions/mean_length": 120.3595962524414, "eval_completions/mean_terminated_length": 120.3595962524414, "eval_completions/min_length": 66.5, "eval_completions/min_terminated_length": 66.5, "eval_loss": 0.0, "eval_num_tokens": 334889215.0, "eval_reward": 1.0504685044288635, "eval_reward_std": 0.2088107019662857, "eval_rewards/accuracy_reward": 0.318359375, "eval_rewards/brier_reward": 0.7826702892780304, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4114453196525574, "eval_rewards/smce": -0.09349317103624344, "eval_runtime": 9.4015, "eval_samples_per_second": 53.183, "eval_steps_per_second": 0.213, "step": 100 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 534.8, "completions/max_terminated_length": 289.0, "completions/mean_length": 117.00869140625, "completions/mean_terminated_length": 116.86989135742188, "completions/min_length": 57.6, "completions/min_terminated_length": 57.6, "epoch": 0.336, "grad_norm": 0.0009752543410286307, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 350809816.0, "reward": 1.0989506721496582, "reward_std": 0.1054120883345604, "rewards/accuracy_reward": 0.4306640625, "rewards/brier_reward": 0.7674848198890686, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4077809572219849, "rewards/smce": -0.05311233140528202, "step": 105 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 273.0, "completions/max_terminated_length": 273.0, "completions/mean_length": 118.835546875, "completions/mean_terminated_length": 118.835546875, "completions/min_length": 56.8, "completions/min_terminated_length": 56.8, "epoch": 0.352, "grad_norm": 0.0008025058778002858, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 367287108.0, "reward": 1.0761260509490966, "reward_std": 0.09988621026277542, "rewards/accuracy_reward": 0.37802734375, "rewards/brier_reward": 0.7743457198143006, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.39769991040229796, "rewards/smce": -0.02415503077208996, "step": 110 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 303.8, "completions/max_terminated_length": 303.8, "completions/mean_length": 124.12080078125, "completions/mean_terminated_length": 124.12080078125, "completions/min_length": 60.6, "completions/min_terminated_length": 60.6, "epoch": 0.368, "grad_norm": 0.0008689153473824263, "learning_rate": 1e-06, "loss": -0.0002, "num_tokens": 383623577.0, "reward": 1.0884521007537842, "reward_std": 0.09868751466274261, "rewards/accuracy_reward": 0.39931640625, "rewards/brier_reward": 0.7777201294898987, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4137275397777557, "rewards/smce": -0.03548511806875467, "step": 115 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 559.2, "completions/max_terminated_length": 303.4, "completions/mean_length": 130.14931640625, "completions/mean_terminated_length": 130.01228942871094, "completions/min_length": 61.8, "completions/min_terminated_length": 61.8, "epoch": 0.384, "grad_norm": 0.0009342688135802746, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 399812818.0, "reward": 1.1017422437667848, "reward_std": 0.10270306766033173, "rewards/accuracy_reward": 0.42939453125, "rewards/brier_reward": 0.7743234157562255, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.41212044954299926, "rewards/smce": -0.03892633318901062, "step": 120 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 838.0, "completions/max_terminated_length": 341.0, "completions/mean_length": 138.803515625, "completions/mean_terminated_length": 138.5307159423828, "completions/min_length": 57.0, "completions/min_terminated_length": 57.0, "epoch": 0.4, "grad_norm": 0.0009262771927751601, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 416270614.0, "reward": 1.0920633316040038, "reward_std": 0.10999695956707001, "rewards/accuracy_reward": 0.40185546875, "rewards/brier_reward": 0.7826905965805053, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.4225615203380585, "rewards/smce": -0.02956448132172227, "step": 125 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00048828125, "completions/max_length": 1100.8, "completions/max_terminated_length": 430.0, "completions/mean_length": 149.11748046875, "completions/mean_terminated_length": 148.43955993652344, "completions/min_length": 70.6, "completions/min_terminated_length": 70.6, "epoch": 0.416, "grad_norm": 0.0009191810968331993, "learning_rate": 1e-06, "loss": 0.002, "num_tokens": 432678761.0, "reward": 1.0989661931991577, "reward_std": 0.11646136194467545, "rewards/accuracy_reward": 0.421875, "rewards/brier_reward": 0.7765630722045899, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.42575586438179014, "rewards/smce": -0.018213643692433834, "step": 130 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 1067.2, "completions/max_terminated_length": 394.8, "completions/mean_length": 152.19921875, "completions/mean_terminated_length": 151.7940216064453, "completions/min_length": 69.6, "completions/min_terminated_length": 69.6, "epoch": 0.432, "grad_norm": 0.0008486987790092826, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 449251617.0, "reward": 1.1061060428619385, "reward_std": 0.10609066933393478, "rewards/accuracy_reward": 0.43486328125, "rewards/brier_reward": 0.777781867980957, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.4303037166595459, "rewards/smce": -0.04326645694673061, "step": 135 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 859.4, "completions/max_terminated_length": 387.0, "completions/mean_length": 152.56455078125, "completions/mean_terminated_length": 152.29410705566406, "completions/min_length": 70.6, "completions/min_terminated_length": 70.6, "epoch": 0.448, "grad_norm": 0.0008342221262864769, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 465766678.0, "reward": 1.101392149925232, "reward_std": 0.1103613555431366, "rewards/accuracy_reward": 0.425, "rewards/brier_reward": 0.778016197681427, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4370110690593719, "rewards/smce": -0.037433745712041853, "step": 140 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 847.6, "completions/max_terminated_length": 389.0, "completions/mean_length": 154.2365234375, "completions/mean_terminated_length": 153.96669311523436, "completions/min_length": 67.6, "completions/min_terminated_length": 67.6, "epoch": 0.464, "grad_norm": 0.0007826949004083872, "learning_rate": 1e-06, "loss": 0.0014, "num_tokens": 482516876.0, "reward": 1.0809687376022339, "reward_std": 0.10724099725484848, "rewards/accuracy_reward": 0.37392578125, "rewards/brier_reward": 0.7883727312088012, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4257929682731628, "rewards/smce": -0.0688736081123352, "step": 145 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 589.0, "completions/max_terminated_length": 358.6, "completions/mean_length": 150.88056640625, "completions/mean_terminated_length": 150.74555358886718, "completions/min_length": 70.4, "completions/min_terminated_length": 70.4, "epoch": 0.48, "grad_norm": 0.0009045750484801829, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 499109925.0, "reward": 1.1116684198379516, "reward_std": 0.10544286221265793, "rewards/accuracy_reward": 0.4521484375, "rewards/brier_reward": 0.7713385105133057, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4053613364696503, "rewards/smce": -0.0533256433904171, "step": 150 }, { "epoch": 0.48, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 330.5, "eval_completions/max_terminated_length": 330.5, "eval_completions/mean_length": 151.33397674560547, "eval_completions/mean_terminated_length": 151.33397674560547, "eval_completions/min_length": 86.5, "eval_completions/min_terminated_length": 86.5, "eval_loss": 0.0, "eval_num_tokens": 499109925.0, "eval_reward": 1.0832970142364502, "eval_reward_std": 0.21396854519844055, "eval_rewards/accuracy_reward": 0.37109375, "eval_rewards/brier_reward": 0.7955322265625, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.40166015923023224, "eval_rewards/smce": -0.03275391226634383, "eval_runtime": 11.706, "eval_samples_per_second": 42.713, "eval_steps_per_second": 0.171, "step": 150 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 633.4, "completions/max_terminated_length": 399.6, "completions/mean_length": 146.53203125, "completions/mean_terminated_length": 146.39669189453124, "completions/min_length": 75.2, "completions/min_terminated_length": 75.2, "epoch": 0.496, "grad_norm": 0.001076922519132495, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 515918253.0, "reward": 1.1104183912277221, "reward_std": 0.11128229051828384, "rewards/accuracy_reward": 0.44697265625, "rewards/brier_reward": 0.7739948391914367, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4203671872615814, "rewards/smce": -0.03376689273864031, "step": 155 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 396.8, "completions/max_terminated_length": 396.8, "completions/mean_length": 146.97705078125, "completions/mean_terminated_length": 146.97705078125, "completions/min_length": 73.0, "completions/min_terminated_length": 73.0, "epoch": 0.512, "grad_norm": 0.000939170247875154, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 532568962.0, "reward": 1.1241384267807006, "reward_std": 0.1158647045493126, "rewards/accuracy_reward": 0.46650390625, "rewards/brier_reward": 0.7818114757537842, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.46039745807647703, "rewards/smce": -0.039459402486681935, "step": 160 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 599.0, "completions/max_terminated_length": 371.8, "completions/mean_length": 153.444921875, "completions/mean_terminated_length": 153.31015014648438, "completions/min_length": 77.6, "completions/min_terminated_length": 77.6, "epoch": 0.528, "grad_norm": 0.0010173226473852992, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 549169774.0, "reward": 1.115717363357544, "reward_std": 0.12228884994983673, "rewards/accuracy_reward": 0.443359375, "rewards/brier_reward": 0.7883172273635864, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.48874316215515134, "rewards/smce": -0.047541006654500964, "step": 165 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 354.0, "completions/max_terminated_length": 354.0, "completions/mean_length": 154.3060546875, "completions/mean_terminated_length": 154.3060546875, "completions/min_length": 72.4, "completions/min_terminated_length": 72.4, "epoch": 0.544, "grad_norm": 0.000949379347730428, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 565913452.0, "reward": 1.1308411598205566, "reward_std": 0.12379086166620254, "rewards/accuracy_reward": 0.48779296875, "rewards/brier_reward": 0.7740159749984741, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4975504994392395, "rewards/smce": -0.029937567189335823, "step": 170 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 449.4, "completions/max_terminated_length": 449.4, "completions/mean_length": 152.90234375, "completions/mean_terminated_length": 152.90234375, "completions/min_length": 75.0, "completions/min_terminated_length": 75.0, "epoch": 0.56, "grad_norm": 0.0007226327434182167, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 582300580.0, "reward": 1.1068236112594605, "reward_std": 0.10674239695072174, "rewards/accuracy_reward": 0.4203125, "rewards/brier_reward": 0.793374526500702, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4588545083999634, "rewards/smce": -0.04062653072178364, "step": 175 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 405.6, "completions/max_terminated_length": 405.6, "completions/mean_length": 154.55283203125, "completions/mean_terminated_length": 154.55283203125, "completions/min_length": 75.0, "completions/min_terminated_length": 75.0, "epoch": 0.576, "grad_norm": 0.0007687191246077418, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 599069825.0, "reward": 1.1068036079406738, "reward_std": 0.08727760165929795, "rewards/accuracy_reward": 0.4296875, "rewards/brier_reward": 0.7839844107627869, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.3939257860183716, "rewards/smce": -0.06543606072664261, "step": 180 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 681.2, "completions/max_terminated_length": 451.2, "completions/mean_length": 157.06923828125, "completions/mean_terminated_length": 156.93432006835937, "completions/min_length": 77.8, "completions/min_terminated_length": 77.8, "epoch": 0.592, "grad_norm": 0.0007763951434753835, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 615845926.0, "reward": 1.1132864475250244, "reward_std": 0.09145474284887314, "rewards/accuracy_reward": 0.4427734375, "rewards/brier_reward": 0.7840348005294799, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4050097703933716, "rewards/smce": -0.04077463261783123, "step": 185 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 839.2, "completions/max_terminated_length": 394.2, "completions/mean_length": 160.98759765625, "completions/mean_terminated_length": 160.71965942382812, "completions/min_length": 78.8, "completions/min_terminated_length": 78.8, "epoch": 0.608, "grad_norm": 0.0008273112471215427, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 632493927.0, "reward": 1.1193994283676147, "reward_std": 0.10334119200706482, "rewards/accuracy_reward": 0.43974609375, "rewards/brier_reward": 0.7992767453193664, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.43347753286361695, "rewards/smce": -0.02950924411416054, "step": 190 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 662.8, "completions/max_terminated_length": 487.4, "completions/mean_length": 166.30029296875, "completions/mean_terminated_length": 166.16661071777344, "completions/min_length": 81.6, "completions/min_terminated_length": 81.6, "epoch": 0.624, "grad_norm": 0.001086984295397997, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 649540746.0, "reward": 1.110041618347168, "reward_std": 0.11234649866819382, "rewards/accuracy_reward": 0.43046875, "rewards/brier_reward": 0.7898535251617431, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4636425793170929, "rewards/smce": -0.04460376352071762, "step": 195 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 415.8, "completions/max_terminated_length": 415.8, "completions/mean_length": 166.65166015625, "completions/mean_terminated_length": 166.65166015625, "completions/min_length": 85.2, "completions/min_terminated_length": 85.2, "epoch": 0.64, "grad_norm": 0.0008566314354538918, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 666589947.0, "reward": 1.1489856958389282, "reward_std": 0.10852711945772171, "rewards/accuracy_reward": 0.50615234375, "rewards/brier_reward": 0.7918631672859192, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.49933398962020875, "rewards/smce": -0.045029871026054026, "step": 200 }, { "epoch": 0.64, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 366.0, "eval_completions/max_terminated_length": 366.0, "eval_completions/mean_length": 166.41905212402344, "eval_completions/mean_terminated_length": 166.41905212402344, "eval_completions/min_length": 86.0, "eval_completions/min_terminated_length": 86.0, "eval_loss": 0.0, "eval_num_tokens": 666589947.0, "eval_reward": 1.0801602602005005, "eval_reward_std": 0.2655382305383682, "eval_rewards/accuracy_reward": 0.3671875, "eval_rewards/brier_reward": 0.7932590246200562, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4937109351158142, "eval_rewards/smce": -0.12693358957767487, "eval_runtime": 12.0819, "eval_samples_per_second": 41.384, "eval_steps_per_second": 0.166, "step": 200 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 873.0, "completions/max_terminated_length": 431.8, "completions/mean_length": 163.4501953125, "completions/mean_terminated_length": 163.04900207519532, "completions/min_length": 84.6, "completions/min_terminated_length": 84.6, "epoch": 0.656, "grad_norm": 0.0011378537165001035, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 683120221.0, "reward": 1.1000010251998902, "reward_std": 0.11923554241657257, "rewards/accuracy_reward": 0.4224609375, "rewards/brier_reward": 0.7779239058494568, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.5125224709510803, "rewards/smce": -0.09081879109144211, "step": 205 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 1085.8, "completions/max_terminated_length": 413.4, "completions/mean_length": 163.33291015625, "completions/mean_terminated_length": 162.79642333984376, "completions/min_length": 78.6, "completions/min_terminated_length": 78.6, "epoch": 0.672, "grad_norm": 0.0006824737647548318, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 699706190.0, "reward": 1.1262107133865356, "reward_std": 0.10164524465799332, "rewards/accuracy_reward": 0.4546875, "rewards/brier_reward": 0.7982605934143067, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.4685644626617432, "rewards/smce": -0.03929471783339977, "step": 210 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 375.2, "completions/max_terminated_length": 375.2, "completions/mean_length": 161.79052734375, "completions/mean_terminated_length": 161.79052734375, "completions/min_length": 75.4, "completions/min_terminated_length": 75.4, "epoch": 0.688, "grad_norm": 0.0009157939930446446, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 716316845.0, "reward": 1.1289687395095824, "reward_std": 0.09724249839782714, "rewards/accuracy_reward": 0.4564453125, "rewards/brier_reward": 0.8015339732170105, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4269199311733246, "rewards/smce": -0.04264727309346199, "step": 215 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 865.2, "completions/max_terminated_length": 491.2, "completions/mean_length": 162.37724609375, "completions/mean_terminated_length": 162.10917053222656, "completions/min_length": 84.2, "completions/min_terminated_length": 84.2, "epoch": 0.704, "grad_norm": 0.0008008752483874559, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 732845732.0, "reward": 1.1373294115066528, "reward_std": 0.0963486909866333, "rewards/accuracy_reward": 0.4806640625, "rewards/brier_reward": 0.7942330479621887, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.43858984112739563, "rewards/smce": -0.04382671192288399, "step": 220 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 857.6, "completions/max_terminated_length": 392.6, "completions/mean_length": 161.50390625, "completions/mean_terminated_length": 161.1012969970703, "completions/min_length": 69.6, "completions/min_terminated_length": 69.6, "epoch": 0.72, "grad_norm": 0.000978630967438221, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 749509388.0, "reward": 1.1495690107345582, "reward_std": 0.10505697280168533, "rewards/accuracy_reward": 0.4892578125, "rewards/brier_reward": 0.8102994561195374, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.49882130026817323, "rewards/smce": -0.02960990173742175, "step": 225 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 826.8, "completions/max_terminated_length": 391.4, "completions/mean_length": 159.55869140625, "completions/mean_terminated_length": 159.15543518066406, "completions/min_length": 83.0, "completions/min_terminated_length": 83.0, "epoch": 0.736, "grad_norm": 0.0009836617391556501, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 766082853.0, "reward": 1.1441120862960816, "reward_std": 0.10725852996110916, "rewards/accuracy_reward": 0.496484375, "rewards/brier_reward": 0.7920857787132263, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.5160800814628601, "rewards/smce": -0.05401993244886398, "step": 230 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 600.8, "completions/max_terminated_length": 384.6, "completions/mean_length": 159.32666015625, "completions/mean_terminated_length": 159.19202270507813, "completions/min_length": 76.2, "completions/min_terminated_length": 76.2, "epoch": 0.752, "grad_norm": 0.0010423972271382809, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 782941558.0, "reward": 1.1330698251724243, "reward_std": 0.10301287025213242, "rewards/accuracy_reward": 0.470703125, "rewards/brier_reward": 0.7955723881721497, "rewards/confidence_one_or_zero": 0.00048828125, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4687236487865448, "rewards/smce": -0.039099032431840895, "step": 235 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 858.4, "completions/max_terminated_length": 438.0, "completions/mean_length": 165.8583984375, "completions/mean_terminated_length": 165.32333984375, "completions/min_length": 80.0, "completions/min_terminated_length": 80.0, "epoch": 0.768, "grad_norm": 0.0008880188688635826, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 799572652.0, "reward": 1.1417131900787354, "reward_std": 0.0946048617362976, "rewards/accuracy_reward": 0.4765625, "rewards/brier_reward": 0.8072758674621582, "rewards/confidence_one_or_zero": 0.00048828125, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.4606347680091858, "rewards/smce": -0.02230907715857029, "step": 240 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 455.6, "completions/max_terminated_length": 455.6, "completions/mean_length": 166.615234375, "completions/mean_terminated_length": 166.615234375, "completions/min_length": 81.0, "completions/min_terminated_length": 81.0, "epoch": 0.784, "grad_norm": 0.0008837269269861281, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 816453160.0, "reward": 1.142217469215393, "reward_std": 0.10188348740339279, "rewards/accuracy_reward": 0.49140625, "rewards/brier_reward": 0.7931073784828186, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4713378846645355, "rewards/smce": -0.07962073720991611, "step": 245 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 1034.2, "completions/max_terminated_length": 594.8, "completions/mean_length": 165.168359375, "completions/mean_terminated_length": 164.90091857910156, "completions/min_length": 85.0, "completions/min_terminated_length": 85.0, "epoch": 0.8, "grad_norm": 0.001023618970066309, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 833155044.0, "reward": 1.1532862901687622, "reward_std": 0.11082861423492432, "rewards/accuracy_reward": 0.50244140625, "rewards/brier_reward": 0.8043578028678894, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.5133701205253601, "rewards/smce": -0.03228470087051392, "step": 250 }, { "epoch": 0.8, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 346.0, "eval_completions/max_terminated_length": 346.0, "eval_completions/mean_length": 169.8637924194336, "eval_completions/mean_terminated_length": 169.8637924194336, "eval_completions/min_length": 96.0, "eval_completions/min_terminated_length": 96.0, "eval_loss": 0.0, "eval_num_tokens": 833155044.0, "eval_reward": 1.0787333250045776, "eval_reward_std": 0.2926654517650604, "eval_rewards/accuracy_reward": 0.376953125, "eval_rewards/brier_reward": 0.7806787192821503, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.5430664122104645, "eval_rewards/smce": -0.16611327975988388, "eval_runtime": 12.1823, "eval_samples_per_second": 41.043, "eval_steps_per_second": 0.164, "step": 250 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 589.6, "completions/max_terminated_length": 589.6, "completions/mean_length": 160.68017578125, "completions/mean_terminated_length": 160.68017578125, "completions/min_length": 82.2, "completions/min_terminated_length": 82.2, "epoch": 0.816, "grad_norm": 0.0013014374999329448, "learning_rate": 1e-06, "loss": -0.0002, "num_tokens": 849899577.0, "reward": 1.144900870323181, "reward_std": 0.12142248898744583, "rewards/accuracy_reward": 0.49375, "rewards/brier_reward": 0.796118152141571, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.5607226490974426, "rewards/smce": -0.06757495403289795, "step": 255 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 644.2, "completions/max_terminated_length": 443.2, "completions/mean_length": 163.26494140625, "completions/mean_terminated_length": 163.13106994628907, "completions/min_length": 83.2, "completions/min_terminated_length": 83.2, "epoch": 0.832, "grad_norm": 0.0008710598922334611, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 866579762.0, "reward": 1.1563215732574463, "reward_std": 0.09912585914134979, "rewards/accuracy_reward": 0.50263671875, "rewards/brier_reward": 0.8101438045501709, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4804052829742432, "rewards/smce": -0.04059399105608463, "step": 260 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 390.2, "completions/max_terminated_length": 390.2, "completions/mean_length": 160.62734375, "completions/mean_terminated_length": 160.62734375, "completions/min_length": 83.2, "completions/min_terminated_length": 83.2, "epoch": 0.848, "grad_norm": 0.0011734205763787031, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 883238954.0, "reward": 1.139661955833435, "reward_std": 0.0918037623167038, "rewards/accuracy_reward": 0.47666015625, "rewards/brier_reward": 0.8027248620986939, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4152099609375, "rewards/smce": -0.06195409372448921, "step": 265 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 678.8, "completions/max_terminated_length": 450.0, "completions/mean_length": 168.620703125, "completions/mean_terminated_length": 168.48779296875, "completions/min_length": 85.2, "completions/min_terminated_length": 85.2, "epoch": 0.864, "grad_norm": 0.0009395951055921614, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 899952446.0, "reward": 1.150853943824768, "reward_std": 0.09571457803249359, "rewards/accuracy_reward": 0.5037109375, "rewards/brier_reward": 0.7982314586639404, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4721386730670929, "rewards/smce": -0.04000121615827083, "step": 270 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 460.8, "completions/max_terminated_length": 460.8, "completions/mean_length": 176.70556640625, "completions/mean_terminated_length": 176.70556640625, "completions/min_length": 81.8, "completions/min_terminated_length": 81.8, "epoch": 0.88, "grad_norm": 0.001014765352010727, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 916908983.0, "reward": 1.1355210781097411, "reward_std": 0.09766919016838074, "rewards/accuracy_reward": 0.46201171875, "rewards/brier_reward": 0.8090754985809326, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4960673928260803, "rewards/smce": -0.04611279219388962, "step": 275 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 871.6, "completions/max_terminated_length": 611.4, "completions/mean_length": 184.18115234375, "completions/mean_terminated_length": 183.91699829101563, "completions/min_length": 91.6, "completions/min_terminated_length": 91.6, "epoch": 0.896, "grad_norm": 0.0010061163920909166, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 933905846.0, "reward": 1.1420814514160156, "reward_std": 0.09720710813999175, "rewards/accuracy_reward": 0.478125, "rewards/brier_reward": 0.8063697218894958, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.48104395270347594, "rewards/smce": -0.039731436781585215, "step": 280 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 1086.6, "completions/max_terminated_length": 438.4, "completions/mean_length": 186.08662109375, "completions/mean_terminated_length": 185.6914276123047, "completions/min_length": 94.8, "completions/min_terminated_length": 94.8, "epoch": 0.912, "grad_norm": 0.00092413614038378, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 950862669.0, "reward": 1.13735990524292, "reward_std": 0.09484986811876298, "rewards/accuracy_reward": 0.47255859375, "rewards/brier_reward": 0.8024985313415527, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4633896589279175, "rewards/smce": -0.04536058492958546, "step": 285 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 1162.8, "completions/max_terminated_length": 554.2, "completions/mean_length": 186.5712890625, "completions/mean_terminated_length": 186.04344482421874, "completions/min_length": 91.0, "completions/min_terminated_length": 91.0, "epoch": 0.928, "grad_norm": 0.0008485357975587249, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 967799975.0, "reward": 1.1491246938705444, "reward_std": 0.08749246448278428, "rewards/accuracy_reward": 0.505859375, "rewards/brier_reward": 0.7928359508514404, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.45264649391174316, "rewards/smce": -0.05614354871213436, "step": 290 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 1129.4, "completions/max_terminated_length": 641.6, "completions/mean_length": 187.45673828125, "completions/mean_terminated_length": 187.06110534667968, "completions/min_length": 89.2, "completions/min_terminated_length": 89.2, "epoch": 0.944, "grad_norm": 0.0009085385827347636, "learning_rate": 1e-06, "loss": 0.0011, "num_tokens": 984694956.0, "reward": 1.1451599836349486, "reward_std": 0.09136117100715638, "rewards/accuracy_reward": 0.4962890625, "rewards/brier_reward": 0.7944116115570068, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4501757860183716, "rewards/smce": -0.0885808089748025, "step": 295 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 755.6, "completions/max_terminated_length": 587.0, "completions/mean_length": 187.70068359375, "completions/mean_terminated_length": 187.43772583007814, "completions/min_length": 91.4, "completions/min_terminated_length": 91.4, "epoch": 0.96, "grad_norm": 0.000844655092805624, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 1001557331.0, "reward": 1.1376837253570558, "reward_std": 0.0915878638625145, "rewards/accuracy_reward": 0.4642578125, "rewards/brier_reward": 0.8113359332084655, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4858310461044312, "rewards/smce": -0.031868165358901024, "step": 300 }, { "epoch": 0.96, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 392.5, "eval_completions/max_terminated_length": 392.5, "eval_completions/mean_length": 191.19613647460938, "eval_completions/mean_terminated_length": 191.19613647460938, "eval_completions/min_length": 92.0, "eval_completions/min_terminated_length": 92.0, "eval_loss": 0.0, "eval_num_tokens": 1001557331.0, "eval_reward": 1.1049875020980835, "eval_reward_std": 0.26456505060195923, "eval_rewards/accuracy_reward": 0.40625, "eval_rewards/brier_reward": 0.803798645734787, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4793164134025574, "eval_rewards/smce": -0.07452617585659027, "eval_runtime": 12.6872, "eval_samples_per_second": 39.41, "eval_steps_per_second": 0.158, "step": 300 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 958.4, "completions/max_terminated_length": 630.4, "completions/mean_length": 189.08857421875, "completions/mean_terminated_length": 188.6948211669922, "completions/min_length": 96.4, "completions/min_terminated_length": 96.4, "epoch": 0.976, "grad_norm": 0.000775482039898634, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 1018354718.0, "reward": 1.1589978218078614, "reward_std": 0.08740981668233871, "rewards/accuracy_reward": 0.50595703125, "rewards/brier_reward": 0.8123676776885986, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4781347870826721, "rewards/smce": -0.037019770964980124, "step": 305 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0005859375, "completions/max_length": 1359.8, "completions/max_terminated_length": 520.6, "completions/mean_length": 187.88623046875, "completions/mean_terminated_length": 187.09657287597656, "completions/min_length": 92.6, "completions/min_terminated_length": 92.6, "epoch": 0.992, "grad_norm": 0.0007895891321823001, "learning_rate": 1e-06, "loss": 0.0015, "num_tokens": 1035407153.0, "reward": 1.1390615701675415, "reward_std": 0.09859338849782943, "rewards/accuracy_reward": 0.4849609375, "rewards/brier_reward": 0.7937985897064209, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.9994140625, "rewards/mean_confidence_reward": 0.44668457508087156, "rewards/smce": -0.05136181376874447, "step": 310 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000732421875, "completions/max_length": 989.0, "completions/max_terminated_length": 730.5, "completions/mean_length": 187.27367401123047, "completions/mean_terminated_length": 186.28497314453125, "completions/min_length": 88.0, "completions/min_terminated_length": 88.0, "epoch": 0.9984, "num_tokens": 1042184222.0, "reward": 1.138842523097992, "reward_std": 0.09942503273487091, "rewards/accuracy_reward": 0.478759765625, "rewards/brier_reward": 0.7997052073478699, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.999267578125, "rewards/mean_confidence_reward": 0.4984985440969467, "rewards/smce": -0.048381347209215164, "step": 312, "total_flos": 0.0, "train_loss": 0.004145986298401392, "train_runtime": 42424.6813, "train_samples_per_second": 0.471, "train_steps_per_second": 0.007 } ], "logging_steps": 5, "max_steps": 312, "num_input_tokens_seen": 1042184222, "num_train_epochs": 1, "save_steps": 60, "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": 4, "trial_name": null, "trial_params": null }