234 lines
8.1 KiB
JSON
234 lines
8.1 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 2.0,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 20,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"completion_length": 20.84375,
|
||
|
|
"epoch": 0.21052631578947367,
|
||
|
|
"grad_norm": 18.707380294799805,
|
||
|
|
"kl": 0.0,
|
||
|
|
"learning_rate": 5e-07,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 2.036191360792145,
|
||
|
|
"reward_std": 0.3408745886699762,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.9699999988079071,
|
||
|
|
"rewards/consensus_reward_func": 1.0,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.0,
|
||
|
|
"rewards/question_recreation_reward_func": 0.06619135761866346,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.0,
|
||
|
|
"step": 2
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 20.96875,
|
||
|
|
"epoch": 0.42105263157894735,
|
||
|
|
"grad_norm": 10.176990509033203,
|
||
|
|
"kl": 0.0014072673075133935,
|
||
|
|
"learning_rate": 4.864543104251586e-07,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 1.7920612578745931,
|
||
|
|
"reward_std": 0.007781138978316449,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.7351249922066927,
|
||
|
|
"rewards/consensus_reward_func": 1.0,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.0,
|
||
|
|
"rewards/question_recreation_reward_func": 0.05693625888670795,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.0,
|
||
|
|
"step": 4
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 21.15625,
|
||
|
|
"epoch": 0.631578947368421,
|
||
|
|
"grad_norm": 37.0001220703125,
|
||
|
|
"kl": 0.014826144542894326,
|
||
|
|
"learning_rate": 4.472851273490984e-07,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 1.9358307735528797,
|
||
|
|
"reward_std": 0.22114181726647075,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.8700000029057264,
|
||
|
|
"rewards/consensus_reward_func": 1.0,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.0,
|
||
|
|
"rewards/question_recreation_reward_func": 0.06583075039088726,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.0,
|
||
|
|
"step": 6
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 18.875,
|
||
|
|
"epoch": 0.8421052631578947,
|
||
|
|
"grad_norm": 16.728694915771484,
|
||
|
|
"kl": 0.007564154569990933,
|
||
|
|
"learning_rate": 3.867370395306068e-07,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 1.5429301662370563,
|
||
|
|
"reward_std": 0.03401432590908371,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.7192500084638596,
|
||
|
|
"rewards/consensus_reward_func": 0.75,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.0,
|
||
|
|
"rewards/question_recreation_reward_func": 0.07368014845997095,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.0,
|
||
|
|
"step": 8
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 15.041666666666666,
|
||
|
|
"epoch": 1.0,
|
||
|
|
"grad_norm": 4.049251556396484,
|
||
|
|
"kl": 0.005367677387160559,
|
||
|
|
"learning_rate": 3.1137137178519977e-07,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 1.909670709981583,
|
||
|
|
"reward_std": 0.0025655655675412468,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.872499997417132,
|
||
|
|
"rewards/consensus_reward_func": 1.0,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.0,
|
||
|
|
"rewards/question_recreation_reward_func": 0.03717067763985445,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.0,
|
||
|
|
"step": 10
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 24.0,
|
||
|
|
"epoch": 1.2105263157894737,
|
||
|
|
"grad_norm": 49.12007522583008,
|
||
|
|
"kl": 0.06435095646884292,
|
||
|
|
"learning_rate": 2.2935516363191693e-07,
|
||
|
|
"loss": 0.0001,
|
||
|
|
"reward": 1.7464143311372027,
|
||
|
|
"reward_std": 0.04995012634026352,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.8133749961853027,
|
||
|
|
"rewards/consensus_reward_func": 0.875,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.0,
|
||
|
|
"rewards/question_recreation_reward_func": 0.05803933663992211,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.0,
|
||
|
|
"step": 12
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 18.90625,
|
||
|
|
"epoch": 1.4210526315789473,
|
||
|
|
"grad_norm": 10.88599681854248,
|
||
|
|
"kl": 0.06291647371836007,
|
||
|
|
"learning_rate": 1.4957614383675767e-07,
|
||
|
|
"loss": 0.0001,
|
||
|
|
"reward": 1.973099282477051,
|
||
|
|
"reward_std": 0.5165156495277188,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.7177499998360872,
|
||
|
|
"rewards/consensus_reward_func": 1.1875,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.0,
|
||
|
|
"rewards/question_recreation_reward_func": 0.06784925094689243,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.0,
|
||
|
|
"step": 14
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 20.46875,
|
||
|
|
"epoch": 1.631578947368421,
|
||
|
|
"grad_norm": 17.875568389892578,
|
||
|
|
"kl": 0.0638568983413279,
|
||
|
|
"learning_rate": 8.067960709356478e-08,
|
||
|
|
"loss": 0.0001,
|
||
|
|
"reward": 1.400315276987385,
|
||
|
|
"reward_std": 0.2234831563400803,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.6239375043660402,
|
||
|
|
"rewards/consensus_reward_func": 0.6875,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.0,
|
||
|
|
"rewards/question_recreation_reward_func": 0.08887775178300217,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.0,
|
||
|
|
"step": 16
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 12.15625,
|
||
|
|
"epoch": 1.8421052631578947,
|
||
|
|
"grad_norm": 4.0479865074157715,
|
||
|
|
"kl": 0.00258190295426175,
|
||
|
|
"learning_rate": 3.013156219837776e-08,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 2.75199212890584,
|
||
|
|
"reward_std": 0.0005554038507398218,
|
||
|
|
"rewards/concensus_correctness_reward_func": 1.3408750034868717,
|
||
|
|
"rewards/consensus_reward_func": 1.375,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.0,
|
||
|
|
"rewards/question_recreation_reward_func": 0.03611713461577892,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.0,
|
||
|
|
"step": 18
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 14.458333333333334,
|
||
|
|
"epoch": 2.0,
|
||
|
|
"grad_norm": 4.316440582275391,
|
||
|
|
"kl": 0.028813007520511746,
|
||
|
|
"learning_rate": 3.4096741493194193e-09,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 2.0803211331367493,
|
||
|
|
"reward_std": 0.06626794133141327,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.9498333260416985,
|
||
|
|
"rewards/consensus_reward_func": 1.0833333333333333,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.0,
|
||
|
|
"rewards/question_recreation_reward_func": 0.04715444279524187,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.0,
|
||
|
|
"step": 20
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0,
|
||
|
|
"step": 20,
|
||
|
|
"total_flos": 0.0,
|
||
|
|
"train_loss": 2.4321150897321785e-05,
|
||
|
|
"train_runtime": 5664.4959,
|
||
|
|
"train_samples_per_second": 0.056,
|
||
|
|
"train_steps_per_second": 0.004
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 2,
|
||
|
|
"max_steps": 20,
|
||
|
|
"num_input_tokens_seen": 0,
|
||
|
|
"num_train_epochs": 3,
|
||
|
|
"save_steps": 25,
|
||
|
|
"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": 2,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|