234 lines
8.2 KiB
JSON
234 lines
8.2 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 5.0,
|
||
|
|
"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": 359.0,
|
||
|
|
"epoch": 0.5714285714285714,
|
||
|
|
"grad_norm": 3.656216859817505,
|
||
|
|
"kl": 0.0,
|
||
|
|
"learning_rate": 5e-07,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 1.5537184616550803,
|
||
|
|
"reward_std": 0.8284978335723281,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.18393750488758087,
|
||
|
|
"rewards/consensus_reward_func": 0.125,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.1875,
|
||
|
|
"rewards/question_recreation_reward_func": 0.5922809606418014,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.46500000171363354,
|
||
|
|
"step": 2
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 219.0,
|
||
|
|
"epoch": 1.0,
|
||
|
|
"grad_norm": 3.6821982860565186,
|
||
|
|
"kl": 0.001071901002433151,
|
||
|
|
"learning_rate": 4.864543104251586e-07,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 1.7033713484803836,
|
||
|
|
"reward_std": 0.6021518250927329,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.0,
|
||
|
|
"rewards/consensus_reward_func": 0.0,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.5,
|
||
|
|
"rewards/question_recreation_reward_func": 0.47470465178291005,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.7286666855216026,
|
||
|
|
"step": 4
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 343.25,
|
||
|
|
"epoch": 1.5714285714285714,
|
||
|
|
"grad_norm": 5.157704830169678,
|
||
|
|
"kl": 0.0013515729988284875,
|
||
|
|
"learning_rate": 4.472851273490984e-07,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 1.7683933172374964,
|
||
|
|
"reward_std": 0.9875846401555464,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.10987499728798866,
|
||
|
|
"rewards/consensus_reward_func": 0.0625,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.5,
|
||
|
|
"rewards/question_recreation_reward_func": 0.5343933093827218,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.015625,
|
||
|
|
"rewards/xmlcount_reward_func": 0.5460000047460198,
|
||
|
|
"step": 6
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 322.4166666666667,
|
||
|
|
"epoch": 2.0,
|
||
|
|
"grad_norm": 4.876896858215332,
|
||
|
|
"kl": 0.0025320505228592083,
|
||
|
|
"learning_rate": 3.867370395306068e-07,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 1.7122117380301158,
|
||
|
|
"reward_std": 0.9681948355088631,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.3100000023841858,
|
||
|
|
"rewards/consensus_reward_func": 0.16666666666666666,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.08333333333333333,
|
||
|
|
"rewards/question_recreation_reward_func": 0.6265450178955992,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.5256666677693526,
|
||
|
|
"step": 8
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 388.84375,
|
||
|
|
"epoch": 2.571428571428571,
|
||
|
|
"grad_norm": 4.167433261871338,
|
||
|
|
"kl": 0.0024263455197797157,
|
||
|
|
"learning_rate": 3.1137137178519977e-07,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 1.7153241410851479,
|
||
|
|
"reward_std": 1.1260282609146088,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.21531250327825546,
|
||
|
|
"rewards/consensus_reward_func": 0.1875,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.1875,
|
||
|
|
"rewards/question_recreation_reward_func": 0.55135537404567,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.5736562591046095,
|
||
|
|
"step": 10
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 232.54166666666666,
|
||
|
|
"epoch": 3.0,
|
||
|
|
"grad_norm": 5.813471794128418,
|
||
|
|
"kl": 0.0031861200113780797,
|
||
|
|
"learning_rate": 2.2935516363191693e-07,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 2.5348341912031174,
|
||
|
|
"reward_std": 0.9292004912470778,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.4763333424925804,
|
||
|
|
"rewards/consensus_reward_func": 0.4166666666666667,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.5,
|
||
|
|
"rewards/question_recreation_reward_func": 0.5478341517349085,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.5939999843637148,
|
||
|
|
"step": 12
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 246.21875,
|
||
|
|
"epoch": 3.571428571428571,
|
||
|
|
"grad_norm": 4.882403373718262,
|
||
|
|
"kl": 0.0033746041226550005,
|
||
|
|
"learning_rate": 1.4957614383675767e-07,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 2.103952983394265,
|
||
|
|
"reward_std": 0.985747697763145,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.17518749833106995,
|
||
|
|
"rewards/consensus_reward_func": 0.3125,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.5,
|
||
|
|
"rewards/question_recreation_reward_func": 0.49595295696053654,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.6203125044703484,
|
||
|
|
"step": 14
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 319.5,
|
||
|
|
"epoch": 4.0,
|
||
|
|
"grad_norm": 3.368023633956909,
|
||
|
|
"kl": 0.003793379970981429,
|
||
|
|
"learning_rate": 8.067960709356478e-08,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 1.2786844770113628,
|
||
|
|
"reward_std": 0.657740268856287,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.0,
|
||
|
|
"rewards/consensus_reward_func": 0.0,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.08333333333333333,
|
||
|
|
"rewards/question_recreation_reward_func": 0.46718447857225937,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.7281666770577431,
|
||
|
|
"step": 16
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 305.71875,
|
||
|
|
"epoch": 4.571428571428571,
|
||
|
|
"grad_norm": 6.600229263305664,
|
||
|
|
"kl": 0.0038686709813191555,
|
||
|
|
"learning_rate": 3.013156219837776e-08,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 2.2076907185837626,
|
||
|
|
"reward_std": 0.9895719490014017,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.5291250087320805,
|
||
|
|
"rewards/consensus_reward_func": 0.3125,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.1875,
|
||
|
|
"rewards/question_recreation_reward_func": 0.49190944456495345,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.6866562394425273,
|
||
|
|
"step": 18
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"completion_length": 339.4166666666667,
|
||
|
|
"epoch": 5.0,
|
||
|
|
"grad_norm": 3.489264726638794,
|
||
|
|
"kl": 0.004692898287127416,
|
||
|
|
"learning_rate": 3.4096741493194193e-09,
|
||
|
|
"loss": 0.0,
|
||
|
|
"reward": 2.753822982311249,
|
||
|
|
"reward_std": 2.2114004302032604,
|
||
|
|
"rewards/concensus_correctness_reward_func": 0.9687499950329462,
|
||
|
|
"rewards/consensus_reward_func": 0.16666666666666666,
|
||
|
|
"rewards/cumulative_reward_2": 0.0,
|
||
|
|
"rewards/final_correctness_reward_func": 0.25,
|
||
|
|
"rewards/question_recreation_reward_func": 0.637656305761387,
|
||
|
|
"rewards/soft_format_reward_func": 0.0,
|
||
|
|
"rewards/strict_format_reward_func": 0.0,
|
||
|
|
"rewards/xmlcount_reward_func": 0.7307499969998995,
|
||
|
|
"step": 20
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0,
|
||
|
|
"step": 20,
|
||
|
|
"total_flos": 0.0,
|
||
|
|
"train_loss": 2.3085623979568483e-06,
|
||
|
|
"train_runtime": 227.0165,
|
||
|
|
"train_samples_per_second": 1.41,
|
||
|
|
"train_steps_per_second": 0.088
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 2,
|
||
|
|
"max_steps": 20,
|
||
|
|
"num_input_tokens_seen": 0,
|
||
|
|
"num_train_epochs": 7,
|
||
|
|
"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
|
||
|
|
}
|