llama_binary_4096_5000_dpo_sft001 / trainer_state.json
unfair221's picture
Upload folder using huggingface_hub
6b02c24 verified
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9984,
"eval_steps": 500,
"global_step": 156,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.064,
"grad_norm": 50.866004943847656,
"learning_rate": 6.249999999999999e-07,
"logits/chosen": -1.7505531311035156,
"logits/rejected": -0.9330509305000305,
"logps/chosen": -326.626953125,
"logps/rejected": -795.3060302734375,
"loss": 0.6799,
"rewards/accuracies": 0.44999998807907104,
"rewards/chosen": 0.003760690800845623,
"rewards/margins": 0.03521683067083359,
"rewards/rejected": -0.03145613521337509,
"step": 10
},
{
"epoch": 0.128,
"grad_norm": 11.6607027053833,
"learning_rate": 9.979871469976195e-07,
"logits/chosen": -1.7663414478302002,
"logits/rejected": -0.9799124598503113,
"logps/chosen": -345.8194580078125,
"logps/rejected": -836.4247436523438,
"loss": 0.3675,
"rewards/accuracies": 0.9468749761581421,
"rewards/chosen": 0.04559936374425888,
"rewards/margins": 1.198445439338684,
"rewards/rejected": -1.1528459787368774,
"step": 20
},
{
"epoch": 0.192,
"grad_norm": 0.01964341662824154,
"learning_rate": 9.755282581475767e-07,
"logits/chosen": -1.9397218227386475,
"logits/rejected": -1.3327372074127197,
"logps/chosen": -347.8189697265625,
"logps/rejected": -946.2703857421875,
"loss": 0.114,
"rewards/accuracies": 0.9812500476837158,
"rewards/chosen": -3.344451904296875,
"rewards/margins": 11.386273384094238,
"rewards/rejected": -14.73072624206543,
"step": 30
},
{
"epoch": 0.256,
"grad_norm": 0.047807998955249786,
"learning_rate": 9.29224396800933e-07,
"logits/chosen": -2.099059820175171,
"logits/rejected": -1.7250522375106812,
"logps/chosen": -447.6947937011719,
"logps/rejected": -1181.8309326171875,
"loss": 0.0208,
"rewards/accuracies": 0.9906249642372131,
"rewards/chosen": -13.066911697387695,
"rewards/margins": 28.916919708251953,
"rewards/rejected": -41.98383331298828,
"step": 40
},
{
"epoch": 0.32,
"grad_norm": 0.028338907286524773,
"learning_rate": 8.613974319136957e-07,
"logits/chosen": -2.122957706451416,
"logits/rejected": -1.8380136489868164,
"logps/chosen": -526.5314331054688,
"logps/rejected": -1396.917724609375,
"loss": 0.0409,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -19.01919174194336,
"rewards/margins": 40.06633758544922,
"rewards/rejected": -59.08552932739258,
"step": 50
},
{
"epoch": 0.384,
"grad_norm": 0.02791423164308071,
"learning_rate": 7.754484907260512e-07,
"logits/chosen": -2.0927040576934814,
"logits/rejected": -1.8271617889404297,
"logps/chosen": -525.5350341796875,
"logps/rejected": -1348.9896240234375,
"loss": 0.0127,
"rewards/accuracies": 0.996874988079071,
"rewards/chosen": -19.48218536376953,
"rewards/margins": 38.847755432128906,
"rewards/rejected": -58.32994079589844,
"step": 60
},
{
"epoch": 0.448,
"grad_norm": 0.032935090363025665,
"learning_rate": 6.756874120406714e-07,
"logits/chosen": -2.068298578262329,
"logits/rejected": -1.7861359119415283,
"logps/chosen": -446.0367431640625,
"logps/rejected": -1329.217041015625,
"loss": 0.0165,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -15.754602432250977,
"rewards/margins": 40.86208724975586,
"rewards/rejected": -56.61668395996094,
"step": 70
},
{
"epoch": 0.512,
"grad_norm": 0.02759699709713459,
"learning_rate": 5.671166329088277e-07,
"logits/chosen": -2.023137331008911,
"logits/rejected": -1.7376649379730225,
"logps/chosen": -480.0372314453125,
"logps/rejected": -1358.5458984375,
"loss": 0.0127,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -16.736797332763672,
"rewards/margins": 40.583927154541016,
"rewards/rejected": -57.32072448730469,
"step": 80
},
{
"epoch": 0.576,
"grad_norm": 0.023939350619912148,
"learning_rate": 4.5518034554828327e-07,
"logits/chosen": -1.9989243745803833,
"logits/rejected": -1.6865986585617065,
"logps/chosen": -483.11419677734375,
"logps/rejected": -1387.4888916015625,
"loss": 0.0312,
"rewards/accuracies": 0.984375,
"rewards/chosen": -16.20440673828125,
"rewards/margins": 40.58293914794922,
"rewards/rejected": -56.7873420715332,
"step": 90
},
{
"epoch": 0.64,
"grad_norm": 0.02108897641301155,
"learning_rate": 3.454915028125263e-07,
"logits/chosen": -1.9607043266296387,
"logits/rejected": -1.6439076662063599,
"logps/chosen": -454.13653564453125,
"logps/rejected": -1346.6138916015625,
"loss": 0.0088,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -14.807940483093262,
"rewards/margins": 40.53756332397461,
"rewards/rejected": -55.34550476074219,
"step": 100
},
{
"epoch": 0.704,
"grad_norm": 0.073819100856781,
"learning_rate": 2.4355036129704696e-07,
"logits/chosen": -1.9478294849395752,
"logits/rejected": -1.6269619464874268,
"logps/chosen": -465.25494384765625,
"logps/rejected": -1353.726318359375,
"loss": 0.0179,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -14.820326805114746,
"rewards/margins": 39.9066047668457,
"rewards/rejected": -54.726932525634766,
"step": 110
},
{
"epoch": 0.768,
"grad_norm": 0.10615050792694092,
"learning_rate": 1.5446867550656767e-07,
"logits/chosen": -1.9225519895553589,
"logits/rejected": -1.6315886974334717,
"logps/chosen": -510.450927734375,
"logps/rejected": -1320.8426513671875,
"loss": 0.0228,
"rewards/accuracies": 0.981249988079071,
"rewards/chosen": -16.662580490112305,
"rewards/margins": 36.9322509765625,
"rewards/rejected": -53.59483337402344,
"step": 120
},
{
"epoch": 0.832,
"grad_norm": 0.09960483759641647,
"learning_rate": 8.271337313934867e-08,
"logits/chosen": -1.9384242296218872,
"logits/rejected": -1.6254316568374634,
"logps/chosen": -486.5965270996094,
"logps/rejected": -1338.5518798828125,
"loss": 0.0137,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -15.351827621459961,
"rewards/margins": 38.287864685058594,
"rewards/rejected": -53.639686584472656,
"step": 130
},
{
"epoch": 0.896,
"grad_norm": 0.018777314573526382,
"learning_rate": 3.188256468013139e-08,
"logits/chosen": -1.9388047456741333,
"logits/rejected": -1.5906115770339966,
"logps/chosen": -425.8097229003906,
"logps/rejected": -1347.3880615234375,
"loss": 0.0098,
"rewards/accuracies": 0.996874988079071,
"rewards/chosen": -12.975400924682617,
"rewards/margins": 40.89888000488281,
"rewards/rejected": -53.8742790222168,
"step": 140
},
{
"epoch": 0.96,
"grad_norm": 0.018243663012981415,
"learning_rate": 4.5251191160326495e-09,
"logits/chosen": -1.9214459657669067,
"logits/rejected": -1.607591152191162,
"logps/chosen": -474.6817321777344,
"logps/rejected": -1351.8935546875,
"loss": 0.0147,
"rewards/accuracies": 0.9937499761581421,
"rewards/chosen": -14.741931915283203,
"rewards/margins": 39.045448303222656,
"rewards/rejected": -53.787376403808594,
"step": 150
}
],
"logging_steps": 10,
"max_steps": 156,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 1.1530176818095063e+18,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}