llama_3_1_8B_Pak_DPO / trainer_state.json
usmanxia's picture
Upload folder using huggingface_hub
11bfc31 verified
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 100,
"global_step": 54,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.09345794392523364,
"grad_norm": 27.934567781750822,
"learning_rate": 4.995258321842611e-06,
"logits/chosen": -1.3600707054138184,
"logits/rejected": -1.311720848083496,
"logps/chosen": -259.7884216308594,
"logps/rejected": -641.204833984375,
"loss": 0.5367130279541016,
"rewards/accuracies": 0.4906249940395355,
"rewards/chosen": -0.011346553452312946,
"rewards/margins": 2.010589599609375,
"rewards/rejected": -2.0219361782073975,
"step": 5
},
{
"epoch": 0.18691588785046728,
"grad_norm": 33.47686763978394,
"learning_rate": 4.83118057351089e-06,
"logits/chosen": -1.4325159788131714,
"logits/rejected": -1.419487714767456,
"logps/chosen": -282.45062255859375,
"logps/rejected": -1006.796630859375,
"loss": 0.1962599277496338,
"rewards/accuracies": 0.8937499523162842,
"rewards/chosen": -1.5842396020889282,
"rewards/margins": 30.126033782958984,
"rewards/rejected": -31.710275650024414,
"step": 10
},
{
"epoch": 0.2803738317757009,
"grad_norm": 33.03025938367781,
"learning_rate": 4.447701436314176e-06,
"logits/chosen": -1.4119861125946045,
"logits/rejected": -1.4033880233764648,
"logps/chosen": -316.49224853515625,
"logps/rejected": -1806.5711669921875,
"loss": 0.13303208351135254,
"rewards/accuracies": 0.9312499761581421,
"rewards/chosen": -3.9324283599853516,
"rewards/margins": 79.23664093017578,
"rewards/rejected": -83.16907501220703,
"step": 15
},
{
"epoch": 0.37383177570093457,
"grad_norm": 15.588810818394203,
"learning_rate": 3.880912432401265e-06,
"logits/chosen": -1.4167428016662598,
"logits/rejected": -1.4375219345092773,
"logps/chosen": -278.1930236816406,
"logps/rejected": -1543.1934814453125,
"loss": 0.1300720453262329,
"rewards/accuracies": 0.9375,
"rewards/chosen": -3.5365679264068604,
"rewards/margins": 69.07416534423828,
"rewards/rejected": -72.61072540283203,
"step": 20
},
{
"epoch": 0.4672897196261682,
"grad_norm": 23.635861216384747,
"learning_rate": 3.184157475180208e-06,
"logits/chosen": -1.5420770645141602,
"logits/rejected": -1.5806173086166382,
"logps/chosen": -289.8417053222656,
"logps/rejected": -1373.1412353515625,
"loss": 0.09687024354934692,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": -3.474010944366455,
"rewards/margins": 56.75503921508789,
"rewards/rejected": -60.22904968261719,
"step": 25
},
{
"epoch": 0.5607476635514018,
"grad_norm": 27.742154267321354,
"learning_rate": 2.4230123536095746e-06,
"logits/chosen": -1.456063985824585,
"logits/rejected": -1.4878963232040405,
"logps/chosen": -284.5406188964844,
"logps/rejected": -1384.5396728515625,
"loss": 0.18257052898406984,
"rewards/accuracies": 0.9031249284744263,
"rewards/chosen": -3.6529622077941895,
"rewards/margins": 59.20906066894531,
"rewards/rejected": -62.86201858520508,
"step": 30
},
{
"epoch": 0.6542056074766355,
"grad_norm": 19.195024194064924,
"learning_rate": 1.6691130013008514e-06,
"logits/chosen": -1.478035807609558,
"logits/rejected": -1.4903415441513062,
"logps/chosen": -294.15576171875,
"logps/rejected": -1432.6474609375,
"loss": 0.14389536380767823,
"rewards/accuracies": 0.9375,
"rewards/chosen": -3.8707540035247803,
"rewards/margins": 60.74583053588867,
"rewards/rejected": -64.6165771484375,
"step": 35
},
{
"epoch": 0.7476635514018691,
"grad_norm": 36.27877245192158,
"learning_rate": 9.934134090518593e-07,
"logits/chosen": -1.4703152179718018,
"logits/rejected": -1.4725449085235596,
"logps/chosen": -310.2994689941406,
"logps/rejected": -1594.9718017578125,
"loss": 0.1617864489555359,
"rewards/accuracies": 0.9281250238418579,
"rewards/chosen": -4.363249778747559,
"rewards/margins": 70.96905517578125,
"rewards/rejected": -75.33231353759766,
"step": 40
},
{
"epoch": 0.8411214953271028,
"grad_norm": 15.093660609578944,
"learning_rate": 4.5950771910944603e-07,
"logits/chosen": -1.4355796575546265,
"logits/rejected": -1.439159631729126,
"logps/chosen": -304.1833801269531,
"logps/rejected": -1558.446533203125,
"loss": 0.13369102478027345,
"rewards/accuracies": 0.921875,
"rewards/chosen": -3.854429006576538,
"rewards/margins": 70.72245025634766,
"rewards/rejected": -74.5768814086914,
"step": 45
},
{
"epoch": 0.9345794392523364,
"grad_norm": 21.304476284345572,
"learning_rate": 1.1764499893210879e-07,
"logits/chosen": -1.4929989576339722,
"logits/rejected": -1.5164098739624023,
"logps/chosen": -278.53350830078125,
"logps/rejected": -1530.287841796875,
"loss": 0.14561424255371094,
"rewards/accuracies": 0.9312499761581421,
"rewards/chosen": -3.787602663040161,
"rewards/margins": 66.7558822631836,
"rewards/rejected": -70.54348754882812,
"step": 50
},
{
"epoch": 1.0,
"step": 54,
"total_flos": 4121088491520.0,
"train_loss": 0.18359575779349716,
"train_runtime": 542.9915,
"train_samples_per_second": 6.298,
"train_steps_per_second": 0.099
}
],
"logging_steps": 5,
"max_steps": 54,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 100,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 4121088491520.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}