GaetanMichelet's picture
Model save
585be15 verified
{
"best_metric": 0.875532865524292,
"best_model_checkpoint": "data/Llama-31-8B_task-1_180-samples_config-2_auto/checkpoint-42",
"epoch": 12.0,
"eval_steps": 500,
"global_step": 102,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.11764705882352941,
"grad_norm": 0.7379887700080872,
"learning_rate": 2.5e-06,
"loss": 1.1383,
"step": 1
},
{
"epoch": 0.23529411764705882,
"grad_norm": 0.8795378804206848,
"learning_rate": 5e-06,
"loss": 1.1326,
"step": 2
},
{
"epoch": 0.47058823529411764,
"grad_norm": 0.7902247905731201,
"learning_rate": 1e-05,
"loss": 1.1866,
"step": 4
},
{
"epoch": 0.7058823529411765,
"grad_norm": 0.8350782990455627,
"learning_rate": 1.5e-05,
"loss": 1.1709,
"step": 6
},
{
"epoch": 0.9411764705882353,
"grad_norm": 0.7659512758255005,
"learning_rate": 2e-05,
"loss": 1.178,
"step": 8
},
{
"epoch": 0.9411764705882353,
"eval_loss": 1.1843764781951904,
"eval_runtime": 14.8627,
"eval_samples_per_second": 2.422,
"eval_steps_per_second": 2.422,
"step": 8
},
{
"epoch": 1.1764705882352942,
"grad_norm": 0.7236676812171936,
"learning_rate": 2.5e-05,
"loss": 1.1113,
"step": 10
},
{
"epoch": 1.4117647058823528,
"grad_norm": 0.6123073101043701,
"learning_rate": 3e-05,
"loss": 1.0434,
"step": 12
},
{
"epoch": 1.6470588235294117,
"grad_norm": 0.48625627160072327,
"learning_rate": 3.5e-05,
"loss": 1.0258,
"step": 14
},
{
"epoch": 1.8823529411764706,
"grad_norm": 0.4125094413757324,
"learning_rate": 4e-05,
"loss": 0.9678,
"step": 16
},
{
"epoch": 2.0,
"eval_loss": 1.0291413068771362,
"eval_runtime": 14.8555,
"eval_samples_per_second": 2.423,
"eval_steps_per_second": 2.423,
"step": 17
},
{
"epoch": 2.1176470588235294,
"grad_norm": 0.5882901549339294,
"learning_rate": 4.5e-05,
"loss": 0.9247,
"step": 18
},
{
"epoch": 2.3529411764705883,
"grad_norm": 0.5574043393135071,
"learning_rate": 5e-05,
"loss": 0.9418,
"step": 20
},
{
"epoch": 2.588235294117647,
"grad_norm": 0.5374100208282471,
"learning_rate": 5.500000000000001e-05,
"loss": 0.9337,
"step": 22
},
{
"epoch": 2.8235294117647056,
"grad_norm": 0.4274942874908447,
"learning_rate": 6e-05,
"loss": 0.9547,
"step": 24
},
{
"epoch": 2.9411764705882355,
"eval_loss": 0.9494705200195312,
"eval_runtime": 14.8488,
"eval_samples_per_second": 2.424,
"eval_steps_per_second": 2.424,
"step": 25
},
{
"epoch": 3.0588235294117645,
"grad_norm": 0.3860715627670288,
"learning_rate": 6.500000000000001e-05,
"loss": 0.8978,
"step": 26
},
{
"epoch": 3.2941176470588234,
"grad_norm": 0.36616435647010803,
"learning_rate": 7e-05,
"loss": 0.8437,
"step": 28
},
{
"epoch": 3.5294117647058822,
"grad_norm": 0.4755256474018097,
"learning_rate": 7.500000000000001e-05,
"loss": 0.8743,
"step": 30
},
{
"epoch": 3.764705882352941,
"grad_norm": 0.47232773900032043,
"learning_rate": 8e-05,
"loss": 0.8333,
"step": 32
},
{
"epoch": 4.0,
"grad_norm": 0.4954814016819,
"learning_rate": 8.5e-05,
"loss": 0.8037,
"step": 34
},
{
"epoch": 4.0,
"eval_loss": 0.896953284740448,
"eval_runtime": 14.8521,
"eval_samples_per_second": 2.424,
"eval_steps_per_second": 2.424,
"step": 34
},
{
"epoch": 4.235294117647059,
"grad_norm": 0.4782576262950897,
"learning_rate": 9e-05,
"loss": 0.76,
"step": 36
},
{
"epoch": 4.470588235294118,
"grad_norm": 0.48261916637420654,
"learning_rate": 9.5e-05,
"loss": 0.7239,
"step": 38
},
{
"epoch": 4.705882352941177,
"grad_norm": 0.4759092628955841,
"learning_rate": 0.0001,
"loss": 0.7653,
"step": 40
},
{
"epoch": 4.9411764705882355,
"grad_norm": 0.5005146265029907,
"learning_rate": 9.999238475781957e-05,
"loss": 0.7404,
"step": 42
},
{
"epoch": 4.9411764705882355,
"eval_loss": 0.875532865524292,
"eval_runtime": 14.8455,
"eval_samples_per_second": 2.425,
"eval_steps_per_second": 2.425,
"step": 42
},
{
"epoch": 5.176470588235294,
"grad_norm": 0.5142149329185486,
"learning_rate": 9.99695413509548e-05,
"loss": 0.6768,
"step": 44
},
{
"epoch": 5.411764705882353,
"grad_norm": 0.4862489402294159,
"learning_rate": 9.99314767377287e-05,
"loss": 0.6057,
"step": 46
},
{
"epoch": 5.647058823529412,
"grad_norm": 0.592656672000885,
"learning_rate": 9.987820251299122e-05,
"loss": 0.6084,
"step": 48
},
{
"epoch": 5.882352941176471,
"grad_norm": 0.664348840713501,
"learning_rate": 9.980973490458728e-05,
"loss": 0.6681,
"step": 50
},
{
"epoch": 6.0,
"eval_loss": 0.9058244228363037,
"eval_runtime": 14.8429,
"eval_samples_per_second": 2.425,
"eval_steps_per_second": 2.425,
"step": 51
},
{
"epoch": 6.117647058823529,
"grad_norm": 0.6137334704399109,
"learning_rate": 9.972609476841367e-05,
"loss": 0.5987,
"step": 52
},
{
"epoch": 6.352941176470588,
"grad_norm": 0.6799383163452148,
"learning_rate": 9.962730758206611e-05,
"loss": 0.5314,
"step": 54
},
{
"epoch": 6.588235294117647,
"grad_norm": 0.6993069052696228,
"learning_rate": 9.951340343707852e-05,
"loss": 0.4916,
"step": 56
},
{
"epoch": 6.823529411764706,
"grad_norm": 0.8820928931236267,
"learning_rate": 9.938441702975689e-05,
"loss": 0.4752,
"step": 58
},
{
"epoch": 6.9411764705882355,
"eval_loss": 0.9784876704216003,
"eval_runtime": 14.8528,
"eval_samples_per_second": 2.424,
"eval_steps_per_second": 2.424,
"step": 59
},
{
"epoch": 7.0588235294117645,
"grad_norm": 0.7526403665542603,
"learning_rate": 9.924038765061042e-05,
"loss": 0.4781,
"step": 60
},
{
"epoch": 7.294117647058823,
"grad_norm": 0.8135146498680115,
"learning_rate": 9.908135917238321e-05,
"loss": 0.3983,
"step": 62
},
{
"epoch": 7.529411764705882,
"grad_norm": 0.9539251327514648,
"learning_rate": 9.890738003669029e-05,
"loss": 0.3169,
"step": 64
},
{
"epoch": 7.764705882352941,
"grad_norm": 1.18768310546875,
"learning_rate": 9.871850323926177e-05,
"loss": 0.3389,
"step": 66
},
{
"epoch": 8.0,
"grad_norm": 0.965970516204834,
"learning_rate": 9.851478631379982e-05,
"loss": 0.3663,
"step": 68
},
{
"epoch": 8.0,
"eval_loss": 1.0200661420822144,
"eval_runtime": 14.8448,
"eval_samples_per_second": 2.425,
"eval_steps_per_second": 2.425,
"step": 68
},
{
"epoch": 8.235294117647058,
"grad_norm": 1.0808840990066528,
"learning_rate": 9.829629131445342e-05,
"loss": 0.2348,
"step": 70
},
{
"epoch": 8.470588235294118,
"grad_norm": 0.999593198299408,
"learning_rate": 9.806308479691595e-05,
"loss": 0.2387,
"step": 72
},
{
"epoch": 8.705882352941176,
"grad_norm": 1.381742238998413,
"learning_rate": 9.781523779815179e-05,
"loss": 0.1997,
"step": 74
},
{
"epoch": 8.941176470588236,
"grad_norm": 1.5578360557556152,
"learning_rate": 9.755282581475769e-05,
"loss": 0.2328,
"step": 76
},
{
"epoch": 8.941176470588236,
"eval_loss": 1.250893473625183,
"eval_runtime": 14.8533,
"eval_samples_per_second": 2.424,
"eval_steps_per_second": 2.424,
"step": 76
},
{
"epoch": 9.176470588235293,
"grad_norm": 0.8032568097114563,
"learning_rate": 9.727592877996585e-05,
"loss": 0.1427,
"step": 78
},
{
"epoch": 9.411764705882353,
"grad_norm": 0.8731306195259094,
"learning_rate": 9.698463103929542e-05,
"loss": 0.1289,
"step": 80
},
{
"epoch": 9.647058823529411,
"grad_norm": 1.0448521375656128,
"learning_rate": 9.667902132486009e-05,
"loss": 0.1326,
"step": 82
},
{
"epoch": 9.882352941176471,
"grad_norm": 1.1242917776107788,
"learning_rate": 9.635919272833938e-05,
"loss": 0.1375,
"step": 84
},
{
"epoch": 10.0,
"eval_loss": 1.4119974374771118,
"eval_runtime": 14.8427,
"eval_samples_per_second": 2.425,
"eval_steps_per_second": 2.425,
"step": 85
},
{
"epoch": 10.117647058823529,
"grad_norm": 0.710112988948822,
"learning_rate": 9.602524267262203e-05,
"loss": 0.0971,
"step": 86
},
{
"epoch": 10.352941176470589,
"grad_norm": 0.8764486312866211,
"learning_rate": 9.567727288213005e-05,
"loss": 0.0803,
"step": 88
},
{
"epoch": 10.588235294117647,
"grad_norm": 0.759926438331604,
"learning_rate": 9.53153893518325e-05,
"loss": 0.0828,
"step": 90
},
{
"epoch": 10.823529411764707,
"grad_norm": 1.4702188968658447,
"learning_rate": 9.493970231495835e-05,
"loss": 0.1013,
"step": 92
},
{
"epoch": 10.941176470588236,
"eval_loss": 1.4668859243392944,
"eval_runtime": 14.8429,
"eval_samples_per_second": 2.425,
"eval_steps_per_second": 2.425,
"step": 93
},
{
"epoch": 11.058823529411764,
"grad_norm": 0.6960480213165283,
"learning_rate": 9.45503262094184e-05,
"loss": 0.0845,
"step": 94
},
{
"epoch": 11.294117647058824,
"grad_norm": 0.7549808025360107,
"learning_rate": 9.414737964294636e-05,
"loss": 0.0658,
"step": 96
},
{
"epoch": 11.529411764705882,
"grad_norm": 0.9834457039833069,
"learning_rate": 9.373098535696979e-05,
"loss": 0.0663,
"step": 98
},
{
"epoch": 11.764705882352942,
"grad_norm": 0.7059162855148315,
"learning_rate": 9.330127018922194e-05,
"loss": 0.054,
"step": 100
},
{
"epoch": 12.0,
"grad_norm": 0.7496359944343567,
"learning_rate": 9.285836503510562e-05,
"loss": 0.0523,
"step": 102
},
{
"epoch": 12.0,
"eval_loss": 1.5482221841812134,
"eval_runtime": 14.857,
"eval_samples_per_second": 2.423,
"eval_steps_per_second": 2.423,
"step": 102
},
{
"epoch": 12.0,
"step": 102,
"total_flos": 2.769253681017651e+16,
"train_loss": 0.5588808070649119,
"train_runtime": 1964.6597,
"train_samples_per_second": 3.461,
"train_steps_per_second": 0.204
}
],
"logging_steps": 2,
"max_steps": 400,
"num_input_tokens_seen": 0,
"num_train_epochs": 50,
"save_steps": 25,
"stateful_callbacks": {
"EarlyStoppingCallback": {
"args": {
"early_stopping_patience": 7,
"early_stopping_threshold": 0.0
},
"attributes": {
"early_stopping_patience_counter": 0
}
},
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 2.769253681017651e+16,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}