GaetanMichelet's picture
Model save
22da792 verified
{
"best_metric": 0.9159461259841919,
"best_model_checkpoint": "data/Mistral-7B_task-1_180-samples_config-2_full/checkpoint-51",
"epoch": 12.941176470588236,
"eval_steps": 500,
"global_step": 110,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.11764705882352941,
"grad_norm": 2.5466060638427734,
"learning_rate": 2.5e-06,
"loss": 2.075,
"step": 1
},
{
"epoch": 0.23529411764705882,
"grad_norm": 2.510213851928711,
"learning_rate": 5e-06,
"loss": 2.0283,
"step": 2
},
{
"epoch": 0.47058823529411764,
"grad_norm": 2.568328380584717,
"learning_rate": 1e-05,
"loss": 2.0754,
"step": 4
},
{
"epoch": 0.7058823529411765,
"grad_norm": 2.496351957321167,
"learning_rate": 1.5e-05,
"loss": 2.0418,
"step": 6
},
{
"epoch": 0.9411764705882353,
"grad_norm": 2.0648036003112793,
"learning_rate": 2e-05,
"loss": 1.9695,
"step": 8
},
{
"epoch": 0.9411764705882353,
"eval_loss": 1.8462496995925903,
"eval_runtime": 14.5686,
"eval_samples_per_second": 2.471,
"eval_steps_per_second": 2.471,
"step": 8
},
{
"epoch": 1.1764705882352942,
"grad_norm": 1.5974884033203125,
"learning_rate": 2.5e-05,
"loss": 1.8101,
"step": 10
},
{
"epoch": 1.4117647058823528,
"grad_norm": 1.2964701652526855,
"learning_rate": 3e-05,
"loss": 1.7043,
"step": 12
},
{
"epoch": 1.6470588235294117,
"grad_norm": 1.2164214849472046,
"learning_rate": 3.5e-05,
"loss": 1.6112,
"step": 14
},
{
"epoch": 1.8823529411764706,
"grad_norm": 1.3542946577072144,
"learning_rate": 4e-05,
"loss": 1.4658,
"step": 16
},
{
"epoch": 2.0,
"eval_loss": 1.3228856325149536,
"eval_runtime": 14.5474,
"eval_samples_per_second": 2.475,
"eval_steps_per_second": 2.475,
"step": 17
},
{
"epoch": 2.1176470588235294,
"grad_norm": 1.3138030767440796,
"learning_rate": 4.5e-05,
"loss": 1.3523,
"step": 18
},
{
"epoch": 2.3529411764705883,
"grad_norm": 1.0838539600372314,
"learning_rate": 5e-05,
"loss": 1.2067,
"step": 20
},
{
"epoch": 2.588235294117647,
"grad_norm": 1.0649068355560303,
"learning_rate": 5.500000000000001e-05,
"loss": 1.1172,
"step": 22
},
{
"epoch": 2.8235294117647056,
"grad_norm": 0.7390829920768738,
"learning_rate": 6e-05,
"loss": 1.0873,
"step": 24
},
{
"epoch": 2.9411764705882355,
"eval_loss": 1.0281312465667725,
"eval_runtime": 14.5775,
"eval_samples_per_second": 2.47,
"eval_steps_per_second": 2.47,
"step": 25
},
{
"epoch": 3.0588235294117645,
"grad_norm": 0.6466070413589478,
"learning_rate": 6.500000000000001e-05,
"loss": 0.9773,
"step": 26
},
{
"epoch": 3.2941176470588234,
"grad_norm": 0.6372568607330322,
"learning_rate": 7e-05,
"loss": 1.0028,
"step": 28
},
{
"epoch": 3.5294117647058822,
"grad_norm": 0.6881040334701538,
"learning_rate": 7.500000000000001e-05,
"loss": 0.9714,
"step": 30
},
{
"epoch": 3.764705882352941,
"grad_norm": 0.5336817502975464,
"learning_rate": 8e-05,
"loss": 0.9352,
"step": 32
},
{
"epoch": 4.0,
"grad_norm": 0.5645790100097656,
"learning_rate": 8.5e-05,
"loss": 0.8963,
"step": 34
},
{
"epoch": 4.0,
"eval_loss": 0.9577707648277283,
"eval_runtime": 14.5662,
"eval_samples_per_second": 2.471,
"eval_steps_per_second": 2.471,
"step": 34
},
{
"epoch": 4.235294117647059,
"grad_norm": 0.5516751408576965,
"learning_rate": 9e-05,
"loss": 0.8726,
"step": 36
},
{
"epoch": 4.470588235294118,
"grad_norm": 0.6818626523017883,
"learning_rate": 9.5e-05,
"loss": 0.8671,
"step": 38
},
{
"epoch": 4.705882352941177,
"grad_norm": 0.6951989531517029,
"learning_rate": 0.0001,
"loss": 0.8326,
"step": 40
},
{
"epoch": 4.9411764705882355,
"grad_norm": 0.6834418773651123,
"learning_rate": 9.999238475781957e-05,
"loss": 0.8173,
"step": 42
},
{
"epoch": 4.9411764705882355,
"eval_loss": 0.9242374897003174,
"eval_runtime": 14.5691,
"eval_samples_per_second": 2.471,
"eval_steps_per_second": 2.471,
"step": 42
},
{
"epoch": 5.176470588235294,
"grad_norm": 0.6817136406898499,
"learning_rate": 9.99695413509548e-05,
"loss": 0.7695,
"step": 44
},
{
"epoch": 5.411764705882353,
"grad_norm": 0.6675918698310852,
"learning_rate": 9.99314767377287e-05,
"loss": 0.7399,
"step": 46
},
{
"epoch": 5.647058823529412,
"grad_norm": 0.8630544543266296,
"learning_rate": 9.987820251299122e-05,
"loss": 0.6773,
"step": 48
},
{
"epoch": 5.882352941176471,
"grad_norm": 0.8795169591903687,
"learning_rate": 9.980973490458728e-05,
"loss": 0.7207,
"step": 50
},
{
"epoch": 6.0,
"eval_loss": 0.9159461259841919,
"eval_runtime": 14.5733,
"eval_samples_per_second": 2.47,
"eval_steps_per_second": 2.47,
"step": 51
},
{
"epoch": 6.117647058823529,
"grad_norm": 0.7988505363464355,
"learning_rate": 9.972609476841367e-05,
"loss": 0.6305,
"step": 52
},
{
"epoch": 6.352941176470588,
"grad_norm": 0.8636096715927124,
"learning_rate": 9.962730758206611e-05,
"loss": 0.5937,
"step": 54
},
{
"epoch": 6.588235294117647,
"grad_norm": 1.080081582069397,
"learning_rate": 9.951340343707852e-05,
"loss": 0.5753,
"step": 56
},
{
"epoch": 6.823529411764706,
"grad_norm": 1.0529221296310425,
"learning_rate": 9.938441702975689e-05,
"loss": 0.5468,
"step": 58
},
{
"epoch": 6.9411764705882355,
"eval_loss": 0.9483461380004883,
"eval_runtime": 14.5621,
"eval_samples_per_second": 2.472,
"eval_steps_per_second": 2.472,
"step": 59
},
{
"epoch": 7.0588235294117645,
"grad_norm": 0.9509832262992859,
"learning_rate": 9.924038765061042e-05,
"loss": 0.5346,
"step": 60
},
{
"epoch": 7.294117647058823,
"grad_norm": 0.9746370315551758,
"learning_rate": 9.908135917238321e-05,
"loss": 0.4425,
"step": 62
},
{
"epoch": 7.529411764705882,
"grad_norm": 1.384165644645691,
"learning_rate": 9.890738003669029e-05,
"loss": 0.4108,
"step": 64
},
{
"epoch": 7.764705882352941,
"grad_norm": 1.441620945930481,
"learning_rate": 9.871850323926177e-05,
"loss": 0.3867,
"step": 66
},
{
"epoch": 8.0,
"grad_norm": 1.274566650390625,
"learning_rate": 9.851478631379982e-05,
"loss": 0.4071,
"step": 68
},
{
"epoch": 8.0,
"eval_loss": 1.0524382591247559,
"eval_runtime": 14.5602,
"eval_samples_per_second": 2.472,
"eval_steps_per_second": 2.472,
"step": 68
},
{
"epoch": 8.235294117647058,
"grad_norm": 1.0214779376983643,
"learning_rate": 9.829629131445342e-05,
"loss": 0.3106,
"step": 70
},
{
"epoch": 8.470588235294118,
"grad_norm": 1.312712550163269,
"learning_rate": 9.806308479691595e-05,
"loss": 0.303,
"step": 72
},
{
"epoch": 8.705882352941176,
"grad_norm": 1.6440112590789795,
"learning_rate": 9.781523779815179e-05,
"loss": 0.282,
"step": 74
},
{
"epoch": 8.941176470588236,
"grad_norm": 1.8456093072891235,
"learning_rate": 9.755282581475769e-05,
"loss": 0.2885,
"step": 76
},
{
"epoch": 8.941176470588236,
"eval_loss": 1.212545394897461,
"eval_runtime": 14.5543,
"eval_samples_per_second": 2.473,
"eval_steps_per_second": 2.473,
"step": 76
},
{
"epoch": 9.176470588235293,
"grad_norm": 1.2559385299682617,
"learning_rate": 9.727592877996585e-05,
"loss": 0.2313,
"step": 78
},
{
"epoch": 9.411764705882353,
"grad_norm": 1.2266454696655273,
"learning_rate": 9.698463103929542e-05,
"loss": 0.2054,
"step": 80
},
{
"epoch": 9.647058823529411,
"grad_norm": 1.3700865507125854,
"learning_rate": 9.667902132486009e-05,
"loss": 0.2041,
"step": 82
},
{
"epoch": 9.882352941176471,
"grad_norm": 1.4827758073806763,
"learning_rate": 9.635919272833938e-05,
"loss": 0.2064,
"step": 84
},
{
"epoch": 10.0,
"eval_loss": 1.290916919708252,
"eval_runtime": 14.5516,
"eval_samples_per_second": 2.474,
"eval_steps_per_second": 2.474,
"step": 85
},
{
"epoch": 10.117647058823529,
"grad_norm": 1.2037745714187622,
"learning_rate": 9.602524267262203e-05,
"loss": 0.1821,
"step": 86
},
{
"epoch": 10.352941176470589,
"grad_norm": 1.2108136415481567,
"learning_rate": 9.567727288213005e-05,
"loss": 0.1467,
"step": 88
},
{
"epoch": 10.588235294117647,
"grad_norm": 1.2231992483139038,
"learning_rate": 9.53153893518325e-05,
"loss": 0.147,
"step": 90
},
{
"epoch": 10.823529411764707,
"grad_norm": 1.1290245056152344,
"learning_rate": 9.493970231495835e-05,
"loss": 0.1585,
"step": 92
},
{
"epoch": 10.941176470588236,
"eval_loss": 1.3460688591003418,
"eval_runtime": 14.5514,
"eval_samples_per_second": 2.474,
"eval_steps_per_second": 2.474,
"step": 93
},
{
"epoch": 11.058823529411764,
"grad_norm": 0.832689106464386,
"learning_rate": 9.45503262094184e-05,
"loss": 0.1399,
"step": 94
},
{
"epoch": 11.294117647058824,
"grad_norm": 0.9425649642944336,
"learning_rate": 9.414737964294636e-05,
"loss": 0.1147,
"step": 96
},
{
"epoch": 11.529411764705882,
"grad_norm": 1.262665033340454,
"learning_rate": 9.373098535696979e-05,
"loss": 0.1255,
"step": 98
},
{
"epoch": 11.764705882352942,
"grad_norm": 1.311782717704773,
"learning_rate": 9.330127018922194e-05,
"loss": 0.125,
"step": 100
},
{
"epoch": 12.0,
"grad_norm": 0.9994375109672546,
"learning_rate": 9.285836503510562e-05,
"loss": 0.125,
"step": 102
},
{
"epoch": 12.0,
"eval_loss": 1.3770337104797363,
"eval_runtime": 14.5637,
"eval_samples_per_second": 2.472,
"eval_steps_per_second": 2.472,
"step": 102
},
{
"epoch": 12.235294117647058,
"grad_norm": 0.99504554271698,
"learning_rate": 9.24024048078213e-05,
"loss": 0.1028,
"step": 104
},
{
"epoch": 12.470588235294118,
"grad_norm": 1.4479221105575562,
"learning_rate": 9.193352839727121e-05,
"loss": 0.1057,
"step": 106
},
{
"epoch": 12.705882352941176,
"grad_norm": 1.4204106330871582,
"learning_rate": 9.145187862775209e-05,
"loss": 0.1063,
"step": 108
},
{
"epoch": 12.941176470588236,
"grad_norm": 1.1615417003631592,
"learning_rate": 9.09576022144496e-05,
"loss": 0.1121,
"step": 110
},
{
"epoch": 12.941176470588236,
"eval_loss": 1.4891993999481201,
"eval_runtime": 14.5634,
"eval_samples_per_second": 2.472,
"eval_steps_per_second": 2.472,
"step": 110
},
{
"epoch": 12.941176470588236,
"step": 110,
"total_flos": 2.772769687207936e+16,
"train_loss": 0.7063802681524646,
"train_runtime": 2094.2057,
"train_samples_per_second": 3.247,
"train_steps_per_second": 0.191
}
],
"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": 2.772769687207936e+16,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}