c1_science_nod_4s_1k / trainer_state.json
ryanmarten's picture
Upload model
586afd9 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 6.666666666666667,
"eval_steps": 500,
"global_step": 70,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.09523809523809523,
"grad_norm": 5.819805402410181,
"learning_rate": 2.8571428571428573e-06,
"loss": 0.9424,
"step": 1
},
{
"epoch": 0.19047619047619047,
"grad_norm": 6.085616084729907,
"learning_rate": 5.7142857142857145e-06,
"loss": 0.9899,
"step": 2
},
{
"epoch": 0.2857142857142857,
"grad_norm": 5.5923202079310865,
"learning_rate": 8.571428571428571e-06,
"loss": 0.9706,
"step": 3
},
{
"epoch": 0.38095238095238093,
"grad_norm": 2.1897098214413964,
"learning_rate": 1.1428571428571429e-05,
"loss": 0.8771,
"step": 4
},
{
"epoch": 0.47619047619047616,
"grad_norm": 4.429501283468818,
"learning_rate": 1.4285714285714287e-05,
"loss": 0.8997,
"step": 5
},
{
"epoch": 0.5714285714285714,
"grad_norm": 5.269619067480585,
"learning_rate": 1.7142857142857142e-05,
"loss": 0.9664,
"step": 6
},
{
"epoch": 0.6666666666666666,
"grad_norm": 4.899422165128255,
"learning_rate": 2e-05,
"loss": 0.9064,
"step": 7
},
{
"epoch": 0.7619047619047619,
"grad_norm": 3.476175801082942,
"learning_rate": 1.9987569212189224e-05,
"loss": 0.8641,
"step": 8
},
{
"epoch": 0.8571428571428571,
"grad_norm": 3.6780372747132266,
"learning_rate": 1.9950307753654016e-05,
"loss": 0.8402,
"step": 9
},
{
"epoch": 0.9523809523809523,
"grad_norm": 2.493327013437351,
"learning_rate": 1.9888308262251286e-05,
"loss": 0.7979,
"step": 10
},
{
"epoch": 1.0476190476190477,
"grad_norm": 3.6744744349502834,
"learning_rate": 1.9801724878485438e-05,
"loss": 1.2105,
"step": 11
},
{
"epoch": 1.1428571428571428,
"grad_norm": 1.7671764727198294,
"learning_rate": 1.969077286229078e-05,
"loss": 0.749,
"step": 12
},
{
"epoch": 1.2380952380952381,
"grad_norm": 1.4978045175063366,
"learning_rate": 1.955572805786141e-05,
"loss": 0.6853,
"step": 13
},
{
"epoch": 1.3333333333333333,
"grad_norm": 1.3478813403098633,
"learning_rate": 1.9396926207859085e-05,
"loss": 0.747,
"step": 14
},
{
"epoch": 1.4285714285714286,
"grad_norm": 0.9590167626186159,
"learning_rate": 1.921476211870408e-05,
"loss": 0.6744,
"step": 15
},
{
"epoch": 1.5238095238095237,
"grad_norm": 0.9731802549185152,
"learning_rate": 1.900968867902419e-05,
"loss": 0.673,
"step": 16
},
{
"epoch": 1.619047619047619,
"grad_norm": 0.928056147924769,
"learning_rate": 1.8782215733702286e-05,
"loss": 0.7702,
"step": 17
},
{
"epoch": 1.7142857142857144,
"grad_norm": 0.733669073161484,
"learning_rate": 1.8532908816321557e-05,
"loss": 0.6813,
"step": 18
},
{
"epoch": 1.8095238095238095,
"grad_norm": 0.9243076881149141,
"learning_rate": 1.826238774315995e-05,
"loss": 0.6769,
"step": 19
},
{
"epoch": 1.9047619047619047,
"grad_norm": 0.8671655193703672,
"learning_rate": 1.7971325072229227e-05,
"loss": 0.7193,
"step": 20
},
{
"epoch": 2.0,
"grad_norm": 1.0450887389047114,
"learning_rate": 1.766044443118978e-05,
"loss": 0.9601,
"step": 21
},
{
"epoch": 2.0952380952380953,
"grad_norm": 0.8151044835586954,
"learning_rate": 1.7330518718298263e-05,
"loss": 0.6002,
"step": 22
},
{
"epoch": 2.1904761904761907,
"grad_norm": 0.7801083830181565,
"learning_rate": 1.698236818086073e-05,
"loss": 0.5637,
"step": 23
},
{
"epoch": 2.2857142857142856,
"grad_norm": 0.7742257428459897,
"learning_rate": 1.6616858375968596e-05,
"loss": 0.5664,
"step": 24
},
{
"epoch": 2.380952380952381,
"grad_norm": 0.7290723848938037,
"learning_rate": 1.6234898018587336e-05,
"loss": 0.5687,
"step": 25
},
{
"epoch": 2.4761904761904763,
"grad_norm": 0.7365502425368368,
"learning_rate": 1.5837436722347902e-05,
"loss": 0.5409,
"step": 26
},
{
"epoch": 2.571428571428571,
"grad_norm": 0.6308333956116082,
"learning_rate": 1.5425462638657597e-05,
"loss": 0.5332,
"step": 27
},
{
"epoch": 2.6666666666666665,
"grad_norm": 0.772759462141294,
"learning_rate": 1.5000000000000002e-05,
"loss": 0.5531,
"step": 28
},
{
"epoch": 2.761904761904762,
"grad_norm": 0.5994678472921114,
"learning_rate": 1.4562106573531632e-05,
"loss": 0.5501,
"step": 29
},
{
"epoch": 2.857142857142857,
"grad_norm": 0.6994647576330689,
"learning_rate": 1.4112871031306118e-05,
"loss": 0.5463,
"step": 30
},
{
"epoch": 2.9523809523809526,
"grad_norm": 0.8213660779889166,
"learning_rate": 1.3653410243663953e-05,
"loss": 0.5643,
"step": 31
},
{
"epoch": 3.0476190476190474,
"grad_norm": 0.8916175795136774,
"learning_rate": 1.3184866502516846e-05,
"loss": 0.781,
"step": 32
},
{
"epoch": 3.142857142857143,
"grad_norm": 0.6046691363322109,
"learning_rate": 1.2708404681430054e-05,
"loss": 0.455,
"step": 33
},
{
"epoch": 3.238095238095238,
"grad_norm": 0.5024745205067274,
"learning_rate": 1.2225209339563144e-05,
"loss": 0.4365,
"step": 34
},
{
"epoch": 3.3333333333333335,
"grad_norm": 0.6489149270351462,
"learning_rate": 1.1736481776669307e-05,
"loss": 0.4653,
"step": 35
},
{
"epoch": 3.4285714285714284,
"grad_norm": 0.7014997304052129,
"learning_rate": 1.1243437046474854e-05,
"loss": 0.4583,
"step": 36
},
{
"epoch": 3.5238095238095237,
"grad_norm": 0.5055406433345587,
"learning_rate": 1.0747300935864245e-05,
"loss": 0.4212,
"step": 37
},
{
"epoch": 3.619047619047619,
"grad_norm": 0.568362993161034,
"learning_rate": 1.0249306917380731e-05,
"loss": 0.4422,
"step": 38
},
{
"epoch": 3.7142857142857144,
"grad_norm": 0.5445268085375626,
"learning_rate": 9.750693082619274e-06,
"loss": 0.4299,
"step": 39
},
{
"epoch": 3.8095238095238093,
"grad_norm": 0.5495036065138471,
"learning_rate": 9.252699064135759e-06,
"loss": 0.4266,
"step": 40
},
{
"epoch": 3.9047619047619047,
"grad_norm": 0.5251539592257585,
"learning_rate": 8.756562953525151e-06,
"loss": 0.4456,
"step": 41
},
{
"epoch": 4.0,
"grad_norm": 0.7935992073599648,
"learning_rate": 8.263518223330698e-06,
"loss": 0.6293,
"step": 42
},
{
"epoch": 4.095238095238095,
"grad_norm": 0.5242630244061871,
"learning_rate": 7.774790660436857e-06,
"loss": 0.3796,
"step": 43
},
{
"epoch": 4.190476190476191,
"grad_norm": 0.4810932819003828,
"learning_rate": 7.291595318569951e-06,
"loss": 0.3828,
"step": 44
},
{
"epoch": 4.285714285714286,
"grad_norm": 0.5183196152897702,
"learning_rate": 6.815133497483157e-06,
"loss": 0.3653,
"step": 45
},
{
"epoch": 4.380952380952381,
"grad_norm": 0.4963807604319561,
"learning_rate": 6.34658975633605e-06,
"loss": 0.3502,
"step": 46
},
{
"epoch": 4.476190476190476,
"grad_norm": 0.5921404393572599,
"learning_rate": 5.887128968693887e-06,
"loss": 0.3713,
"step": 47
},
{
"epoch": 4.571428571428571,
"grad_norm": 0.5283351908427641,
"learning_rate": 5.43789342646837e-06,
"loss": 0.3663,
"step": 48
},
{
"epoch": 4.666666666666667,
"grad_norm": 0.4698027047945985,
"learning_rate": 5.000000000000003e-06,
"loss": 0.3339,
"step": 49
},
{
"epoch": 4.761904761904762,
"grad_norm": 0.50436633837415,
"learning_rate": 4.5745373613424075e-06,
"loss": 0.3283,
"step": 50
},
{
"epoch": 4.857142857142857,
"grad_norm": 0.4648987567966693,
"learning_rate": 4.162563277652104e-06,
"loss": 0.3595,
"step": 51
},
{
"epoch": 4.9523809523809526,
"grad_norm": 0.44686230781148556,
"learning_rate": 3.7651019814126656e-06,
"loss": 0.3533,
"step": 52
},
{
"epoch": 5.0476190476190474,
"grad_norm": 0.7268213912352287,
"learning_rate": 3.3831416240314085e-06,
"loss": 0.5114,
"step": 53
},
{
"epoch": 5.142857142857143,
"grad_norm": 0.4495286976768143,
"learning_rate": 3.017631819139273e-06,
"loss": 0.3045,
"step": 54
},
{
"epoch": 5.238095238095238,
"grad_norm": 0.44902485684265286,
"learning_rate": 2.669481281701739e-06,
"loss": 0.3205,
"step": 55
},
{
"epoch": 5.333333333333333,
"grad_norm": 0.41841894575195615,
"learning_rate": 2.339555568810221e-06,
"loss": 0.3004,
"step": 56
},
{
"epoch": 5.428571428571429,
"grad_norm": 0.3975111111726599,
"learning_rate": 2.0286749277707783e-06,
"loss": 0.2959,
"step": 57
},
{
"epoch": 5.523809523809524,
"grad_norm": 0.5210539661827408,
"learning_rate": 1.7376122568400533e-06,
"loss": 0.3069,
"step": 58
},
{
"epoch": 5.619047619047619,
"grad_norm": 0.49237993808054326,
"learning_rate": 1.467091183678444e-06,
"loss": 0.3089,
"step": 59
},
{
"epoch": 5.714285714285714,
"grad_norm": 0.465696761569748,
"learning_rate": 1.2177842662977136e-06,
"loss": 0.3308,
"step": 60
},
{
"epoch": 5.809523809523809,
"grad_norm": 0.7224482177425864,
"learning_rate": 9.903113209758098e-07,
"loss": 0.3125,
"step": 61
},
{
"epoch": 5.904761904761905,
"grad_norm": 0.36459721085643404,
"learning_rate": 7.852378812959227e-07,
"loss": 0.2866,
"step": 62
},
{
"epoch": 6.0,
"grad_norm": 0.5992301818821822,
"learning_rate": 6.030737921409169e-07,
"loss": 0.4461,
"step": 63
},
{
"epoch": 6.095238095238095,
"grad_norm": 0.3930680441039287,
"learning_rate": 4.4427194213859216e-07,
"loss": 0.3123,
"step": 64
},
{
"epoch": 6.190476190476191,
"grad_norm": 0.3930849129863722,
"learning_rate": 3.0922713770922155e-07,
"loss": 0.2864,
"step": 65
},
{
"epoch": 6.285714285714286,
"grad_norm": 0.3860341508088273,
"learning_rate": 1.9827512151456175e-07,
"loss": 0.2898,
"step": 66
},
{
"epoch": 6.380952380952381,
"grad_norm": 0.401081252191647,
"learning_rate": 1.1169173774871478e-07,
"loss": 0.2838,
"step": 67
},
{
"epoch": 6.476190476190476,
"grad_norm": 0.39802647328109886,
"learning_rate": 4.9692246345985905e-08,
"loss": 0.294,
"step": 68
},
{
"epoch": 6.571428571428571,
"grad_norm": 0.3777001913339216,
"learning_rate": 1.2430787810776556e-08,
"loss": 0.2669,
"step": 69
},
{
"epoch": 6.666666666666667,
"grad_norm": 0.3777433547291866,
"learning_rate": 0.0,
"loss": 0.2928,
"step": 70
},
{
"epoch": 6.666666666666667,
"step": 70,
"total_flos": 1.0455522678616883e+17,
"train_loss": 0.5417226936135974,
"train_runtime": 2252.2011,
"train_samples_per_second": 3.108,
"train_steps_per_second": 0.031
}
],
"logging_steps": 1,
"max_steps": 70,
"num_input_tokens_seen": 0,
"num_train_epochs": 7,
"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": 1.0455522678616883e+17,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}