opt-invariants / trainer_state.json
sallywww's picture
commit
f7060f2
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 44.61538461538461,
"global_step": 45,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.62,
"learning_rate": 1e-05,
"loss": 5.5122,
"step": 1
},
{
"epoch": 1.62,
"learning_rate": 2e-05,
"loss": 14.9312,
"step": 2
},
{
"epoch": 2.62,
"learning_rate": 1.9973322836635517e-05,
"loss": 4.8242,
"step": 3
},
{
"epoch": 3.62,
"learning_rate": 1.9893433680751105e-05,
"loss": 5.7161,
"step": 4
},
{
"epoch": 4.62,
"learning_rate": 1.9760758775559275e-05,
"loss": 2.5249,
"step": 5
},
{
"epoch": 5.62,
"learning_rate": 1.957600599908406e-05,
"loss": 2.6135,
"step": 6
},
{
"epoch": 6.62,
"learning_rate": 1.9340161087325483e-05,
"loss": 1.7229,
"step": 7
},
{
"epoch": 7.62,
"learning_rate": 1.905448237493147e-05,
"loss": 1.7059,
"step": 8
},
{
"epoch": 8.62,
"learning_rate": 1.872049408143808e-05,
"loss": 1.4524,
"step": 9
},
{
"epoch": 9.62,
"learning_rate": 1.833997817889878e-05,
"loss": 2.1753,
"step": 10
},
{
"epoch": 10.62,
"learning_rate": 1.7914964884292543e-05,
"loss": 1.9059,
"step": 11
},
{
"epoch": 11.62,
"learning_rate": 1.744772182743782e-05,
"loss": 1.0491,
"step": 12
},
{
"epoch": 12.62,
"learning_rate": 1.6940741952206342e-05,
"loss": 1.1342,
"step": 13
},
{
"epoch": 13.62,
"learning_rate": 1.6396730215588913e-05,
"loss": 1.0765,
"step": 14
},
{
"epoch": 14.62,
"learning_rate": 1.581858915557953e-05,
"loss": 1.0615,
"step": 15
},
{
"epoch": 15.62,
"learning_rate": 1.5209403404879305e-05,
"loss": 0.6614,
"step": 16
},
{
"epoch": 16.62,
"learning_rate": 1.4572423233046386e-05,
"loss": 1.1994,
"step": 17
},
{
"epoch": 17.62,
"learning_rate": 1.391104720490156e-05,
"loss": 0.7368,
"step": 18
},
{
"epoch": 18.62,
"learning_rate": 1.3228804047714462e-05,
"loss": 0.6948,
"step": 19
},
{
"epoch": 19.62,
"learning_rate": 1.2529333823916807e-05,
"loss": 0.4572,
"step": 20
},
{
"epoch": 20.62,
"learning_rate": 1.1816368509794365e-05,
"loss": 0.4539,
"step": 21
},
{
"epoch": 21.62,
"learning_rate": 1.1093712083778748e-05,
"loss": 0.596,
"step": 22
},
{
"epoch": 22.62,
"learning_rate": 1.0365220230576592e-05,
"loss": 0.6658,
"step": 23
},
{
"epoch": 23.62,
"learning_rate": 9.634779769423412e-06,
"loss": 0.449,
"step": 24
},
{
"epoch": 24.62,
"learning_rate": 8.906287916221259e-06,
"loss": 0.416,
"step": 25
},
{
"epoch": 25.62,
"learning_rate": 8.183631490205636e-06,
"loss": 0.7238,
"step": 26
},
{
"epoch": 26.62,
"learning_rate": 7.470666176083193e-06,
"loss": 0.2661,
"step": 27
},
{
"epoch": 27.62,
"learning_rate": 6.771195952285541e-06,
"loss": 0.4413,
"step": 28
},
{
"epoch": 28.62,
"learning_rate": 6.088952795098442e-06,
"loss": 0.2416,
"step": 29
},
{
"epoch": 29.62,
"learning_rate": 5.427576766953615e-06,
"loss": 0.1973,
"step": 30
},
{
"epoch": 30.62,
"learning_rate": 4.790596595120699e-06,
"loss": 0.253,
"step": 31
},
{
"epoch": 31.62,
"learning_rate": 4.181410844420473e-06,
"loss": 0.2261,
"step": 32
},
{
"epoch": 32.62,
"learning_rate": 3.6032697844110896e-06,
"loss": 0.195,
"step": 33
},
{
"epoch": 33.62,
"learning_rate": 3.0592580477936606e-06,
"loss": 0.2191,
"step": 34
},
{
"epoch": 34.62,
"learning_rate": 2.5522781725621814e-06,
"loss": 0.1858,
"step": 35
},
{
"epoch": 35.62,
"learning_rate": 2.08503511570746e-06,
"loss": 0.1376,
"step": 36
},
{
"epoch": 36.62,
"learning_rate": 1.660021821101222e-06,
"loss": 0.1918,
"step": 37
},
{
"epoch": 37.62,
"learning_rate": 1.279505918561923e-06,
"loss": 0.1518,
"step": 38
},
{
"epoch": 38.62,
"learning_rate": 9.455176250685338e-07,
"loss": 0.2115,
"step": 39
},
{
"epoch": 39.62,
"learning_rate": 6.598389126745209e-07,
"loss": 0.1276,
"step": 40
},
{
"epoch": 40.62,
"learning_rate": 4.2399400091594154e-07,
"loss": 0.1432,
"step": 41
},
{
"epoch": 41.62,
"learning_rate": 2.392412244407294e-07,
"loss": 0.1248,
"step": 42
},
{
"epoch": 42.62,
"learning_rate": 1.0656631924889749e-07,
"loss": 0.1649,
"step": 43
},
{
"epoch": 43.62,
"learning_rate": 2.667716336448356e-08,
"loss": 0.1561,
"step": 44
},
{
"epoch": 44.62,
"learning_rate": 0.0,
"loss": 0.1263,
"step": 45
},
{
"epoch": 44.62,
"step": 45,
"total_flos": 2105118038163456.0,
"train_loss": 1.3382400512695312,
"train_runtime": 78.1057,
"train_samples_per_second": 59.343,
"train_steps_per_second": 0.576
}
],
"max_steps": 45,
"num_train_epochs": 45,
"total_flos": 2105118038163456.0,
"trial_name": null,
"trial_params": null
}