gpt2_wt103-40m_12-layer / trainer_state.json
Kristijan's picture
Upload 8 files
41e0c10
{
"best_metric": 3.699657678604126,
"best_model_checkpoint": "/scratch/ka2773/project/lm-mem/checkpoints/gpt2_40m_12-768-1024_a_02/checkpoint-31000",
"epoch": 8.481532147742818,
"global_step": 31000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.14,
"learning_rate": 6e-05,
"loss": 6.6803,
"step": 500
},
{
"epoch": 0.14,
"eval_loss": 6.017919540405273,
"eval_runtime": 9.4636,
"eval_samples_per_second": 24.938,
"step": 500
},
{
"epoch": 0.27,
"learning_rate": 6e-05,
"loss": 6.0117,
"step": 1000
},
{
"epoch": 0.27,
"eval_loss": 5.752170085906982,
"eval_runtime": 9.4619,
"eval_samples_per_second": 24.942,
"step": 1000
},
{
"epoch": 0.41,
"learning_rate": 6e-05,
"loss": 5.8304,
"step": 1500
},
{
"epoch": 0.41,
"eval_loss": 5.598978519439697,
"eval_runtime": 9.4601,
"eval_samples_per_second": 24.947,
"step": 1500
},
{
"epoch": 0.55,
"learning_rate": 6e-05,
"loss": 5.6932,
"step": 2000
},
{
"epoch": 0.55,
"eval_loss": 5.4606218338012695,
"eval_runtime": 9.4514,
"eval_samples_per_second": 24.97,
"step": 2000
},
{
"epoch": 0.68,
"learning_rate": 6e-05,
"loss": 5.5582,
"step": 2500
},
{
"epoch": 0.68,
"eval_loss": 5.324854373931885,
"eval_runtime": 9.4555,
"eval_samples_per_second": 24.959,
"step": 2500
},
{
"epoch": 0.82,
"learning_rate": 6e-05,
"loss": 5.4194,
"step": 3000
},
{
"epoch": 0.82,
"eval_loss": 5.187784194946289,
"eval_runtime": 9.4406,
"eval_samples_per_second": 24.998,
"step": 3000
},
{
"epoch": 0.96,
"learning_rate": 6e-05,
"loss": 5.2874,
"step": 3500
},
{
"epoch": 0.96,
"eval_loss": 5.043916702270508,
"eval_runtime": 9.4594,
"eval_samples_per_second": 24.949,
"step": 3500
},
{
"epoch": 1.09,
"learning_rate": 6e-05,
"loss": 5.1502,
"step": 4000
},
{
"epoch": 1.09,
"eval_loss": 4.9315690994262695,
"eval_runtime": 9.458,
"eval_samples_per_second": 24.952,
"step": 4000
},
{
"epoch": 1.23,
"learning_rate": 6e-05,
"loss": 5.0351,
"step": 4500
},
{
"epoch": 1.23,
"eval_loss": 4.828794479370117,
"eval_runtime": 9.4563,
"eval_samples_per_second": 24.957,
"step": 4500
},
{
"epoch": 1.37,
"learning_rate": 6e-05,
"loss": 4.9495,
"step": 5000
},
{
"epoch": 1.37,
"eval_loss": 4.735019683837891,
"eval_runtime": 9.4648,
"eval_samples_per_second": 24.934,
"step": 5000
},
{
"epoch": 1.5,
"learning_rate": 6e-05,
"loss": 4.8715,
"step": 5500
},
{
"epoch": 1.5,
"eval_loss": 4.6624579429626465,
"eval_runtime": 9.4463,
"eval_samples_per_second": 24.983,
"step": 5500
},
{
"epoch": 1.64,
"learning_rate": 6e-05,
"loss": 4.7962,
"step": 6000
},
{
"epoch": 1.64,
"eval_loss": 4.586391925811768,
"eval_runtime": 9.4563,
"eval_samples_per_second": 24.957,
"step": 6000
},
{
"epoch": 1.78,
"learning_rate": 6e-05,
"loss": 4.7329,
"step": 6500
},
{
"epoch": 1.78,
"eval_loss": 4.520113468170166,
"eval_runtime": 9.4613,
"eval_samples_per_second": 24.944,
"step": 6500
},
{
"epoch": 1.92,
"learning_rate": 6e-05,
"loss": 4.6692,
"step": 7000
},
{
"epoch": 1.92,
"eval_loss": 4.469329357147217,
"eval_runtime": 9.4531,
"eval_samples_per_second": 24.965,
"step": 7000
},
{
"epoch": 2.05,
"learning_rate": 6e-05,
"loss": 4.603,
"step": 7500
},
{
"epoch": 2.05,
"eval_loss": 4.408419132232666,
"eval_runtime": 9.4573,
"eval_samples_per_second": 24.954,
"step": 7500
},
{
"epoch": 2.19,
"learning_rate": 6e-05,
"loss": 4.5274,
"step": 8000
},
{
"epoch": 2.19,
"eval_loss": 4.358486175537109,
"eval_runtime": 9.4556,
"eval_samples_per_second": 24.959,
"step": 8000
},
{
"epoch": 2.33,
"learning_rate": 6e-05,
"loss": 4.482,
"step": 8500
},
{
"epoch": 2.33,
"eval_loss": 4.316709041595459,
"eval_runtime": 9.4572,
"eval_samples_per_second": 24.954,
"step": 8500
},
{
"epoch": 2.46,
"learning_rate": 6e-05,
"loss": 4.4431,
"step": 9000
},
{
"epoch": 2.46,
"eval_loss": 4.261863708496094,
"eval_runtime": 9.4644,
"eval_samples_per_second": 24.935,
"step": 9000
},
{
"epoch": 2.6,
"learning_rate": 6e-05,
"loss": 4.4003,
"step": 9500
},
{
"epoch": 2.6,
"eval_loss": 4.2196125984191895,
"eval_runtime": 9.4647,
"eval_samples_per_second": 24.935,
"step": 9500
},
{
"epoch": 2.74,
"learning_rate": 6e-05,
"loss": 4.3602,
"step": 10000
},
{
"epoch": 2.74,
"eval_loss": 4.1680402755737305,
"eval_runtime": 9.4399,
"eval_samples_per_second": 25.0,
"step": 10000
},
{
"epoch": 2.87,
"learning_rate": 6e-05,
"loss": 4.3072,
"step": 10500
},
{
"epoch": 2.87,
"eval_loss": 4.134276390075684,
"eval_runtime": 9.4305,
"eval_samples_per_second": 25.025,
"step": 10500
},
{
"epoch": 3.01,
"learning_rate": 6e-05,
"loss": 4.2849,
"step": 11000
},
{
"epoch": 3.01,
"eval_loss": 4.093378067016602,
"eval_runtime": 9.4676,
"eval_samples_per_second": 24.927,
"step": 11000
},
{
"epoch": 3.15,
"learning_rate": 6e-05,
"loss": 4.2092,
"step": 11500
},
{
"epoch": 3.15,
"eval_loss": 4.067226886749268,
"eval_runtime": 9.4548,
"eval_samples_per_second": 24.961,
"step": 11500
},
{
"epoch": 3.28,
"learning_rate": 6e-05,
"loss": 4.1923,
"step": 12000
},
{
"epoch": 3.28,
"eval_loss": 4.044227600097656,
"eval_runtime": 9.4733,
"eval_samples_per_second": 24.912,
"step": 12000
},
{
"epoch": 3.42,
"learning_rate": 6e-05,
"loss": 4.1774,
"step": 12500
},
{
"epoch": 3.42,
"eval_loss": 4.027223587036133,
"eval_runtime": 9.4672,
"eval_samples_per_second": 24.928,
"step": 12500
},
{
"epoch": 3.56,
"learning_rate": 6e-05,
"loss": 4.1547,
"step": 13000
},
{
"epoch": 3.56,
"eval_loss": 4.007537841796875,
"eval_runtime": 9.4577,
"eval_samples_per_second": 24.953,
"step": 13000
},
{
"epoch": 3.69,
"learning_rate": 6e-05,
"loss": 4.141,
"step": 13500
},
{
"epoch": 3.69,
"eval_loss": 3.9847309589385986,
"eval_runtime": 9.464,
"eval_samples_per_second": 24.937,
"step": 13500
},
{
"epoch": 3.83,
"learning_rate": 6e-05,
"loss": 4.1375,
"step": 14000
},
{
"epoch": 3.83,
"eval_loss": 3.9671761989593506,
"eval_runtime": 9.4732,
"eval_samples_per_second": 24.912,
"step": 14000
},
{
"epoch": 3.97,
"learning_rate": 6e-05,
"loss": 4.1064,
"step": 14500
},
{
"epoch": 3.97,
"eval_loss": 3.9399986267089844,
"eval_runtime": 9.4647,
"eval_samples_per_second": 24.935,
"step": 14500
},
{
"epoch": 4.1,
"learning_rate": 6e-05,
"loss": 4.0485,
"step": 15000
},
{
"epoch": 4.1,
"eval_loss": 3.932234525680542,
"eval_runtime": 9.4672,
"eval_samples_per_second": 24.928,
"step": 15000
},
{
"epoch": 4.24,
"learning_rate": 6e-05,
"loss": 4.0335,
"step": 15500
},
{
"epoch": 4.24,
"eval_loss": 3.9215309619903564,
"eval_runtime": 9.4597,
"eval_samples_per_second": 24.948,
"step": 15500
},
{
"epoch": 4.38,
"learning_rate": 6e-05,
"loss": 4.0358,
"step": 16000
},
{
"epoch": 4.38,
"eval_loss": 3.898069381713867,
"eval_runtime": 9.4635,
"eval_samples_per_second": 24.938,
"step": 16000
},
{
"epoch": 4.51,
"learning_rate": 6e-05,
"loss": 4.0187,
"step": 16500
},
{
"epoch": 4.51,
"eval_loss": 3.893036127090454,
"eval_runtime": 9.4622,
"eval_samples_per_second": 24.941,
"step": 16500
},
{
"epoch": 4.65,
"learning_rate": 6e-05,
"loss": 4.0188,
"step": 17000
},
{
"epoch": 4.65,
"eval_loss": 3.8782527446746826,
"eval_runtime": 9.459,
"eval_samples_per_second": 24.95,
"step": 17000
},
{
"epoch": 4.79,
"learning_rate": 6e-05,
"loss": 4.0176,
"step": 17500
},
{
"epoch": 4.79,
"eval_loss": 3.8655033111572266,
"eval_runtime": 9.4578,
"eval_samples_per_second": 24.953,
"step": 17500
},
{
"epoch": 4.92,
"learning_rate": 6e-05,
"loss": 4.0025,
"step": 18000
},
{
"epoch": 4.92,
"eval_loss": 3.8568382263183594,
"eval_runtime": 9.4581,
"eval_samples_per_second": 24.952,
"step": 18000
},
{
"epoch": 5.06,
"learning_rate": 6e-05,
"loss": 3.9554,
"step": 18500
},
{
"epoch": 5.06,
"eval_loss": 3.849705696105957,
"eval_runtime": 9.4471,
"eval_samples_per_second": 24.981,
"step": 18500
},
{
"epoch": 5.2,
"learning_rate": 6e-05,
"loss": 3.9252,
"step": 19000
},
{
"epoch": 5.2,
"eval_loss": 3.838930130004883,
"eval_runtime": 9.4605,
"eval_samples_per_second": 24.946,
"step": 19000
},
{
"epoch": 5.34,
"learning_rate": 6e-05,
"loss": 3.9239,
"step": 19500
},
{
"epoch": 5.34,
"eval_loss": 3.8333945274353027,
"eval_runtime": 9.462,
"eval_samples_per_second": 24.942,
"step": 19500
},
{
"epoch": 5.47,
"learning_rate": 6e-05,
"loss": 3.9354,
"step": 20000
},
{
"epoch": 5.47,
"eval_loss": 3.8130805492401123,
"eval_runtime": 9.4596,
"eval_samples_per_second": 24.948,
"step": 20000
},
{
"epoch": 5.61,
"learning_rate": 6e-05,
"loss": 3.9418,
"step": 20500
},
{
"epoch": 5.61,
"eval_loss": 3.8175787925720215,
"eval_runtime": 9.4602,
"eval_samples_per_second": 24.947,
"step": 20500
},
{
"epoch": 5.75,
"learning_rate": 6e-05,
"loss": 3.9291,
"step": 21000
},
{
"epoch": 5.75,
"eval_loss": 3.807772636413574,
"eval_runtime": 9.459,
"eval_samples_per_second": 24.95,
"step": 21000
},
{
"epoch": 5.88,
"learning_rate": 6e-05,
"loss": 3.9309,
"step": 21500
},
{
"epoch": 5.88,
"eval_loss": 3.7906742095947266,
"eval_runtime": 9.4675,
"eval_samples_per_second": 24.927,
"step": 21500
},
{
"epoch": 6.02,
"learning_rate": 6e-05,
"loss": 3.9105,
"step": 22000
},
{
"epoch": 6.02,
"eval_loss": 3.7875397205352783,
"eval_runtime": 9.4606,
"eval_samples_per_second": 24.946,
"step": 22000
},
{
"epoch": 6.16,
"learning_rate": 6e-05,
"loss": 3.8424,
"step": 22500
},
{
"epoch": 6.16,
"eval_loss": 3.782013416290283,
"eval_runtime": 9.4639,
"eval_samples_per_second": 24.937,
"step": 22500
},
{
"epoch": 6.29,
"learning_rate": 6e-05,
"loss": 3.8579,
"step": 23000
},
{
"epoch": 6.29,
"eval_loss": 3.781845808029175,
"eval_runtime": 9.4621,
"eval_samples_per_second": 24.941,
"step": 23000
},
{
"epoch": 6.43,
"learning_rate": 6e-05,
"loss": 3.868,
"step": 23500
},
{
"epoch": 6.43,
"eval_loss": 3.76729679107666,
"eval_runtime": 9.4657,
"eval_samples_per_second": 24.932,
"step": 23500
},
{
"epoch": 6.57,
"learning_rate": 6e-05,
"loss": 3.8631,
"step": 24000
},
{
"epoch": 6.57,
"eval_loss": 3.7602250576019287,
"eval_runtime": 9.4649,
"eval_samples_per_second": 24.934,
"step": 24000
},
{
"epoch": 6.7,
"learning_rate": 6e-05,
"loss": 3.8635,
"step": 24500
},
{
"epoch": 6.7,
"eval_loss": 3.7623238563537598,
"eval_runtime": 9.456,
"eval_samples_per_second": 24.958,
"step": 24500
},
{
"epoch": 6.84,
"learning_rate": 6e-05,
"loss": 3.8632,
"step": 25000
},
{
"epoch": 6.84,
"eval_loss": 3.7607743740081787,
"eval_runtime": 9.4601,
"eval_samples_per_second": 24.947,
"step": 25000
},
{
"epoch": 6.98,
"learning_rate": 6e-05,
"loss": 3.873,
"step": 25500
},
{
"epoch": 6.98,
"eval_loss": 3.749258279800415,
"eval_runtime": 9.4598,
"eval_samples_per_second": 24.948,
"step": 25500
},
{
"epoch": 7.11,
"learning_rate": 6e-05,
"loss": 3.7911,
"step": 26000
},
{
"epoch": 7.11,
"eval_loss": 3.744161367416382,
"eval_runtime": 9.4582,
"eval_samples_per_second": 24.952,
"step": 26000
},
{
"epoch": 7.25,
"learning_rate": 6e-05,
"loss": 3.8039,
"step": 26500
},
{
"epoch": 7.25,
"eval_loss": 3.739082098007202,
"eval_runtime": 9.4581,
"eval_samples_per_second": 24.952,
"step": 26500
},
{
"epoch": 7.39,
"learning_rate": 6e-05,
"loss": 3.7996,
"step": 27000
},
{
"epoch": 7.39,
"eval_loss": 3.738431692123413,
"eval_runtime": 9.4714,
"eval_samples_per_second": 24.917,
"step": 27000
},
{
"epoch": 7.52,
"learning_rate": 6e-05,
"loss": 3.8157,
"step": 27500
},
{
"epoch": 7.52,
"eval_loss": 3.7242279052734375,
"eval_runtime": 9.4626,
"eval_samples_per_second": 24.94,
"step": 27500
},
{
"epoch": 7.66,
"learning_rate": 6e-05,
"loss": 3.8224,
"step": 28000
},
{
"epoch": 7.66,
"eval_loss": 3.7267138957977295,
"eval_runtime": 9.4527,
"eval_samples_per_second": 24.966,
"step": 28000
},
{
"epoch": 7.8,
"learning_rate": 6e-05,
"loss": 3.8246,
"step": 28500
},
{
"epoch": 7.8,
"eval_loss": 3.711819887161255,
"eval_runtime": 9.4622,
"eval_samples_per_second": 24.941,
"step": 28500
},
{
"epoch": 7.93,
"learning_rate": 6e-05,
"loss": 3.8176,
"step": 29000
},
{
"epoch": 7.93,
"eval_loss": 3.707026958465576,
"eval_runtime": 9.4595,
"eval_samples_per_second": 24.948,
"step": 29000
},
{
"epoch": 8.07,
"learning_rate": 6e-05,
"loss": 3.7763,
"step": 29500
},
{
"epoch": 8.07,
"eval_loss": 3.7134788036346436,
"eval_runtime": 9.4603,
"eval_samples_per_second": 24.946,
"step": 29500
},
{
"epoch": 8.21,
"learning_rate": 6e-05,
"loss": 3.7557,
"step": 30000
},
{
"epoch": 8.21,
"eval_loss": 3.7067065238952637,
"eval_runtime": 9.4683,
"eval_samples_per_second": 24.925,
"step": 30000
},
{
"epoch": 8.34,
"learning_rate": 6e-05,
"loss": 3.7662,
"step": 30500
},
{
"epoch": 8.34,
"eval_loss": 3.7023823261260986,
"eval_runtime": 9.4321,
"eval_samples_per_second": 25.021,
"step": 30500
},
{
"epoch": 8.48,
"learning_rate": 6e-05,
"loss": 3.7677,
"step": 31000
},
{
"epoch": 8.48,
"eval_loss": 3.699657678604126,
"eval_runtime": 9.4485,
"eval_samples_per_second": 24.978,
"step": 31000
}
],
"max_steps": 36550,
"num_train_epochs": 10,
"total_flos": 3969647640576000.0,
"trial_name": null,
"trial_params": null
}