stage2-10240-1920 / trainer_state.json
gowitheflow1998
upload checkpoint
c406b3d
{
"best_metric": 0.7677604385680659,
"best_model_checkpoint": "./checkpoints/clip-stage2-10240/checkpoint-860",
"epoch": 1.001564945226917,
"global_step": 1920,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 6e-06,
"loss": 4.375,
"step": 20
},
{
"epoch": 0.02,
"learning_rate": 1.2e-05,
"loss": 2.2072,
"step": 40
},
{
"epoch": 0.03,
"learning_rate": 1.8e-05,
"loss": 1.7421,
"step": 60
},
{
"epoch": 0.04,
"learning_rate": 2.4e-05,
"loss": 1.5179,
"step": 80
},
{
"epoch": 0.05,
"learning_rate": 3e-05,
"loss": 1.3767,
"step": 100
},
{
"epoch": 0.06,
"learning_rate": 2.9670329670329673e-05,
"loss": 1.2672,
"step": 120
},
{
"epoch": 0.07,
"learning_rate": 2.934065934065934e-05,
"loss": 1.1824,
"step": 140
},
{
"epoch": 0.08,
"learning_rate": 2.9010989010989013e-05,
"loss": 1.1082,
"step": 160
},
{
"epoch": 0.09,
"learning_rate": 2.8681318681318685e-05,
"loss": 1.0627,
"step": 180
},
{
"epoch": 0.1,
"learning_rate": 2.8351648351648353e-05,
"loss": 1.0224,
"step": 200
},
{
"epoch": 0.11,
"learning_rate": 2.802197802197802e-05,
"loss": 0.991,
"step": 220
},
{
"epoch": 0.13,
"learning_rate": 2.7692307692307694e-05,
"loss": 0.9595,
"step": 240
},
{
"epoch": 0.14,
"learning_rate": 2.7362637362637362e-05,
"loss": 0.9364,
"step": 260
},
{
"epoch": 0.15,
"learning_rate": 2.7032967032967034e-05,
"loss": 0.9123,
"step": 280
},
{
"epoch": 0.16,
"learning_rate": 2.6703296703296702e-05,
"loss": 0.8941,
"step": 300
},
{
"epoch": 0.17,
"learning_rate": 2.6373626373626374e-05,
"loss": 0.8776,
"step": 320
},
{
"epoch": 0.18,
"learning_rate": 2.6043956043956046e-05,
"loss": 0.8598,
"step": 340
},
{
"epoch": 0.19,
"learning_rate": 2.5714285714285714e-05,
"loss": 0.8323,
"step": 360
},
{
"epoch": 0.2,
"learning_rate": 2.5384615384615386e-05,
"loss": 0.8301,
"step": 380
},
{
"epoch": 0.21,
"learning_rate": 2.5054945054945058e-05,
"loss": 0.8138,
"step": 400
},
{
"epoch": 0.22,
"learning_rate": 2.4725274725274723e-05,
"loss": 0.8084,
"step": 420
},
{
"epoch": 0.23,
"learning_rate": 2.4395604395604395e-05,
"loss": 0.7964,
"step": 440
},
{
"epoch": 0.24,
"learning_rate": 2.4065934065934067e-05,
"loss": 0.7846,
"step": 460
},
{
"epoch": 0.25,
"learning_rate": 2.3736263736263735e-05,
"loss": 0.7686,
"step": 480
},
{
"epoch": 0.26,
"learning_rate": 2.3406593406593407e-05,
"loss": 0.7574,
"step": 500
},
{
"epoch": 0.27,
"learning_rate": 2.307692307692308e-05,
"loss": 0.7542,
"step": 520
},
{
"epoch": 0.28,
"learning_rate": 2.2747252747252748e-05,
"loss": 0.7474,
"step": 540
},
{
"epoch": 0.29,
"learning_rate": 2.241758241758242e-05,
"loss": 0.7369,
"step": 560
},
{
"epoch": 0.3,
"learning_rate": 2.208791208791209e-05,
"loss": 0.7296,
"step": 580
},
{
"epoch": 0.31,
"learning_rate": 2.175824175824176e-05,
"loss": 0.7167,
"step": 600
},
{
"epoch": 0.32,
"learning_rate": 2.1428571428571428e-05,
"loss": 0.7155,
"step": 620
},
{
"epoch": 0.33,
"learning_rate": 2.10989010989011e-05,
"loss": 0.7052,
"step": 640
},
{
"epoch": 0.34,
"learning_rate": 2.076923076923077e-05,
"loss": 0.6997,
"step": 660
},
{
"epoch": 0.35,
"learning_rate": 2.043956043956044e-05,
"loss": 0.6944,
"step": 680
},
{
"epoch": 0.37,
"learning_rate": 2.010989010989011e-05,
"loss": 0.6896,
"step": 700
},
{
"epoch": 0.38,
"learning_rate": 1.978021978021978e-05,
"loss": 0.6802,
"step": 720
},
{
"epoch": 0.39,
"learning_rate": 1.9450549450549452e-05,
"loss": 0.6777,
"step": 740
},
{
"epoch": 0.4,
"learning_rate": 1.912087912087912e-05,
"loss": 0.6707,
"step": 760
},
{
"epoch": 0.41,
"learning_rate": 1.8791208791208793e-05,
"loss": 0.6679,
"step": 780
},
{
"epoch": 0.42,
"learning_rate": 1.8461538461538465e-05,
"loss": 0.6575,
"step": 800
},
{
"epoch": 0.43,
"learning_rate": 1.813186813186813e-05,
"loss": 0.6589,
"step": 820
},
{
"epoch": 0.44,
"learning_rate": 1.78021978021978e-05,
"loss": 0.6514,
"step": 840
},
{
"epoch": 0.45,
"learning_rate": 1.7472527472527473e-05,
"loss": 0.6494,
"step": 860
},
{
"epoch": 0.46,
"learning_rate": 1.7142857142857142e-05,
"loss": 0.6441,
"step": 880
},
{
"epoch": 0.47,
"learning_rate": 1.6813186813186814e-05,
"loss": 0.6364,
"step": 900
},
{
"epoch": 0.48,
"learning_rate": 1.6483516483516486e-05,
"loss": 0.6335,
"step": 920
},
{
"epoch": 0.49,
"learning_rate": 1.6153846153846154e-05,
"loss": 0.63,
"step": 940
},
{
"epoch": 0.5,
"learning_rate": 1.5824175824175826e-05,
"loss": 0.6278,
"step": 960
},
{
"epoch": 0.51,
"learning_rate": 1.5494505494505498e-05,
"loss": 0.6214,
"step": 980
},
{
"epoch": 0.52,
"learning_rate": 1.5164835164835164e-05,
"loss": 0.6174,
"step": 1000
},
{
"epoch": 0.53,
"learning_rate": 1.4835164835164836e-05,
"loss": 0.6184,
"step": 1020
},
{
"epoch": 0.54,
"learning_rate": 1.4505494505494506e-05,
"loss": 0.6128,
"step": 1040
},
{
"epoch": 0.55,
"learning_rate": 1.4175824175824177e-05,
"loss": 0.6146,
"step": 1060
},
{
"epoch": 0.56,
"learning_rate": 1.3846153846153847e-05,
"loss": 0.6073,
"step": 1080
},
{
"epoch": 0.57,
"learning_rate": 1.3516483516483517e-05,
"loss": 0.6037,
"step": 1100
},
{
"epoch": 0.58,
"learning_rate": 1.3186813186813187e-05,
"loss": 0.594,
"step": 1120
},
{
"epoch": 0.59,
"learning_rate": 1.2857142857142857e-05,
"loss": 0.597,
"step": 1140
},
{
"epoch": 0.61,
"learning_rate": 1.2527472527472529e-05,
"loss": 0.5906,
"step": 1160
},
{
"epoch": 0.62,
"learning_rate": 1.2197802197802198e-05,
"loss": 0.5941,
"step": 1180
},
{
"epoch": 0.63,
"learning_rate": 1.1868131868131868e-05,
"loss": 0.591,
"step": 1200
},
{
"epoch": 0.64,
"learning_rate": 1.153846153846154e-05,
"loss": 0.5878,
"step": 1220
},
{
"epoch": 0.65,
"learning_rate": 1.120879120879121e-05,
"loss": 0.5876,
"step": 1240
},
{
"epoch": 0.66,
"learning_rate": 1.087912087912088e-05,
"loss": 0.5778,
"step": 1260
},
{
"epoch": 0.67,
"learning_rate": 1.054945054945055e-05,
"loss": 0.5835,
"step": 1280
},
{
"epoch": 0.68,
"learning_rate": 1.021978021978022e-05,
"loss": 0.5781,
"step": 1300
},
{
"epoch": 0.69,
"learning_rate": 9.89010989010989e-06,
"loss": 0.5805,
"step": 1320
},
{
"epoch": 0.7,
"learning_rate": 9.56043956043956e-06,
"loss": 0.5718,
"step": 1340
},
{
"epoch": 0.71,
"learning_rate": 9.230769230769232e-06,
"loss": 0.5721,
"step": 1360
},
{
"epoch": 0.72,
"learning_rate": 8.9010989010989e-06,
"loss": 0.5767,
"step": 1380
},
{
"epoch": 0.73,
"learning_rate": 8.571428571428571e-06,
"loss": 0.5742,
"step": 1400
},
{
"epoch": 0.74,
"learning_rate": 8.241758241758243e-06,
"loss": 0.5662,
"step": 1420
},
{
"epoch": 0.75,
"learning_rate": 7.912087912087913e-06,
"loss": 0.5658,
"step": 1440
},
{
"epoch": 0.76,
"learning_rate": 7.582417582417582e-06,
"loss": 0.5664,
"step": 1460
},
{
"epoch": 0.77,
"learning_rate": 7.252747252747253e-06,
"loss": 0.5637,
"step": 1480
},
{
"epoch": 0.78,
"learning_rate": 6.923076923076923e-06,
"loss": 0.5571,
"step": 1500
},
{
"epoch": 0.79,
"learning_rate": 6.5934065934065935e-06,
"loss": 0.5599,
"step": 1520
},
{
"epoch": 0.8,
"learning_rate": 6.2637362637362645e-06,
"loss": 0.5551,
"step": 1540
},
{
"epoch": 0.81,
"learning_rate": 5.934065934065934e-06,
"loss": 0.554,
"step": 1560
},
{
"epoch": 0.82,
"learning_rate": 5.604395604395605e-06,
"loss": 0.5573,
"step": 1580
},
{
"epoch": 0.83,
"learning_rate": 5.274725274725275e-06,
"loss": 0.5571,
"step": 1600
},
{
"epoch": 0.85,
"learning_rate": 4.945054945054945e-06,
"loss": 0.5491,
"step": 1620
},
{
"epoch": 0.86,
"learning_rate": 4.615384615384616e-06,
"loss": 0.5537,
"step": 1640
},
{
"epoch": 0.87,
"learning_rate": 4.2857142857142855e-06,
"loss": 0.5491,
"step": 1660
},
{
"epoch": 0.88,
"learning_rate": 3.9560439560439565e-06,
"loss": 0.5474,
"step": 1680
},
{
"epoch": 0.89,
"learning_rate": 3.6263736263736266e-06,
"loss": 0.552,
"step": 1700
},
{
"epoch": 0.9,
"learning_rate": 3.2967032967032968e-06,
"loss": 0.5483,
"step": 1720
},
{
"epoch": 0.91,
"learning_rate": 2.967032967032967e-06,
"loss": 0.5412,
"step": 1740
},
{
"epoch": 0.92,
"learning_rate": 2.6373626373626375e-06,
"loss": 0.543,
"step": 1760
},
{
"epoch": 0.93,
"learning_rate": 2.307692307692308e-06,
"loss": 0.5431,
"step": 1780
},
{
"epoch": 0.94,
"learning_rate": 1.9780219780219782e-06,
"loss": 0.5385,
"step": 1800
},
{
"epoch": 0.95,
"learning_rate": 1.6483516483516484e-06,
"loss": 0.5432,
"step": 1820
},
{
"epoch": 0.96,
"learning_rate": 1.3186813186813187e-06,
"loss": 0.5424,
"step": 1840
},
{
"epoch": 0.97,
"learning_rate": 9.890109890109891e-07,
"loss": 0.5419,
"step": 1860
},
{
"epoch": 0.98,
"learning_rate": 6.593406593406594e-07,
"loss": 0.5388,
"step": 1880
},
{
"epoch": 0.99,
"learning_rate": 3.296703296703297e-07,
"loss": 0.5351,
"step": 1900
},
{
"epoch": 1.0,
"learning_rate": 0.0,
"loss": 0.5365,
"step": 1920
}
],
"max_steps": 1920,
"num_train_epochs": 2,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}