anuragshas's picture
End of training
84fcc84
raw
history blame
12.4 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 60.0,
"global_step": 6660,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.9,
"learning_rate": 7.275e-06,
"loss": 13.5875,
"step": 100
},
{
"epoch": 1.8,
"learning_rate": 1.4775e-05,
"loss": 5.5556,
"step": 200
},
{
"epoch": 2.7,
"learning_rate": 2.2274999999999996e-05,
"loss": 4.0171,
"step": 300
},
{
"epoch": 3.6,
"learning_rate": 2.9775e-05,
"loss": 3.4219,
"step": 400
},
{
"epoch": 3.6,
"eval_loss": 3.3127081394195557,
"eval_runtime": 73.7584,
"eval_samples_per_second": 22.031,
"eval_steps_per_second": 1.383,
"eval_wer": 1.0,
"step": 400
},
{
"epoch": 4.5,
"learning_rate": 3.7275e-05,
"loss": 3.2104,
"step": 500
},
{
"epoch": 5.41,
"learning_rate": 4.4775e-05,
"loss": 3.152,
"step": 600
},
{
"epoch": 6.31,
"learning_rate": 5.227499999999999e-05,
"loss": 3.0987,
"step": 700
},
{
"epoch": 7.21,
"learning_rate": 5.9774999999999996e-05,
"loss": 3.0399,
"step": 800
},
{
"epoch": 7.21,
"eval_loss": 3.0330417156219482,
"eval_runtime": 71.6796,
"eval_samples_per_second": 22.67,
"eval_steps_per_second": 1.423,
"eval_wer": 1.0,
"step": 800
},
{
"epoch": 8.11,
"learning_rate": 6.7275e-05,
"loss": 3.0035,
"step": 900
},
{
"epoch": 9.01,
"learning_rate": 7.477499999999999e-05,
"loss": 2.7063,
"step": 1000
},
{
"epoch": 9.91,
"learning_rate": 7.371466431095406e-05,
"loss": 2.0137,
"step": 1100
},
{
"epoch": 10.81,
"learning_rate": 7.238957597173144e-05,
"loss": 1.5756,
"step": 1200
},
{
"epoch": 10.81,
"eval_loss": 0.6108289361000061,
"eval_runtime": 70.7969,
"eval_samples_per_second": 22.953,
"eval_steps_per_second": 1.441,
"eval_wer": 0.572425678586816,
"step": 1200
},
{
"epoch": 11.71,
"learning_rate": 7.106448763250883e-05,
"loss": 1.3527,
"step": 1300
},
{
"epoch": 12.61,
"learning_rate": 6.973939929328621e-05,
"loss": 1.2067,
"step": 1400
},
{
"epoch": 13.51,
"learning_rate": 6.841431095406359e-05,
"loss": 1.1508,
"step": 1500
},
{
"epoch": 14.41,
"learning_rate": 6.708922261484098e-05,
"loss": 1.0995,
"step": 1600
},
{
"epoch": 14.41,
"eval_loss": 0.30910709500312805,
"eval_runtime": 70.2841,
"eval_samples_per_second": 23.12,
"eval_steps_per_second": 1.451,
"eval_wer": 0.31538130116329166,
"step": 1600
},
{
"epoch": 15.32,
"learning_rate": 6.576413427561838e-05,
"loss": 1.0495,
"step": 1700
},
{
"epoch": 16.22,
"learning_rate": 6.443904593639576e-05,
"loss": 1.0183,
"step": 1800
},
{
"epoch": 17.12,
"learning_rate": 6.311395759717314e-05,
"loss": 0.9867,
"step": 1900
},
{
"epoch": 18.02,
"learning_rate": 6.178886925795053e-05,
"loss": 0.9639,
"step": 2000
},
{
"epoch": 18.02,
"eval_loss": 0.2596471905708313,
"eval_runtime": 70.5333,
"eval_samples_per_second": 23.039,
"eval_steps_per_second": 1.446,
"eval_wer": 0.2841016803102111,
"step": 2000
},
{
"epoch": 18.92,
"learning_rate": 6.046378091872791e-05,
"loss": 0.9383,
"step": 2100
},
{
"epoch": 19.82,
"learning_rate": 5.913869257950529e-05,
"loss": 0.9041,
"step": 2200
},
{
"epoch": 20.72,
"learning_rate": 5.781360424028268e-05,
"loss": 0.8936,
"step": 2300
},
{
"epoch": 21.62,
"learning_rate": 5.648851590106007e-05,
"loss": 0.9032,
"step": 2400
},
{
"epoch": 21.62,
"eval_loss": 0.22702418267726898,
"eval_runtime": 71.1235,
"eval_samples_per_second": 22.848,
"eval_steps_per_second": 1.434,
"eval_wer": 0.2513571736320552,
"step": 2400
},
{
"epoch": 22.52,
"learning_rate": 5.516342756183745e-05,
"loss": 0.8739,
"step": 2500
},
{
"epoch": 23.42,
"learning_rate": 5.383833922261484e-05,
"loss": 0.8602,
"step": 2600
},
{
"epoch": 24.32,
"learning_rate": 5.2513250883392223e-05,
"loss": 0.8483,
"step": 2700
},
{
"epoch": 25.23,
"learning_rate": 5.11881625441696e-05,
"loss": 0.8145,
"step": 2800
},
{
"epoch": 25.23,
"eval_loss": 0.21715673804283142,
"eval_runtime": 70.4062,
"eval_samples_per_second": 23.08,
"eval_steps_per_second": 1.449,
"eval_wer": 0.24834123222748816,
"step": 2800
},
{
"epoch": 26.13,
"learning_rate": 4.986307420494699e-05,
"loss": 0.8245,
"step": 2900
},
{
"epoch": 27.03,
"learning_rate": 4.853798586572438e-05,
"loss": 0.8085,
"step": 3000
},
{
"epoch": 27.93,
"learning_rate": 4.721289752650177e-05,
"loss": 0.8101,
"step": 3100
},
{
"epoch": 28.83,
"learning_rate": 4.588780918727915e-05,
"loss": 0.7845,
"step": 3200
},
{
"epoch": 28.83,
"eval_loss": 0.2083793580532074,
"eval_runtime": 70.1439,
"eval_samples_per_second": 23.167,
"eval_steps_per_second": 1.454,
"eval_wer": 0.23326152520465315,
"step": 3200
},
{
"epoch": 29.73,
"learning_rate": 4.4562720848056537e-05,
"loss": 0.7816,
"step": 3300
},
{
"epoch": 30.63,
"learning_rate": 4.3237632508833916e-05,
"loss": 0.7655,
"step": 3400
},
{
"epoch": 31.53,
"learning_rate": 4.19125441696113e-05,
"loss": 0.762,
"step": 3500
},
{
"epoch": 32.43,
"learning_rate": 4.058745583038869e-05,
"loss": 0.7694,
"step": 3600
},
{
"epoch": 32.43,
"eval_loss": 0.19744575023651123,
"eval_runtime": 70.5594,
"eval_samples_per_second": 23.03,
"eval_steps_per_second": 1.446,
"eval_wer": 0.22343817320120637,
"step": 3600
},
{
"epoch": 33.33,
"learning_rate": 3.926236749116607e-05,
"loss": 0.7517,
"step": 3700
},
{
"epoch": 34.23,
"learning_rate": 3.7937279151943456e-05,
"loss": 0.7401,
"step": 3800
},
{
"epoch": 35.14,
"learning_rate": 3.661219081272084e-05,
"loss": 0.7196,
"step": 3900
},
{
"epoch": 36.04,
"learning_rate": 3.528710247349823e-05,
"loss": 0.7333,
"step": 4000
},
{
"epoch": 36.04,
"eval_loss": 0.20204411447048187,
"eval_runtime": 69.6934,
"eval_samples_per_second": 23.316,
"eval_steps_per_second": 1.464,
"eval_wer": 0.21852649719948297,
"step": 4000
},
{
"epoch": 36.94,
"learning_rate": 3.3962014134275616e-05,
"loss": 0.7251,
"step": 4100
},
{
"epoch": 37.84,
"learning_rate": 3.2636925795053e-05,
"loss": 0.7116,
"step": 4200
},
{
"epoch": 38.74,
"learning_rate": 3.131183745583038e-05,
"loss": 0.7236,
"step": 4300
},
{
"epoch": 39.64,
"learning_rate": 2.998674911660777e-05,
"loss": 0.693,
"step": 4400
},
{
"epoch": 39.64,
"eval_loss": 0.194662407040596,
"eval_runtime": 69.8883,
"eval_samples_per_second": 23.251,
"eval_steps_per_second": 1.459,
"eval_wer": 0.21482119775958639,
"step": 4400
},
{
"epoch": 40.54,
"learning_rate": 2.866166077738516e-05,
"loss": 0.6943,
"step": 4500
},
{
"epoch": 41.44,
"learning_rate": 2.7336572438162543e-05,
"loss": 0.6938,
"step": 4600
},
{
"epoch": 42.34,
"learning_rate": 2.6011484098939926e-05,
"loss": 0.6831,
"step": 4700
},
{
"epoch": 43.24,
"learning_rate": 2.469964664310954e-05,
"loss": 0.6802,
"step": 4800
},
{
"epoch": 43.24,
"eval_loss": 0.19601202011108398,
"eval_runtime": 69.3985,
"eval_samples_per_second": 23.416,
"eval_steps_per_second": 1.47,
"eval_wer": 0.2101680310211116,
"step": 4800
},
{
"epoch": 44.14,
"learning_rate": 2.3374558303886924e-05,
"loss": 0.6756,
"step": 4900
},
{
"epoch": 45.05,
"learning_rate": 2.2049469964664307e-05,
"loss": 0.67,
"step": 5000
},
{
"epoch": 45.95,
"learning_rate": 2.0724381625441694e-05,
"loss": 0.6753,
"step": 5100
},
{
"epoch": 46.85,
"learning_rate": 1.939929328621908e-05,
"loss": 0.667,
"step": 5200
},
{
"epoch": 46.85,
"eval_loss": 0.19041335582733154,
"eval_runtime": 70.1706,
"eval_samples_per_second": 23.158,
"eval_steps_per_second": 1.454,
"eval_wer": 0.2072382593709608,
"step": 5200
},
{
"epoch": 47.75,
"learning_rate": 1.8074204946996464e-05,
"loss": 0.6562,
"step": 5300
},
{
"epoch": 48.65,
"learning_rate": 1.674911660777385e-05,
"loss": 0.666,
"step": 5400
},
{
"epoch": 49.55,
"learning_rate": 1.5424028268551237e-05,
"loss": 0.6572,
"step": 5500
},
{
"epoch": 50.45,
"learning_rate": 1.409893992932862e-05,
"loss": 0.6486,
"step": 5600
},
{
"epoch": 50.45,
"eval_loss": 0.18806982040405273,
"eval_runtime": 70.1099,
"eval_samples_per_second": 23.178,
"eval_steps_per_second": 1.455,
"eval_wer": 0.2009478672985782,
"step": 5600
},
{
"epoch": 51.35,
"learning_rate": 1.2773851590106007e-05,
"loss": 0.6484,
"step": 5700
},
{
"epoch": 52.25,
"learning_rate": 1.1448763250883392e-05,
"loss": 0.6549,
"step": 5800
},
{
"epoch": 53.15,
"learning_rate": 1.0123674911660777e-05,
"loss": 0.6322,
"step": 5900
},
{
"epoch": 54.05,
"learning_rate": 8.811837455830388e-06,
"loss": 0.6339,
"step": 6000
},
{
"epoch": 54.05,
"eval_loss": 0.1877446174621582,
"eval_runtime": 70.1417,
"eval_samples_per_second": 23.167,
"eval_steps_per_second": 1.454,
"eval_wer": 0.1988797931925894,
"step": 6000
},
{
"epoch": 54.95,
"learning_rate": 7.486749116607773e-06,
"loss": 0.6413,
"step": 6100
},
{
"epoch": 55.86,
"learning_rate": 6.161660777385159e-06,
"loss": 0.6392,
"step": 6200
},
{
"epoch": 56.76,
"learning_rate": 4.836572438162544e-06,
"loss": 0.6363,
"step": 6300
},
{
"epoch": 57.66,
"learning_rate": 3.511484098939929e-06,
"loss": 0.6254,
"step": 6400
},
{
"epoch": 57.66,
"eval_loss": 0.18933533132076263,
"eval_runtime": 69.6111,
"eval_samples_per_second": 23.344,
"eval_steps_per_second": 1.465,
"eval_wer": 0.2002585092632486,
"step": 6400
},
{
"epoch": 58.56,
"learning_rate": 2.1863957597173144e-06,
"loss": 0.625,
"step": 6500
},
{
"epoch": 59.46,
"learning_rate": 8.613074204946996e-07,
"loss": 0.6379,
"step": 6600
},
{
"epoch": 60.0,
"step": 6660,
"total_flos": 3.0269424793140363e+19,
"train_loss": 1.3688410950852585,
"train_runtime": 13291.4016,
"train_samples_per_second": 16.016,
"train_steps_per_second": 0.501
}
],
"max_steps": 6660,
"num_train_epochs": 60,
"total_flos": 3.0269424793140363e+19,
"trial_name": null,
"trial_params": null
}