sanchit-gandhi's picture
End of training
6611ffe
raw
history blame
36.6 kB
{
"best_metric": 0.8860394021739131,
"best_model_checkpoint": "whisper-small-ft-common-language-id/checkpoint-6246",
"epoch": 10.0,
"global_step": 6940,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.04,
"learning_rate": 3.602305475504323e-07,
"loss": 3.8187,
"step": 25
},
{
"epoch": 0.07,
"learning_rate": 7.204610951008646e-07,
"loss": 3.8248,
"step": 50
},
{
"epoch": 0.11,
"learning_rate": 1.0806916426512968e-06,
"loss": 3.8197,
"step": 75
},
{
"epoch": 0.14,
"learning_rate": 1.4409221902017292e-06,
"loss": 3.8101,
"step": 100
},
{
"epoch": 0.18,
"learning_rate": 1.8011527377521614e-06,
"loss": 3.8011,
"step": 125
},
{
"epoch": 0.22,
"learning_rate": 2.1613832853025936e-06,
"loss": 3.7947,
"step": 150
},
{
"epoch": 0.25,
"learning_rate": 2.521613832853026e-06,
"loss": 3.7571,
"step": 175
},
{
"epoch": 0.29,
"learning_rate": 2.8818443804034585e-06,
"loss": 3.6929,
"step": 200
},
{
"epoch": 0.32,
"learning_rate": 3.242074927953891e-06,
"loss": 3.5792,
"step": 225
},
{
"epoch": 0.36,
"learning_rate": 3.602305475504323e-06,
"loss": 3.3495,
"step": 250
},
{
"epoch": 0.4,
"learning_rate": 3.962536023054755e-06,
"loss": 3.0844,
"step": 275
},
{
"epoch": 0.43,
"learning_rate": 4.322766570605187e-06,
"loss": 2.7972,
"step": 300
},
{
"epoch": 0.47,
"learning_rate": 4.68299711815562e-06,
"loss": 2.5878,
"step": 325
},
{
"epoch": 0.5,
"learning_rate": 5.043227665706052e-06,
"loss": 2.3991,
"step": 350
},
{
"epoch": 0.54,
"learning_rate": 5.403458213256485e-06,
"loss": 2.192,
"step": 375
},
{
"epoch": 0.58,
"learning_rate": 5.763688760806917e-06,
"loss": 2.0522,
"step": 400
},
{
"epoch": 0.61,
"learning_rate": 6.109510086455331e-06,
"loss": 1.9705,
"step": 425
},
{
"epoch": 0.65,
"learning_rate": 6.469740634005765e-06,
"loss": 1.8122,
"step": 450
},
{
"epoch": 0.68,
"learning_rate": 6.815561959654179e-06,
"loss": 1.7162,
"step": 475
},
{
"epoch": 0.72,
"learning_rate": 7.175792507204612e-06,
"loss": 1.632,
"step": 500
},
{
"epoch": 0.76,
"learning_rate": 7.536023054755044e-06,
"loss": 1.5107,
"step": 525
},
{
"epoch": 0.79,
"learning_rate": 7.896253602305477e-06,
"loss": 1.4526,
"step": 550
},
{
"epoch": 0.83,
"learning_rate": 8.25648414985591e-06,
"loss": 1.3565,
"step": 575
},
{
"epoch": 0.86,
"learning_rate": 8.61671469740634e-06,
"loss": 1.3658,
"step": 600
},
{
"epoch": 0.9,
"learning_rate": 8.976945244956773e-06,
"loss": 1.2536,
"step": 625
},
{
"epoch": 0.94,
"learning_rate": 9.337175792507205e-06,
"loss": 1.171,
"step": 650
},
{
"epoch": 0.97,
"learning_rate": 9.697406340057637e-06,
"loss": 1.1767,
"step": 675
},
{
"epoch": 1.0,
"eval_accuracy": 0.751358695652174,
"eval_loss": 1.1062874794006348,
"eval_runtime": 157.9227,
"eval_samples_per_second": 37.284,
"eval_steps_per_second": 2.33,
"step": 694
},
{
"epoch": 1.01,
"learning_rate": 9.995196926032662e-06,
"loss": 1.2008,
"step": 700
},
{
"epoch": 1.04,
"learning_rate": 9.955171309638169e-06,
"loss": 1.0066,
"step": 725
},
{
"epoch": 1.08,
"learning_rate": 9.915145693243676e-06,
"loss": 0.9397,
"step": 750
},
{
"epoch": 1.12,
"learning_rate": 9.875120076849185e-06,
"loss": 0.8813,
"step": 775
},
{
"epoch": 1.15,
"learning_rate": 9.835094460454691e-06,
"loss": 0.9198,
"step": 800
},
{
"epoch": 1.19,
"learning_rate": 9.7950688440602e-06,
"loss": 0.7896,
"step": 825
},
{
"epoch": 1.22,
"learning_rate": 9.755043227665707e-06,
"loss": 0.8289,
"step": 850
},
{
"epoch": 1.26,
"learning_rate": 9.715017611271215e-06,
"loss": 0.771,
"step": 875
},
{
"epoch": 1.3,
"learning_rate": 9.674991994876722e-06,
"loss": 0.7692,
"step": 900
},
{
"epoch": 1.33,
"learning_rate": 9.63496637848223e-06,
"loss": 0.7407,
"step": 925
},
{
"epoch": 1.37,
"learning_rate": 9.594940762087736e-06,
"loss": 0.757,
"step": 950
},
{
"epoch": 1.4,
"learning_rate": 9.554915145693245e-06,
"loss": 0.7397,
"step": 975
},
{
"epoch": 1.44,
"learning_rate": 9.514889529298752e-06,
"loss": 0.7409,
"step": 1000
},
{
"epoch": 1.48,
"learning_rate": 9.47486391290426e-06,
"loss": 0.701,
"step": 1025
},
{
"epoch": 1.51,
"learning_rate": 9.434838296509767e-06,
"loss": 0.6916,
"step": 1050
},
{
"epoch": 1.55,
"learning_rate": 9.394812680115276e-06,
"loss": 0.6818,
"step": 1075
},
{
"epoch": 1.59,
"learning_rate": 9.354787063720781e-06,
"loss": 0.6406,
"step": 1100
},
{
"epoch": 1.62,
"learning_rate": 9.31476144732629e-06,
"loss": 0.6734,
"step": 1125
},
{
"epoch": 1.66,
"learning_rate": 9.274735830931796e-06,
"loss": 0.6168,
"step": 1150
},
{
"epoch": 1.69,
"learning_rate": 9.234710214537305e-06,
"loss": 0.6079,
"step": 1175
},
{
"epoch": 1.73,
"learning_rate": 9.194684598142812e-06,
"loss": 0.6342,
"step": 1200
},
{
"epoch": 1.77,
"learning_rate": 9.15465898174832e-06,
"loss": 0.6685,
"step": 1225
},
{
"epoch": 1.8,
"learning_rate": 9.114633365353827e-06,
"loss": 0.6057,
"step": 1250
},
{
"epoch": 1.84,
"learning_rate": 9.074607748959334e-06,
"loss": 0.6247,
"step": 1275
},
{
"epoch": 1.87,
"learning_rate": 9.034582132564841e-06,
"loss": 0.6223,
"step": 1300
},
{
"epoch": 1.91,
"learning_rate": 8.99455651617035e-06,
"loss": 0.5928,
"step": 1325
},
{
"epoch": 1.95,
"learning_rate": 8.954530899775857e-06,
"loss": 0.496,
"step": 1350
},
{
"epoch": 1.98,
"learning_rate": 8.914505283381365e-06,
"loss": 0.582,
"step": 1375
},
{
"epoch": 2.0,
"eval_accuracy": 0.8327105978260869,
"eval_loss": 0.6595200896263123,
"eval_runtime": 158.3053,
"eval_samples_per_second": 37.194,
"eval_steps_per_second": 2.325,
"step": 1388
},
{
"epoch": 2.02,
"learning_rate": 8.874479666986872e-06,
"loss": 0.4673,
"step": 1400
},
{
"epoch": 2.05,
"learning_rate": 8.83445405059238e-06,
"loss": 0.3253,
"step": 1425
},
{
"epoch": 2.09,
"learning_rate": 8.794428434197888e-06,
"loss": 0.3313,
"step": 1450
},
{
"epoch": 2.13,
"learning_rate": 8.754402817803395e-06,
"loss": 0.4111,
"step": 1475
},
{
"epoch": 2.16,
"learning_rate": 8.714377201408901e-06,
"loss": 0.3823,
"step": 1500
},
{
"epoch": 2.2,
"learning_rate": 8.67435158501441e-06,
"loss": 0.405,
"step": 1525
},
{
"epoch": 2.23,
"learning_rate": 8.634325968619917e-06,
"loss": 0.369,
"step": 1550
},
{
"epoch": 2.27,
"learning_rate": 8.594300352225426e-06,
"loss": 0.3928,
"step": 1575
},
{
"epoch": 2.31,
"learning_rate": 8.554274735830932e-06,
"loss": 0.3726,
"step": 1600
},
{
"epoch": 2.34,
"learning_rate": 8.51424911943644e-06,
"loss": 0.3263,
"step": 1625
},
{
"epoch": 2.38,
"learning_rate": 8.474223503041948e-06,
"loss": 0.3034,
"step": 1650
},
{
"epoch": 2.41,
"learning_rate": 8.434197886647455e-06,
"loss": 0.3467,
"step": 1675
},
{
"epoch": 2.45,
"learning_rate": 8.394172270252963e-06,
"loss": 0.2828,
"step": 1700
},
{
"epoch": 2.49,
"learning_rate": 8.35414665385847e-06,
"loss": 0.3231,
"step": 1725
},
{
"epoch": 2.52,
"learning_rate": 8.314121037463977e-06,
"loss": 0.3015,
"step": 1750
},
{
"epoch": 2.56,
"learning_rate": 8.274095421069484e-06,
"loss": 0.3196,
"step": 1775
},
{
"epoch": 2.59,
"learning_rate": 8.234069804674993e-06,
"loss": 0.31,
"step": 1800
},
{
"epoch": 2.63,
"learning_rate": 8.1940441882805e-06,
"loss": 0.231,
"step": 1825
},
{
"epoch": 2.67,
"learning_rate": 8.154018571886008e-06,
"loss": 0.271,
"step": 1850
},
{
"epoch": 2.7,
"learning_rate": 8.113992955491515e-06,
"loss": 0.3012,
"step": 1875
},
{
"epoch": 2.74,
"learning_rate": 8.073967339097024e-06,
"loss": 0.3028,
"step": 1900
},
{
"epoch": 2.77,
"learning_rate": 8.03394172270253e-06,
"loss": 0.3093,
"step": 1925
},
{
"epoch": 2.81,
"learning_rate": 7.993916106308038e-06,
"loss": 0.3267,
"step": 1950
},
{
"epoch": 2.85,
"learning_rate": 7.953890489913544e-06,
"loss": 0.3391,
"step": 1975
},
{
"epoch": 2.88,
"learning_rate": 7.913864873519053e-06,
"loss": 0.299,
"step": 2000
},
{
"epoch": 2.92,
"learning_rate": 7.87383925712456e-06,
"loss": 0.304,
"step": 2025
},
{
"epoch": 2.95,
"learning_rate": 7.833813640730068e-06,
"loss": 0.2359,
"step": 2050
},
{
"epoch": 2.99,
"learning_rate": 7.793788024335575e-06,
"loss": 0.3172,
"step": 2075
},
{
"epoch": 3.0,
"eval_accuracy": 0.852921195652174,
"eval_loss": 0.5887084603309631,
"eval_runtime": 159.3156,
"eval_samples_per_second": 36.958,
"eval_steps_per_second": 2.31,
"step": 2082
},
{
"epoch": 3.03,
"learning_rate": 7.753762407941084e-06,
"loss": 0.2212,
"step": 2100
},
{
"epoch": 3.06,
"learning_rate": 7.71373679154659e-06,
"loss": 0.1606,
"step": 2125
},
{
"epoch": 3.1,
"learning_rate": 7.673711175152098e-06,
"loss": 0.1502,
"step": 2150
},
{
"epoch": 3.13,
"learning_rate": 7.633685558757605e-06,
"loss": 0.1679,
"step": 2175
},
{
"epoch": 3.17,
"learning_rate": 7.593659942363113e-06,
"loss": 0.1284,
"step": 2200
},
{
"epoch": 3.21,
"learning_rate": 7.553634325968621e-06,
"loss": 0.1763,
"step": 2225
},
{
"epoch": 3.24,
"learning_rate": 7.513608709574129e-06,
"loss": 0.201,
"step": 2250
},
{
"epoch": 3.28,
"learning_rate": 7.473583093179636e-06,
"loss": 0.1596,
"step": 2275
},
{
"epoch": 3.31,
"learning_rate": 7.4335574767851426e-06,
"loss": 0.1809,
"step": 2300
},
{
"epoch": 3.35,
"learning_rate": 7.39353186039065e-06,
"loss": 0.147,
"step": 2325
},
{
"epoch": 3.39,
"learning_rate": 7.353506243996158e-06,
"loss": 0.2199,
"step": 2350
},
{
"epoch": 3.42,
"learning_rate": 7.313480627601666e-06,
"loss": 0.204,
"step": 2375
},
{
"epoch": 3.46,
"learning_rate": 7.2734550112071735e-06,
"loss": 0.1524,
"step": 2400
},
{
"epoch": 3.49,
"learning_rate": 7.233429394812681e-06,
"loss": 0.1668,
"step": 2425
},
{
"epoch": 3.53,
"learning_rate": 7.193403778418189e-06,
"loss": 0.1332,
"step": 2450
},
{
"epoch": 3.57,
"learning_rate": 7.153378162023695e-06,
"loss": 0.1485,
"step": 2475
},
{
"epoch": 3.6,
"learning_rate": 7.113352545629203e-06,
"loss": 0.1908,
"step": 2500
},
{
"epoch": 3.64,
"learning_rate": 7.0733269292347106e-06,
"loss": 0.1895,
"step": 2525
},
{
"epoch": 3.67,
"learning_rate": 7.033301312840218e-06,
"loss": 0.1378,
"step": 2550
},
{
"epoch": 3.71,
"learning_rate": 6.993275696445726e-06,
"loss": 0.182,
"step": 2575
},
{
"epoch": 3.75,
"learning_rate": 6.953250080051234e-06,
"loss": 0.1899,
"step": 2600
},
{
"epoch": 3.78,
"learning_rate": 6.9132244636567415e-06,
"loss": 0.1497,
"step": 2625
},
{
"epoch": 3.82,
"learning_rate": 6.873198847262248e-06,
"loss": 0.1613,
"step": 2650
},
{
"epoch": 3.85,
"learning_rate": 6.833173230867755e-06,
"loss": 0.1975,
"step": 2675
},
{
"epoch": 3.89,
"learning_rate": 6.793147614473263e-06,
"loss": 0.1377,
"step": 2700
},
{
"epoch": 3.93,
"learning_rate": 6.753121998078771e-06,
"loss": 0.1923,
"step": 2725
},
{
"epoch": 3.96,
"learning_rate": 6.713096381684279e-06,
"loss": 0.1679,
"step": 2750
},
{
"epoch": 4.0,
"learning_rate": 6.673070765289786e-06,
"loss": 0.196,
"step": 2775
},
{
"epoch": 4.0,
"eval_accuracy": 0.8700747282608695,
"eval_loss": 0.533203125,
"eval_runtime": 157.4321,
"eval_samples_per_second": 37.4,
"eval_steps_per_second": 2.338,
"step": 2776
},
{
"epoch": 4.03,
"learning_rate": 6.633045148895294e-06,
"loss": 0.0967,
"step": 2800
},
{
"epoch": 4.07,
"learning_rate": 6.593019532500801e-06,
"loss": 0.0737,
"step": 2825
},
{
"epoch": 4.11,
"learning_rate": 6.552993916106309e-06,
"loss": 0.1141,
"step": 2850
},
{
"epoch": 4.14,
"learning_rate": 6.514569324367596e-06,
"loss": 0.1032,
"step": 2875
},
{
"epoch": 4.18,
"learning_rate": 6.4745437079731035e-06,
"loss": 0.0914,
"step": 2900
},
{
"epoch": 4.21,
"learning_rate": 6.434518091578611e-06,
"loss": 0.0749,
"step": 2925
},
{
"epoch": 4.25,
"learning_rate": 6.394492475184119e-06,
"loss": 0.0905,
"step": 2950
},
{
"epoch": 4.29,
"learning_rate": 6.354466858789627e-06,
"loss": 0.053,
"step": 2975
},
{
"epoch": 4.32,
"learning_rate": 6.314441242395133e-06,
"loss": 0.0892,
"step": 3000
},
{
"epoch": 4.36,
"learning_rate": 6.27601665065642e-06,
"loss": 0.1321,
"step": 3025
},
{
"epoch": 4.39,
"learning_rate": 6.2359910342619275e-06,
"loss": 0.104,
"step": 3050
},
{
"epoch": 4.43,
"learning_rate": 6.195965417867435e-06,
"loss": 0.1087,
"step": 3075
},
{
"epoch": 4.47,
"learning_rate": 6.155939801472943e-06,
"loss": 0.0969,
"step": 3100
},
{
"epoch": 4.5,
"learning_rate": 6.115914185078451e-06,
"loss": 0.0813,
"step": 3125
},
{
"epoch": 4.54,
"learning_rate": 6.0758885686839585e-06,
"loss": 0.0826,
"step": 3150
},
{
"epoch": 4.57,
"learning_rate": 6.035862952289466e-06,
"loss": 0.122,
"step": 3175
},
{
"epoch": 4.61,
"learning_rate": 5.995837335894973e-06,
"loss": 0.0906,
"step": 3200
},
{
"epoch": 4.65,
"learning_rate": 5.955811719500481e-06,
"loss": 0.131,
"step": 3225
},
{
"epoch": 4.68,
"learning_rate": 5.915786103105988e-06,
"loss": 0.1152,
"step": 3250
},
{
"epoch": 4.72,
"learning_rate": 5.8757604867114956e-06,
"loss": 0.133,
"step": 3275
},
{
"epoch": 4.76,
"learning_rate": 5.835734870317003e-06,
"loss": 0.0999,
"step": 3300
},
{
"epoch": 4.79,
"learning_rate": 5.795709253922511e-06,
"loss": 0.1129,
"step": 3325
},
{
"epoch": 4.83,
"learning_rate": 5.755683637528019e-06,
"loss": 0.1011,
"step": 3350
},
{
"epoch": 4.86,
"learning_rate": 5.715658021133526e-06,
"loss": 0.0728,
"step": 3375
},
{
"epoch": 4.9,
"learning_rate": 5.6756324047390334e-06,
"loss": 0.1007,
"step": 3400
},
{
"epoch": 4.94,
"learning_rate": 5.635606788344541e-06,
"loss": 0.0648,
"step": 3425
},
{
"epoch": 4.97,
"learning_rate": 5.595581171950049e-06,
"loss": 0.0858,
"step": 3450
},
{
"epoch": 5.0,
"eval_accuracy": 0.8733016304347826,
"eval_loss": 0.5704634189605713,
"eval_runtime": 157.7884,
"eval_samples_per_second": 37.316,
"eval_steps_per_second": 2.332,
"step": 3470
},
{
"epoch": 5.01,
"learning_rate": 5.555555555555557e-06,
"loss": 0.1054,
"step": 3475
},
{
"epoch": 5.04,
"learning_rate": 5.5155299391610636e-06,
"loss": 0.0637,
"step": 3500
},
{
"epoch": 5.08,
"learning_rate": 5.4755043227665705e-06,
"loss": 0.0593,
"step": 3525
},
{
"epoch": 5.12,
"learning_rate": 5.435478706372078e-06,
"loss": 0.0762,
"step": 3550
},
{
"epoch": 5.15,
"learning_rate": 5.395453089977586e-06,
"loss": 0.0422,
"step": 3575
},
{
"epoch": 5.19,
"learning_rate": 5.355427473583094e-06,
"loss": 0.0659,
"step": 3600
},
{
"epoch": 5.22,
"learning_rate": 5.3154018571886014e-06,
"loss": 0.0697,
"step": 3625
},
{
"epoch": 5.26,
"learning_rate": 5.275376240794109e-06,
"loss": 0.0331,
"step": 3650
},
{
"epoch": 5.3,
"learning_rate": 5.235350624399617e-06,
"loss": 0.0855,
"step": 3675
},
{
"epoch": 5.33,
"learning_rate": 5.195325008005123e-06,
"loss": 0.0715,
"step": 3700
},
{
"epoch": 5.37,
"learning_rate": 5.155299391610631e-06,
"loss": 0.0724,
"step": 3725
},
{
"epoch": 5.4,
"learning_rate": 5.1152737752161385e-06,
"loss": 0.0802,
"step": 3750
},
{
"epoch": 5.44,
"learning_rate": 5.075248158821646e-06,
"loss": 0.0694,
"step": 3775
},
{
"epoch": 5.48,
"learning_rate": 5.035222542427154e-06,
"loss": 0.0611,
"step": 3800
},
{
"epoch": 5.51,
"learning_rate": 4.995196926032661e-06,
"loss": 0.0453,
"step": 3825
},
{
"epoch": 5.55,
"learning_rate": 4.955171309638169e-06,
"loss": 0.058,
"step": 3850
},
{
"epoch": 5.58,
"learning_rate": 4.915145693243676e-06,
"loss": 0.0582,
"step": 3875
},
{
"epoch": 5.62,
"learning_rate": 4.875120076849184e-06,
"loss": 0.0567,
"step": 3900
},
{
"epoch": 5.66,
"learning_rate": 4.835094460454691e-06,
"loss": 0.0541,
"step": 3925
},
{
"epoch": 5.69,
"learning_rate": 4.795068844060199e-06,
"loss": 0.098,
"step": 3950
},
{
"epoch": 5.73,
"learning_rate": 4.7550432276657065e-06,
"loss": 0.0747,
"step": 3975
},
{
"epoch": 5.76,
"learning_rate": 4.715017611271213e-06,
"loss": 0.0363,
"step": 4000
},
{
"epoch": 5.8,
"learning_rate": 4.674991994876721e-06,
"loss": 0.0719,
"step": 4025
},
{
"epoch": 5.84,
"learning_rate": 4.634966378482229e-06,
"loss": 0.0356,
"step": 4050
},
{
"epoch": 5.87,
"learning_rate": 4.594940762087737e-06,
"loss": 0.0325,
"step": 4075
},
{
"epoch": 5.91,
"learning_rate": 4.5549151456932435e-06,
"loss": 0.0237,
"step": 4100
},
{
"epoch": 5.94,
"learning_rate": 4.514889529298751e-06,
"loss": 0.062,
"step": 4125
},
{
"epoch": 5.98,
"learning_rate": 4.474863912904259e-06,
"loss": 0.0477,
"step": 4150
},
{
"epoch": 6.0,
"eval_accuracy": 0.8778872282608695,
"eval_loss": 0.631108820438385,
"eval_runtime": 157.6105,
"eval_samples_per_second": 37.358,
"eval_steps_per_second": 2.335,
"step": 4164
},
{
"epoch": 6.02,
"learning_rate": 4.434838296509767e-06,
"loss": 0.0733,
"step": 4175
},
{
"epoch": 6.05,
"learning_rate": 4.394812680115274e-06,
"loss": 0.0266,
"step": 4200
},
{
"epoch": 6.09,
"learning_rate": 4.354787063720781e-06,
"loss": 0.0211,
"step": 4225
},
{
"epoch": 6.12,
"learning_rate": 4.314761447326289e-06,
"loss": 0.0383,
"step": 4250
},
{
"epoch": 6.16,
"learning_rate": 4.274735830931797e-06,
"loss": 0.0466,
"step": 4275
},
{
"epoch": 6.2,
"learning_rate": 4.234710214537305e-06,
"loss": 0.023,
"step": 4300
},
{
"epoch": 6.23,
"learning_rate": 4.1946845981428115e-06,
"loss": 0.0366,
"step": 4325
},
{
"epoch": 6.27,
"learning_rate": 4.154658981748319e-06,
"loss": 0.0538,
"step": 4350
},
{
"epoch": 6.3,
"learning_rate": 4.114633365353827e-06,
"loss": 0.0283,
"step": 4375
},
{
"epoch": 6.34,
"learning_rate": 4.074607748959335e-06,
"loss": 0.0241,
"step": 4400
},
{
"epoch": 6.38,
"learning_rate": 4.0345821325648425e-06,
"loss": 0.0621,
"step": 4425
},
{
"epoch": 6.41,
"learning_rate": 3.994556516170349e-06,
"loss": 0.0318,
"step": 4450
},
{
"epoch": 6.45,
"learning_rate": 3.954530899775857e-06,
"loss": 0.0394,
"step": 4475
},
{
"epoch": 6.48,
"learning_rate": 3.914505283381365e-06,
"loss": 0.0455,
"step": 4500
},
{
"epoch": 6.52,
"learning_rate": 3.874479666986872e-06,
"loss": 0.0468,
"step": 4525
},
{
"epoch": 6.56,
"learning_rate": 3.8344540505923795e-06,
"loss": 0.0215,
"step": 4550
},
{
"epoch": 6.59,
"learning_rate": 3.7944284341978873e-06,
"loss": 0.0449,
"step": 4575
},
{
"epoch": 6.63,
"learning_rate": 3.754402817803394e-06,
"loss": 0.0158,
"step": 4600
},
{
"epoch": 6.66,
"learning_rate": 3.714377201408902e-06,
"loss": 0.0416,
"step": 4625
},
{
"epoch": 6.7,
"learning_rate": 3.6743515850144097e-06,
"loss": 0.0357,
"step": 4650
},
{
"epoch": 6.74,
"learning_rate": 3.6343259686199174e-06,
"loss": 0.0505,
"step": 4675
},
{
"epoch": 6.77,
"learning_rate": 3.5943003522254243e-06,
"loss": 0.0375,
"step": 4700
},
{
"epoch": 6.81,
"learning_rate": 3.554274735830932e-06,
"loss": 0.0561,
"step": 4725
},
{
"epoch": 6.84,
"learning_rate": 3.51424911943644e-06,
"loss": 0.0639,
"step": 4750
},
{
"epoch": 6.88,
"learning_rate": 3.4742235030419467e-06,
"loss": 0.0616,
"step": 4775
},
{
"epoch": 6.92,
"learning_rate": 3.4341978866474544e-06,
"loss": 0.0351,
"step": 4800
},
{
"epoch": 6.95,
"learning_rate": 3.394172270252962e-06,
"loss": 0.0464,
"step": 4825
},
{
"epoch": 6.99,
"learning_rate": 3.35414665385847e-06,
"loss": 0.0353,
"step": 4850
},
{
"epoch": 7.0,
"eval_accuracy": 0.8824728260869565,
"eval_loss": 0.6011340022087097,
"eval_runtime": 160.1074,
"eval_samples_per_second": 36.775,
"eval_steps_per_second": 2.298,
"step": 4858
},
{
"epoch": 7.02,
"learning_rate": 3.3141210374639773e-06,
"loss": 0.0298,
"step": 4875
},
{
"epoch": 7.06,
"learning_rate": 3.2740954210694846e-06,
"loss": 0.0391,
"step": 4900
},
{
"epoch": 7.1,
"learning_rate": 3.2340698046749923e-06,
"loss": 0.031,
"step": 4925
},
{
"epoch": 7.13,
"learning_rate": 3.1940441882804996e-06,
"loss": 0.0287,
"step": 4950
},
{
"epoch": 7.17,
"learning_rate": 3.1540185718860074e-06,
"loss": 0.0317,
"step": 4975
},
{
"epoch": 7.2,
"learning_rate": 3.1139929554915147e-06,
"loss": 0.0332,
"step": 5000
},
{
"epoch": 7.24,
"learning_rate": 3.0739673390970225e-06,
"loss": 0.0153,
"step": 5025
},
{
"epoch": 7.28,
"learning_rate": 3.0339417227025298e-06,
"loss": 0.0376,
"step": 5050
},
{
"epoch": 7.31,
"learning_rate": 2.9939161063080375e-06,
"loss": 0.0464,
"step": 5075
},
{
"epoch": 7.35,
"learning_rate": 2.9538904899135453e-06,
"loss": 0.0413,
"step": 5100
},
{
"epoch": 7.38,
"learning_rate": 2.913864873519052e-06,
"loss": 0.0213,
"step": 5125
},
{
"epoch": 7.42,
"learning_rate": 2.87383925712456e-06,
"loss": 0.0214,
"step": 5150
},
{
"epoch": 7.46,
"learning_rate": 2.8338136407300677e-06,
"loss": 0.0109,
"step": 5175
},
{
"epoch": 7.49,
"learning_rate": 2.7937880243355746e-06,
"loss": 0.0361,
"step": 5200
},
{
"epoch": 7.53,
"learning_rate": 2.7537624079410823e-06,
"loss": 0.0222,
"step": 5225
},
{
"epoch": 7.56,
"learning_rate": 2.71373679154659e-06,
"loss": 0.0136,
"step": 5250
},
{
"epoch": 7.6,
"learning_rate": 2.673711175152098e-06,
"loss": 0.0028,
"step": 5275
},
{
"epoch": 7.64,
"learning_rate": 2.6336855587576047e-06,
"loss": 0.0082,
"step": 5300
},
{
"epoch": 7.67,
"learning_rate": 2.5936599423631124e-06,
"loss": 0.0211,
"step": 5325
},
{
"epoch": 7.71,
"learning_rate": 2.55363432596862e-06,
"loss": 0.018,
"step": 5350
},
{
"epoch": 7.74,
"learning_rate": 2.5136087095741275e-06,
"loss": 0.0252,
"step": 5375
},
{
"epoch": 7.78,
"learning_rate": 2.4735830931796352e-06,
"loss": 0.0313,
"step": 5400
},
{
"epoch": 7.82,
"learning_rate": 2.4335574767851426e-06,
"loss": 0.0183,
"step": 5425
},
{
"epoch": 7.85,
"learning_rate": 2.3935318603906503e-06,
"loss": 0.02,
"step": 5450
},
{
"epoch": 7.89,
"learning_rate": 2.3535062439961576e-06,
"loss": 0.0415,
"step": 5475
},
{
"epoch": 7.93,
"learning_rate": 2.3134806276016654e-06,
"loss": 0.0171,
"step": 5500
},
{
"epoch": 7.96,
"learning_rate": 2.2734550112071727e-06,
"loss": 0.01,
"step": 5525
},
{
"epoch": 8.0,
"learning_rate": 2.2334293948126804e-06,
"loss": 0.0033,
"step": 5550
},
{
"epoch": 8.0,
"eval_accuracy": 0.8843410326086957,
"eval_loss": 0.618594765663147,
"eval_runtime": 158.9755,
"eval_samples_per_second": 37.037,
"eval_steps_per_second": 2.315,
"step": 5552
},
{
"epoch": 8.03,
"learning_rate": 2.1934037784181878e-06,
"loss": 0.0298,
"step": 5575
},
{
"epoch": 8.07,
"learning_rate": 2.1533781620236955e-06,
"loss": 0.0171,
"step": 5600
},
{
"epoch": 8.11,
"learning_rate": 2.113352545629203e-06,
"loss": 0.0019,
"step": 5625
},
{
"epoch": 8.14,
"learning_rate": 2.0733269292347106e-06,
"loss": 0.0101,
"step": 5650
},
{
"epoch": 8.18,
"learning_rate": 2.033301312840218e-06,
"loss": 0.0146,
"step": 5675
},
{
"epoch": 8.21,
"learning_rate": 1.9932756964457252e-06,
"loss": 0.0189,
"step": 5700
},
{
"epoch": 8.25,
"learning_rate": 1.953250080051233e-06,
"loss": 0.028,
"step": 5725
},
{
"epoch": 8.29,
"learning_rate": 1.9132244636567403e-06,
"loss": 0.0075,
"step": 5750
},
{
"epoch": 8.32,
"learning_rate": 1.873198847262248e-06,
"loss": 0.0041,
"step": 5775
},
{
"epoch": 8.36,
"learning_rate": 1.8331732308677556e-06,
"loss": 0.036,
"step": 5800
},
{
"epoch": 8.39,
"learning_rate": 1.7931476144732629e-06,
"loss": 0.0062,
"step": 5825
},
{
"epoch": 8.43,
"learning_rate": 1.7531219980787706e-06,
"loss": 0.0102,
"step": 5850
},
{
"epoch": 8.47,
"learning_rate": 1.713096381684278e-06,
"loss": 0.0154,
"step": 5875
},
{
"epoch": 8.5,
"learning_rate": 1.6730707652897857e-06,
"loss": 0.0083,
"step": 5900
},
{
"epoch": 8.54,
"learning_rate": 1.633045148895293e-06,
"loss": 0.0126,
"step": 5925
},
{
"epoch": 8.57,
"learning_rate": 1.5930195325008008e-06,
"loss": 0.0044,
"step": 5950
},
{
"epoch": 8.61,
"learning_rate": 1.552993916106308e-06,
"loss": 0.0221,
"step": 5975
},
{
"epoch": 8.65,
"learning_rate": 1.5129682997118156e-06,
"loss": 0.0164,
"step": 6000
},
{
"epoch": 8.68,
"learning_rate": 1.4729426833173234e-06,
"loss": 0.0247,
"step": 6025
},
{
"epoch": 8.72,
"learning_rate": 1.4329170669228307e-06,
"loss": 0.0101,
"step": 6050
},
{
"epoch": 8.75,
"learning_rate": 1.3928914505283384e-06,
"loss": 0.0053,
"step": 6075
},
{
"epoch": 8.79,
"learning_rate": 1.3528658341338458e-06,
"loss": 0.014,
"step": 6100
},
{
"epoch": 8.83,
"learning_rate": 1.312840217739353e-06,
"loss": 0.0124,
"step": 6125
},
{
"epoch": 8.86,
"learning_rate": 1.2728146013448608e-06,
"loss": 0.0017,
"step": 6150
},
{
"epoch": 8.9,
"learning_rate": 1.2327889849503684e-06,
"loss": 0.0089,
"step": 6175
},
{
"epoch": 8.93,
"learning_rate": 1.1927633685558759e-06,
"loss": 0.0334,
"step": 6200
},
{
"epoch": 8.97,
"learning_rate": 1.1527377521613834e-06,
"loss": 0.0071,
"step": 6225
},
{
"epoch": 9.0,
"eval_accuracy": 0.8860394021739131,
"eval_loss": 0.6408511996269226,
"eval_runtime": 157.4747,
"eval_samples_per_second": 37.39,
"eval_steps_per_second": 2.337,
"step": 6246
},
{
"epoch": 9.01,
"learning_rate": 1.112712135766891e-06,
"loss": 0.0139,
"step": 6250
},
{
"epoch": 9.04,
"learning_rate": 1.0726865193723985e-06,
"loss": 0.0019,
"step": 6275
},
{
"epoch": 9.08,
"learning_rate": 1.032660902977906e-06,
"loss": 0.0061,
"step": 6300
},
{
"epoch": 9.11,
"learning_rate": 9.926352865834136e-07,
"loss": 0.0148,
"step": 6325
},
{
"epoch": 9.15,
"learning_rate": 9.526096701889209e-07,
"loss": 0.0183,
"step": 6350
},
{
"epoch": 9.19,
"learning_rate": 9.125840537944285e-07,
"loss": 0.0044,
"step": 6375
},
{
"epoch": 9.22,
"learning_rate": 8.72558437399936e-07,
"loss": 0.0019,
"step": 6400
},
{
"epoch": 9.26,
"learning_rate": 8.325328210054436e-07,
"loss": 0.0021,
"step": 6425
},
{
"epoch": 9.29,
"learning_rate": 7.925072046109511e-07,
"loss": 0.0105,
"step": 6450
},
{
"epoch": 9.33,
"learning_rate": 7.524815882164586e-07,
"loss": 0.0039,
"step": 6475
},
{
"epoch": 9.37,
"learning_rate": 7.124559718219662e-07,
"loss": 0.0019,
"step": 6500
},
{
"epoch": 9.4,
"learning_rate": 6.724303554274736e-07,
"loss": 0.01,
"step": 6525
},
{
"epoch": 9.44,
"learning_rate": 6.324047390329811e-07,
"loss": 0.0135,
"step": 6550
},
{
"epoch": 9.47,
"learning_rate": 5.923791226384887e-07,
"loss": 0.0016,
"step": 6575
},
{
"epoch": 9.51,
"learning_rate": 5.523535062439962e-07,
"loss": 0.0095,
"step": 6600
},
{
"epoch": 9.55,
"learning_rate": 5.123278898495037e-07,
"loss": 0.0223,
"step": 6625
},
{
"epoch": 9.58,
"learning_rate": 4.723022734550112e-07,
"loss": 0.0128,
"step": 6650
},
{
"epoch": 9.62,
"learning_rate": 4.322766570605188e-07,
"loss": 0.0156,
"step": 6675
},
{
"epoch": 9.65,
"learning_rate": 3.9225104066602624e-07,
"loss": 0.0109,
"step": 6700
},
{
"epoch": 9.69,
"learning_rate": 3.522254242715338e-07,
"loss": 0.0037,
"step": 6725
},
{
"epoch": 9.73,
"learning_rate": 3.121998078770413e-07,
"loss": 0.0155,
"step": 6750
},
{
"epoch": 9.76,
"learning_rate": 2.7217419148254884e-07,
"loss": 0.0068,
"step": 6775
},
{
"epoch": 9.8,
"learning_rate": 2.3214857508805637e-07,
"loss": 0.0101,
"step": 6800
},
{
"epoch": 9.83,
"learning_rate": 1.921229586935639e-07,
"loss": 0.0182,
"step": 6825
},
{
"epoch": 9.87,
"learning_rate": 1.520973422990714e-07,
"loss": 0.002,
"step": 6850
},
{
"epoch": 9.91,
"learning_rate": 1.1207172590457893e-07,
"loss": 0.0017,
"step": 6875
},
{
"epoch": 9.94,
"learning_rate": 7.204610951008646e-08,
"loss": 0.0039,
"step": 6900
},
{
"epoch": 9.98,
"learning_rate": 3.2020493115593984e-08,
"loss": 0.0074,
"step": 6925
},
{
"epoch": 10.0,
"eval_accuracy": 0.8860394021739131,
"eval_loss": 0.6333902478218079,
"eval_runtime": 161.8306,
"eval_samples_per_second": 36.384,
"eval_steps_per_second": 2.274,
"step": 6940
},
{
"epoch": 10.0,
"step": 6940,
"total_flos": 2.7871930756137603e+19,
"train_loss": 0.39663364636880866,
"train_runtime": 15037.7635,
"train_samples_per_second": 14.759,
"train_steps_per_second": 0.462
}
],
"max_steps": 6940,
"num_train_epochs": 10,
"total_flos": 2.7871930756137603e+19,
"trial_name": null,
"trial_params": null
}