novaspeech-test-0.1 / trainer_state.json
Mazino0's picture
Upload checkpoint 14400 folder contents
d8b0400 verified
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.4999826394916843,
"eval_steps": 720,
"global_step": 14400,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 3.472101663136697e-05,
"grad_norm": 12.5,
"learning_rate": 0.0,
"loss": 0.7266,
"step": 1
},
{
"epoch": 0.0034721016631366966,
"grad_norm": 0.2353515625,
"learning_rate": 0.00019800000000000002,
"loss": 0.3923,
"step": 100
},
{
"epoch": 0.006944203326273393,
"grad_norm": 0.1513671875,
"learning_rate": 0.000398,
"loss": 0.2685,
"step": 200
},
{
"epoch": 0.01041630498941009,
"grad_norm": 0.1748046875,
"learning_rate": 0.000598,
"loss": 0.2522,
"step": 300
},
{
"epoch": 0.013888406652546786,
"grad_norm": 0.2353515625,
"learning_rate": 0.0007980000000000001,
"loss": 0.2361,
"step": 400
},
{
"epoch": 0.017360508315683483,
"grad_norm": 0.1533203125,
"learning_rate": 0.000998,
"loss": 0.1964,
"step": 500
},
{
"epoch": 0.02083260997882018,
"grad_norm": 0.1357421875,
"learning_rate": 0.001198,
"loss": 0.1519,
"step": 600
},
{
"epoch": 0.024304711641956876,
"grad_norm": 0.11181640625,
"learning_rate": 0.001398,
"loss": 0.1331,
"step": 700
},
{
"epoch": 0.024999131974584216,
"eval_covost2-en-de_loss": 3.690591812133789,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 77.2672,
"eval_covost2-en-de_samples_per_second": 0.828,
"eval_covost2-en-de_steps_per_second": 0.052,
"step": 720
},
{
"epoch": 0.024999131974584216,
"eval_covost2-zh-en_loss": 4.936097145080566,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 74.4347,
"eval_covost2-zh-en_samples_per_second": 0.86,
"eval_covost2-zh-en_steps_per_second": 0.054,
"step": 720
},
{
"epoch": 0.024999131974584216,
"eval_peoplespeech-clean-transcription_loss": 4.681502819061279,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 78.2658,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.818,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.051,
"step": 720
},
{
"epoch": 0.027776813305093573,
"grad_norm": 0.11328125,
"learning_rate": 0.001598,
"loss": 0.1232,
"step": 800
},
{
"epoch": 0.03124891496823027,
"grad_norm": 0.1298828125,
"learning_rate": 0.0017980000000000001,
"loss": 0.1157,
"step": 900
},
{
"epoch": 0.034721016631366966,
"grad_norm": 0.10693359375,
"learning_rate": 0.001998,
"loss": 0.1108,
"step": 1000
},
{
"epoch": 0.03819311829450366,
"grad_norm": 0.09326171875,
"learning_rate": 0.001999757588312555,
"loss": 0.1128,
"step": 1100
},
{
"epoch": 0.04166521995764036,
"grad_norm": 0.11376953125,
"learning_rate": 0.0019990206677939332,
"loss": 0.1025,
"step": 1200
},
{
"epoch": 0.045137321620777056,
"grad_norm": 0.07763671875,
"learning_rate": 0.0019977896187431264,
"loss": 0.1,
"step": 1300
},
{
"epoch": 0.04860942328391375,
"grad_norm": 0.0634765625,
"learning_rate": 0.0019960651177813524,
"loss": 0.0965,
"step": 1400
},
{
"epoch": 0.04999826394916843,
"eval_covost2-en-de_loss": 3.5930023193359375,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 80.0093,
"eval_covost2-en-de_samples_per_second": 0.8,
"eval_covost2-en-de_steps_per_second": 0.05,
"step": 1440
},
{
"epoch": 0.04999826394916843,
"eval_covost2-zh-en_loss": 4.925909042358398,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 74.5086,
"eval_covost2-zh-en_samples_per_second": 0.859,
"eval_covost2-zh-en_steps_per_second": 0.054,
"step": 1440
},
{
"epoch": 0.04999826394916843,
"eval_peoplespeech-clean-transcription_loss": 4.403814315795898,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 78.7678,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.813,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.051,
"step": 1440
},
{
"epoch": 0.05208152494705045,
"grad_norm": 0.08837890625,
"learning_rate": 0.0019938481127456968,
"loss": 0.0965,
"step": 1500
},
{
"epoch": 0.055553626610187146,
"grad_norm": 0.0634765625,
"learning_rate": 0.0019911398221681507,
"loss": 0.0937,
"step": 1600
},
{
"epoch": 0.05902572827332384,
"grad_norm": 0.0888671875,
"learning_rate": 0.0019879417346058688,
"loss": 0.092,
"step": 1700
},
{
"epoch": 0.06249782993646054,
"grad_norm": 0.07080078125,
"learning_rate": 0.0019842556078230158,
"loss": 0.0903,
"step": 1800
},
{
"epoch": 0.06596993159959724,
"grad_norm": 0.087890625,
"learning_rate": 0.0019800834678246454,
"loss": 0.0915,
"step": 1900
},
{
"epoch": 0.06944203326273393,
"grad_norm": 0.0673828125,
"learning_rate": 0.001975427607743149,
"loss": 0.0894,
"step": 2000
},
{
"epoch": 0.07291413492587064,
"grad_norm": 0.06591796875,
"learning_rate": 0.0019702905865778804,
"loss": 0.0908,
"step": 2100
},
{
"epoch": 0.07499739592375265,
"eval_covost2-en-de_loss": 3.5631778240203857,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 77.789,
"eval_covost2-en-de_samples_per_second": 0.823,
"eval_covost2-en-de_steps_per_second": 0.051,
"step": 2160
},
{
"epoch": 0.07499739592375265,
"eval_covost2-zh-en_loss": 4.879820823669434,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 76.2131,
"eval_covost2-zh-en_samples_per_second": 0.84,
"eval_covost2-zh-en_steps_per_second": 0.052,
"step": 2160
},
{
"epoch": 0.07499739592375265,
"eval_peoplespeech-clean-transcription_loss": 4.396980285644531,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 77.3777,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.827,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.052,
"step": 2160
},
{
"epoch": 0.07638623658900733,
"grad_norm": 0.060791015625,
"learning_rate": 0.0019646752277886546,
"loss": 0.0879,
"step": 2200
},
{
"epoch": 0.07985833825214403,
"grad_norm": 0.05078125,
"learning_rate": 0.0019585846177438897,
"loss": 0.0848,
"step": 2300
},
{
"epoch": 0.08333043991528072,
"grad_norm": 0.0751953125,
"learning_rate": 0.001952022104024247,
"loss": 0.0848,
"step": 2400
},
{
"epoch": 0.08680254157841742,
"grad_norm": 0.0732421875,
"learning_rate": 0.0019449912935827003,
"loss": 0.0849,
"step": 2500
},
{
"epoch": 0.09027464324155411,
"grad_norm": 0.061279296875,
"learning_rate": 0.0019374960507620488,
"loss": 0.0858,
"step": 2600
},
{
"epoch": 0.09374674490469082,
"grad_norm": 0.0615234375,
"learning_rate": 0.0019295404951709576,
"loss": 0.0849,
"step": 2700
},
{
"epoch": 0.0972188465678275,
"grad_norm": 0.0849609375,
"learning_rate": 0.0019211289994196992,
"loss": 0.0852,
"step": 2800
},
{
"epoch": 0.09999652789833686,
"eval_covost2-en-de_loss": 3.5589070320129395,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 78.7708,
"eval_covost2-en-de_samples_per_second": 0.812,
"eval_covost2-en-de_steps_per_second": 0.051,
"step": 2880
},
{
"epoch": 0.09999652789833686,
"eval_covost2-zh-en_loss": 4.881261825561523,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 74.497,
"eval_covost2-zh-en_samples_per_second": 0.859,
"eval_covost2-zh-en_steps_per_second": 0.054,
"step": 2880
},
{
"epoch": 0.09999652789833686,
"eval_peoplespeech-clean-transcription_loss": 4.360118865966797,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 78.7472,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.813,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.051,
"step": 2880
},
{
"epoch": 0.10069094823096421,
"grad_norm": 0.07373046875,
"learning_rate": 0.0019122661867168361,
"loss": 0.0829,
"step": 2900
},
{
"epoch": 0.1041630498941009,
"grad_norm": 0.06005859375,
"learning_rate": 0.001902956928328167,
"loss": 0.0841,
"step": 3000
},
{
"epoch": 0.1076351515572376,
"grad_norm": 0.07080078125,
"learning_rate": 0.001893206340899336,
"loss": 0.0817,
"step": 3100
},
{
"epoch": 0.11110725322037429,
"grad_norm": 0.044677734375,
"learning_rate": 0.0018830197836435698,
"loss": 0.0809,
"step": 3200
},
{
"epoch": 0.114579354883511,
"grad_norm": 0.056396484375,
"learning_rate": 0.001872402855396096,
"loss": 0.0821,
"step": 3300
},
{
"epoch": 0.11805145654664768,
"grad_norm": 0.08349609375,
"learning_rate": 0.0018613613915368545,
"loss": 0.082,
"step": 3400
},
{
"epoch": 0.12152355820978439,
"grad_norm": 0.048828125,
"learning_rate": 0.0018499014607832023,
"loss": 0.0819,
"step": 3500
},
{
"epoch": 0.12499565987292108,
"grad_norm": 0.041748046875,
"learning_rate": 0.0018380293618543617,
"loss": 0.0818,
"step": 3600
},
{
"epoch": 0.12499565987292108,
"eval_covost2-en-de_loss": 3.5445632934570312,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 77.8828,
"eval_covost2-en-de_samples_per_second": 0.822,
"eval_covost2-en-de_steps_per_second": 0.051,
"step": 3600
},
{
"epoch": 0.12499565987292108,
"eval_covost2-zh-en_loss": 4.845459938049316,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 85.5834,
"eval_covost2-zh-en_samples_per_second": 0.748,
"eval_covost2-zh-en_steps_per_second": 0.047,
"step": 3600
},
{
"epoch": 0.12499565987292108,
"eval_peoplespeech-clean-transcription_loss": 4.317172527313232,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 77.703,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.824,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.051,
"step": 3600
},
{
"epoch": 0.12846776153605777,
"grad_norm": 0.0693359375,
"learning_rate": 0.0018257516200094602,
"loss": 0.0808,
"step": 3700
},
{
"epoch": 0.13193986319919448,
"grad_norm": 0.0537109375,
"learning_rate": 0.001813074983461051,
"loss": 0.0799,
"step": 3800
},
{
"epoch": 0.13541196486233117,
"grad_norm": 0.052490234375,
"learning_rate": 0.001800006419666096,
"loss": 0.0794,
"step": 3900
},
{
"epoch": 0.13888406652546786,
"grad_norm": 0.056396484375,
"learning_rate": 0.0017865531114964436,
"loss": 0.0796,
"step": 4000
},
{
"epoch": 0.14235616818860455,
"grad_norm": 0.07861328125,
"learning_rate": 0.0017727224532909083,
"loss": 0.0797,
"step": 4100
},
{
"epoch": 0.14582826985174127,
"grad_norm": 0.0546875,
"learning_rate": 0.001758522046791122,
"loss": 0.0786,
"step": 4200
},
{
"epoch": 0.14930037151487796,
"grad_norm": 0.057373046875,
"learning_rate": 0.0017439596969633904,
"loss": 0.0789,
"step": 4300
},
{
"epoch": 0.1499947918475053,
"eval_covost2-en-de_loss": 3.547935962677002,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 79.1172,
"eval_covost2-en-de_samples_per_second": 0.809,
"eval_covost2-en-de_steps_per_second": 0.051,
"step": 4320
},
{
"epoch": 0.1499947918475053,
"eval_covost2-zh-en_loss": 4.872008800506592,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 75.0696,
"eval_covost2-zh-en_samples_per_second": 0.853,
"eval_covost2-zh-en_steps_per_second": 0.053,
"step": 4320
},
{
"epoch": 0.1499947918475053,
"eval_peoplespeech-clean-transcription_loss": 4.321802139282227,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 77.2149,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.829,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.052,
"step": 4320
},
{
"epoch": 0.15277247317801465,
"grad_norm": 0.057373046875,
"learning_rate": 0.0017290434077088508,
"loss": 0.079,
"step": 4400
},
{
"epoch": 0.15624457484115134,
"grad_norm": 0.055908203125,
"learning_rate": 0.00171378137746429,
"loss": 0.0791,
"step": 4500
},
{
"epoch": 0.15971667650428806,
"grad_norm": 0.047119140625,
"learning_rate": 0.00169818199469604,
"loss": 0.0787,
"step": 4600
},
{
"epoch": 0.16318877816742475,
"grad_norm": 0.0380859375,
"learning_rate": 0.0016822538332894243,
"loss": 0.0783,
"step": 4700
},
{
"epoch": 0.16666087983056144,
"grad_norm": 0.06396484375,
"learning_rate": 0.0016660056478362997,
"loss": 0.077,
"step": 4800
},
{
"epoch": 0.17013298149369813,
"grad_norm": 0.050537109375,
"learning_rate": 0.0016494463688232674,
"loss": 0.0793,
"step": 4900
},
{
"epoch": 0.17360508315683484,
"grad_norm": 0.048583984375,
"learning_rate": 0.0016325850977232151,
"loss": 0.0763,
"step": 5000
},
{
"epoch": 0.1749939238220895,
"eval_covost2-en-de_loss": 3.547793388366699,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 79.8111,
"eval_covost2-en-de_samples_per_second": 0.802,
"eval_covost2-en-de_steps_per_second": 0.05,
"step": 5040
},
{
"epoch": 0.1749939238220895,
"eval_covost2-zh-en_loss": 4.870274543762207,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 75.3371,
"eval_covost2-zh-en_samples_per_second": 0.85,
"eval_covost2-zh-en_steps_per_second": 0.053,
"step": 5040
},
{
"epoch": 0.1749939238220895,
"eval_peoplespeech-clean-transcription_loss": 4.296048164367676,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 78.9834,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.81,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.051,
"step": 5040
},
{
"epoch": 0.17707718481997153,
"grad_norm": 0.068359375,
"learning_rate": 0.0016154311019928736,
"loss": 0.0775,
"step": 5100
},
{
"epoch": 0.18054928648310822,
"grad_norm": 0.0517578125,
"learning_rate": 0.0015979938099791489,
"loss": 0.0748,
"step": 5200
},
{
"epoch": 0.1840213881462449,
"grad_norm": 0.049072265625,
"learning_rate": 0.0015802828057370204,
"loss": 0.076,
"step": 5300
},
{
"epoch": 0.18749348980938163,
"grad_norm": 0.047119140625,
"learning_rate": 0.0015623078237618606,
"loss": 0.0773,
"step": 5400
},
{
"epoch": 0.19096559147251832,
"grad_norm": 0.044921875,
"learning_rate": 0.0015440787436390635,
"loss": 0.0764,
"step": 5500
},
{
"epoch": 0.194437693135655,
"grad_norm": 0.046630859375,
"learning_rate": 0.001525605584613934,
"loss": 0.0773,
"step": 5600
},
{
"epoch": 0.1979097947987917,
"grad_norm": 0.050048828125,
"learning_rate": 0.00150689850008481,
"loss": 0.0764,
"step": 5700
},
{
"epoch": 0.19999305579667373,
"eval_covost2-en-de_loss": 3.5351662635803223,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 78.6769,
"eval_covost2-en-de_samples_per_second": 0.813,
"eval_covost2-en-de_steps_per_second": 0.051,
"step": 5760
},
{
"epoch": 0.19999305579667373,
"eval_covost2-zh-en_loss": 4.827414512634277,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 75.8074,
"eval_covost2-zh-en_samples_per_second": 0.844,
"eval_covost2-zh-en_steps_per_second": 0.053,
"step": 5760
},
{
"epoch": 0.19999305579667373,
"eval_peoplespeech-clean-transcription_loss": 4.274168968200684,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 78.7303,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.813,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.051,
"step": 5760
},
{
"epoch": 0.20138189646192842,
"grad_norm": 0.04736328125,
"learning_rate": 0.001487967772022452,
"loss": 0.0763,
"step": 5800
},
{
"epoch": 0.2048539981250651,
"grad_norm": 0.050537109375,
"learning_rate": 0.0014688238053187678,
"loss": 0.0753,
"step": 5900
},
{
"epoch": 0.2083260997882018,
"grad_norm": 0.04443359375,
"learning_rate": 0.0014494771220679712,
"loss": 0.0757,
"step": 6000
},
{
"epoch": 0.21179820145133849,
"grad_norm": 0.047607421875,
"learning_rate": 0.0014299383557833288,
"loss": 0.0757,
"step": 6100
},
{
"epoch": 0.2152703031144752,
"grad_norm": 0.049072265625,
"learning_rate": 0.0014102182455526626,
"loss": 0.0745,
"step": 6200
},
{
"epoch": 0.2187424047776119,
"grad_norm": 0.048583984375,
"learning_rate": 0.0013903276301358268,
"loss": 0.0723,
"step": 6300
},
{
"epoch": 0.22221450644074858,
"grad_norm": 0.058837890625,
"learning_rate": 0.0013702774420074045,
"loss": 0.0751,
"step": 6400
},
{
"epoch": 0.22499218777125793,
"eval_covost2-en-de_loss": 3.531952381134033,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 77.3327,
"eval_covost2-en-de_samples_per_second": 0.828,
"eval_covost2-en-de_steps_per_second": 0.052,
"step": 6480
},
{
"epoch": 0.22499218777125793,
"eval_covost2-zh-en_loss": 4.849607467651367,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 79.3692,
"eval_covost2-zh-en_samples_per_second": 0.806,
"eval_covost2-zh-en_steps_per_second": 0.05,
"step": 6480
},
{
"epoch": 0.22499218777125793,
"eval_peoplespeech-clean-transcription_loss": 4.26890230178833,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 77.8834,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.822,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.051,
"step": 6480
},
{
"epoch": 0.22568660810388527,
"grad_norm": 0.07080078125,
"learning_rate": 0.001350078701347893,
"loss": 0.0756,
"step": 6500
},
{
"epoch": 0.229158709767022,
"grad_norm": 0.047119140625,
"learning_rate": 0.0013297425099866851,
"loss": 0.0744,
"step": 6600
},
{
"epoch": 0.23263081143015868,
"grad_norm": 0.047119140625,
"learning_rate": 0.0013092800453001724,
"loss": 0.0735,
"step": 6700
},
{
"epoch": 0.23610291309329537,
"grad_norm": 0.046875,
"learning_rate": 0.0012887025540683256,
"loss": 0.0742,
"step": 6800
},
{
"epoch": 0.23957501475643206,
"grad_norm": 0.0595703125,
"learning_rate": 0.0012680213462931311,
"loss": 0.0732,
"step": 6900
},
{
"epoch": 0.24304711641956878,
"grad_norm": 0.04345703125,
"learning_rate": 0.0012472477889822766,
"loss": 0.0746,
"step": 7000
},
{
"epoch": 0.24651921808270547,
"grad_norm": 0.049560546875,
"learning_rate": 0.0012263932999015041,
"loss": 0.0732,
"step": 7100
},
{
"epoch": 0.24999131974584216,
"grad_norm": 0.03857421875,
"learning_rate": 0.001205469341299067,
"loss": 0.0734,
"step": 7200
},
{
"epoch": 0.24999131974584216,
"eval_covost2-en-de_loss": 3.52813720703125,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 80.0562,
"eval_covost2-en-de_samples_per_second": 0.799,
"eval_covost2-en-de_steps_per_second": 0.05,
"step": 7200
},
{
"epoch": 0.24999131974584216,
"eval_covost2-zh-en_loss": 4.8250908851623535,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 76.2029,
"eval_covost2-zh-en_samples_per_second": 0.84,
"eval_covost2-zh-en_steps_per_second": 0.052,
"step": 7200
},
{
"epoch": 0.24999131974584216,
"eval_peoplespeech-clean-transcription_loss": 4.264875888824463,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 79.8691,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.801,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.05,
"step": 7200
},
{
"epoch": 0.2534634214089789,
"grad_norm": 0.052734375,
"learning_rate": 0.0011844874136057357,
"loss": 0.0736,
"step": 7300
},
{
"epoch": 0.25693552307211553,
"grad_norm": 0.045654296875,
"learning_rate": 0.0011634590491138206,
"loss": 0.072,
"step": 7400
},
{
"epoch": 0.26040762473525225,
"grad_norm": 0.06494140625,
"learning_rate": 0.0011423958056386782,
"loss": 0.0725,
"step": 7500
},
{
"epoch": 0.26387972639838897,
"grad_norm": 0.044677734375,
"learning_rate": 0.0011213092601661948,
"loss": 0.0724,
"step": 7600
},
{
"epoch": 0.26735182806152563,
"grad_norm": 0.040283203125,
"learning_rate": 0.0011002110024897263,
"loss": 0.0714,
"step": 7700
},
{
"epoch": 0.27082392972466235,
"grad_norm": 0.05908203125,
"learning_rate": 0.0010791126288400054,
"loss": 0.0716,
"step": 7800
},
{
"epoch": 0.274296031387799,
"grad_norm": 0.0556640625,
"learning_rate": 0.001058025735511508,
"loss": 0.0729,
"step": 7900
},
{
"epoch": 0.27499045172042635,
"eval_covost2-en-de_loss": 3.527186393737793,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 79.5291,
"eval_covost2-en-de_samples_per_second": 0.805,
"eval_covost2-en-de_steps_per_second": 0.05,
"step": 7920
},
{
"epoch": 0.27499045172042635,
"eval_covost2-zh-en_loss": 4.821711540222168,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 75.53,
"eval_covost2-zh-en_samples_per_second": 0.847,
"eval_covost2-zh-en_steps_per_second": 0.053,
"step": 7920
},
{
"epoch": 0.27499045172042635,
"eval_peoplespeech-clean-transcription_loss": 4.27138090133667,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 77.949,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.821,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.051,
"step": 7920
},
{
"epoch": 0.27776813305093573,
"grad_norm": 0.062255859375,
"learning_rate": 0.001036961912488782,
"loss": 0.0724,
"step": 8000
},
{
"epoch": 0.28124023471407245,
"grad_norm": 0.050537109375,
"learning_rate": 0.00101593273707625,
"loss": 0.0734,
"step": 8100
},
{
"epoch": 0.2847123363772091,
"grad_norm": 0.05029296875,
"learning_rate": 0.0009949497675349748,
"loss": 0.0729,
"step": 8200
},
{
"epoch": 0.2881844380403458,
"grad_norm": 0.062255859375,
"learning_rate": 0.0009740245367298972,
"loss": 0.0726,
"step": 8300
},
{
"epoch": 0.29165653970348254,
"grad_norm": 0.043701171875,
"learning_rate": 0.0009531685457910284,
"loss": 0.071,
"step": 8400
},
{
"epoch": 0.2951286413666192,
"grad_norm": 0.038330078125,
"learning_rate": 0.0009323932577920875,
"loss": 0.0718,
"step": 8500
},
{
"epoch": 0.2986007430297559,
"grad_norm": 0.04296875,
"learning_rate": 0.0009117100914500547,
"loss": 0.0713,
"step": 8600
},
{
"epoch": 0.2999895836950106,
"eval_covost2-en-de_loss": 3.5240273475646973,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 77.1532,
"eval_covost2-en-de_samples_per_second": 0.83,
"eval_covost2-en-de_steps_per_second": 0.052,
"step": 8640
},
{
"epoch": 0.2999895836950106,
"eval_covost2-zh-en_loss": 4.82330322265625,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 75.675,
"eval_covost2-zh-en_samples_per_second": 0.846,
"eval_covost2-zh-en_steps_per_second": 0.053,
"step": 8640
},
{
"epoch": 0.2999895836950106,
"eval_peoplespeech-clean-transcription_loss": 4.263067245483398,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 78.697,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.813,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.051,
"step": 8640
},
{
"epoch": 0.3020728446928926,
"grad_norm": 0.042724609375,
"learning_rate": 0.0008911304148491035,
"loss": 0.072,
"step": 8700
},
{
"epoch": 0.3055449463560293,
"grad_norm": 0.046142578125,
"learning_rate": 0.0008706655391923645,
"loss": 0.0703,
"step": 8800
},
{
"epoch": 0.309017048019166,
"grad_norm": 0.0625,
"learning_rate": 0.0008503267125849506,
"loss": 0.0713,
"step": 8900
},
{
"epoch": 0.3124891496823027,
"grad_norm": 0.0498046875,
"learning_rate": 0.0008301251138516616,
"loss": 0.0699,
"step": 9000
},
{
"epoch": 0.3159612513454394,
"grad_norm": 0.057861328125,
"learning_rate": 0.0008100718463927737,
"loss": 0.072,
"step": 9100
},
{
"epoch": 0.3194333530085761,
"grad_norm": 0.06396484375,
"learning_rate": 0.0007901779320812752,
"loss": 0.0709,
"step": 9200
},
{
"epoch": 0.3229054546717128,
"grad_norm": 0.0400390625,
"learning_rate": 0.0007704543052049202,
"loss": 0.0704,
"step": 9300
},
{
"epoch": 0.3249887156695948,
"eval_covost2-en-de_loss": 3.519291400909424,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 76.4905,
"eval_covost2-en-de_samples_per_second": 0.837,
"eval_covost2-en-de_steps_per_second": 0.052,
"step": 9360
},
{
"epoch": 0.3249887156695948,
"eval_covost2-zh-en_loss": 4.809961318969727,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 74.8452,
"eval_covost2-zh-en_samples_per_second": 0.855,
"eval_covost2-zh-en_steps_per_second": 0.053,
"step": 9360
},
{
"epoch": 0.3249887156695948,
"eval_peoplespeech-clean-transcription_loss": 4.249151706695557,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 79.1882,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.808,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.051,
"step": 9360
},
{
"epoch": 0.3263775563348495,
"grad_norm": 0.06298828125,
"learning_rate": 0.0007509118064564167,
"loss": 0.0698,
"step": 9400
},
{
"epoch": 0.32984965799798616,
"grad_norm": 0.05126953125,
"learning_rate": 0.0007315611769750596,
"loss": 0.0708,
"step": 9500
},
{
"epoch": 0.3333217596611229,
"grad_norm": 0.053955078125,
"learning_rate": 0.0007124130524430808,
"loss": 0.0708,
"step": 9600
},
{
"epoch": 0.3367938613242596,
"grad_norm": 0.03857421875,
"learning_rate": 0.0006934779572399614,
"loss": 0.0711,
"step": 9700
},
{
"epoch": 0.34026596298739625,
"grad_norm": 0.059326171875,
"learning_rate": 0.0006747662986579189,
"loss": 0.0711,
"step": 9800
},
{
"epoch": 0.34373806465053297,
"grad_norm": 0.045654296875,
"learning_rate": 0.0006562883611817502,
"loss": 0.0708,
"step": 9900
},
{
"epoch": 0.3472101663136697,
"grad_norm": 0.049560546875,
"learning_rate": 0.0006380543008361721,
"loss": 0.0715,
"step": 10000
},
{
"epoch": 0.349987847644179,
"eval_covost2-en-de_loss": 3.5199830532073975,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 77.6721,
"eval_covost2-en-de_samples_per_second": 0.824,
"eval_covost2-en-de_steps_per_second": 0.051,
"step": 10080
},
{
"epoch": 0.349987847644179,
"eval_covost2-zh-en_loss": 4.81780481338501,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 75.9304,
"eval_covost2-zh-en_samples_per_second": 0.843,
"eval_covost2-zh-en_steps_per_second": 0.053,
"step": 10080
},
{
"epoch": 0.349987847644179,
"eval_peoplespeech-clean-transcription_loss": 4.261136054992676,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 77.5514,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.825,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.052,
"step": 10080
},
{
"epoch": 0.35068226797680635,
"grad_norm": 0.05712890625,
"learning_rate": 0.000620074139603767,
"loss": 0.0714,
"step": 10100
},
{
"epoch": 0.35415436963994307,
"grad_norm": 0.052734375,
"learning_rate": 0.000602357759916605,
"loss": 0.0721,
"step": 10200
},
{
"epoch": 0.35762647130307973,
"grad_norm": 0.05810546875,
"learning_rate": 0.0005849148992245615,
"loss": 0.0704,
"step": 10300
},
{
"epoch": 0.36109857296621645,
"grad_norm": 0.05078125,
"learning_rate": 0.0005677551446433289,
"loss": 0.0713,
"step": 10400
},
{
"epoch": 0.36457067462935316,
"grad_norm": 0.04638671875,
"learning_rate": 0.0005508879276850472,
"loss": 0.0718,
"step": 10500
},
{
"epoch": 0.3680427762924898,
"grad_norm": 0.041259765625,
"learning_rate": 0.0005343225190744699,
"loss": 0.0694,
"step": 10600
},
{
"epoch": 0.37151487795562654,
"grad_norm": 0.054443359375,
"learning_rate": 0.0005180680236534926,
"loss": 0.0708,
"step": 10700
},
{
"epoch": 0.37498697961876326,
"grad_norm": 0.037841796875,
"learning_rate": 0.0005021333753768642,
"loss": 0.0697,
"step": 10800
},
{
"epoch": 0.37498697961876326,
"eval_covost2-en-de_loss": 3.5141408443450928,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 78.2999,
"eval_covost2-en-de_samples_per_second": 0.817,
"eval_covost2-en-de_steps_per_second": 0.051,
"step": 10800
},
{
"epoch": 0.37498697961876326,
"eval_covost2-zh-en_loss": 4.8156962394714355,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 77.4053,
"eval_covost2-zh-en_samples_per_second": 0.827,
"eval_covost2-zh-en_steps_per_second": 0.052,
"step": 10800
},
{
"epoch": 0.37498697961876326,
"eval_peoplespeech-clean-transcription_loss": 4.252511978149414,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 79.1957,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.808,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.051,
"step": 10800
},
{
"epoch": 0.3784590812818999,
"grad_norm": 0.039794921875,
"learning_rate": 0.0004865273324018164,
"loss": 0.0692,
"step": 10900
},
{
"epoch": 0.38193118294503664,
"grad_norm": 0.041015625,
"learning_rate": 0.0004712584722743229,
"loss": 0.0701,
"step": 11000
},
{
"epoch": 0.3854032846081733,
"grad_norm": 0.042236328125,
"learning_rate": 0.00045633518721462145,
"loss": 0.0695,
"step": 11100
},
{
"epoch": 0.38887538627131,
"grad_norm": 0.064453125,
"learning_rate": 0.0004417656795046017,
"loss": 0.0683,
"step": 11200
},
{
"epoch": 0.39234748793444674,
"grad_norm": 0.04931640625,
"learning_rate": 0.000427557956979587,
"loss": 0.0692,
"step": 11300
},
{
"epoch": 0.3958195895975834,
"grad_norm": 0.04248046875,
"learning_rate": 0.0004137198286269876,
"loss": 0.0698,
"step": 11400
},
{
"epoch": 0.3992916912607201,
"grad_norm": 0.037353515625,
"learning_rate": 0.0004002589002942515,
"loss": 0.0691,
"step": 11500
},
{
"epoch": 0.39998611159334746,
"eval_covost2-en-de_loss": 3.5120811462402344,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 80.7687,
"eval_covost2-en-de_samples_per_second": 0.792,
"eval_covost2-en-de_steps_per_second": 0.05,
"step": 11520
},
{
"epoch": 0.39998611159334746,
"eval_covost2-zh-en_loss": 4.810714244842529,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 77.3127,
"eval_covost2-zh-en_samples_per_second": 0.828,
"eval_covost2-zh-en_steps_per_second": 0.052,
"step": 11520
},
{
"epoch": 0.39998611159334746,
"eval_peoplespeech-clean-transcription_loss": 4.240975379943848,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 83.3783,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.768,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.048,
"step": 11520
},
{
"epoch": 0.40276379292385683,
"grad_norm": 0.041015625,
"learning_rate": 0.00038718257050846064,
"loss": 0.0681,
"step": 11600
},
{
"epoch": 0.4062358945869935,
"grad_norm": 0.04541015625,
"learning_rate": 0.00037449802640988406,
"loss": 0.0693,
"step": 11700
},
{
"epoch": 0.4097079962501302,
"grad_norm": 0.040771484375,
"learning_rate": 0.00036221223980170827,
"loss": 0.0699,
"step": 11800
},
{
"epoch": 0.4131800979132669,
"grad_norm": 0.04931640625,
"learning_rate": 0.00035033196331812883,
"loss": 0.0694,
"step": 11900
},
{
"epoch": 0.4166521995764036,
"grad_norm": 0.049072265625,
"learning_rate": 0.00033886372671289983,
"loss": 0.0694,
"step": 12000
},
{
"epoch": 0.4201243012395403,
"grad_norm": 0.05712890625,
"learning_rate": 0.00032781383327038774,
"loss": 0.0702,
"step": 12100
},
{
"epoch": 0.42359640290267697,
"grad_norm": 0.064453125,
"learning_rate": 0.00031718835634109865,
"loss": 0.068,
"step": 12200
},
{
"epoch": 0.42498524356793166,
"eval_covost2-en-de_loss": 3.511444568634033,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 81.5271,
"eval_covost2-en-de_samples_per_second": 0.785,
"eval_covost2-en-de_steps_per_second": 0.049,
"step": 12240
},
{
"epoch": 0.42498524356793166,
"eval_covost2-zh-en_loss": 4.813705921173096,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 77.2782,
"eval_covost2-zh-en_samples_per_second": 0.828,
"eval_covost2-zh-en_steps_per_second": 0.052,
"step": 12240
},
{
"epoch": 0.42498524356793166,
"eval_peoplespeech-clean-transcription_loss": 4.2477288246154785,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 80.326,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.797,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.05,
"step": 12240
},
{
"epoch": 0.4270685045658137,
"grad_norm": 0.045654296875,
"learning_rate": 0.0003069931360035833,
"loss": 0.0688,
"step": 12300
},
{
"epoch": 0.4305406062289504,
"grad_norm": 0.0439453125,
"learning_rate": 0.0002972337758545569,
"loss": 0.0705,
"step": 12400
},
{
"epoch": 0.43401270789208707,
"grad_norm": 0.046630859375,
"learning_rate": 0.0002879156399289945,
"loss": 0.0702,
"step": 12500
},
{
"epoch": 0.4374848095552238,
"grad_norm": 0.050048828125,
"learning_rate": 0.0002790438497518984,
"loss": 0.0699,
"step": 12600
},
{
"epoch": 0.44095691121836045,
"grad_norm": 0.09716796875,
"learning_rate": 0.0002706232815233549,
"loss": 0.0698,
"step": 12700
},
{
"epoch": 0.44442901288149717,
"grad_norm": 0.05615234375,
"learning_rate": 0.00026265856343843017,
"loss": 0.0691,
"step": 12800
},
{
"epoch": 0.4479011145446339,
"grad_norm": 0.0390625,
"learning_rate": 0.00025515407314337596,
"loss": 0.0701,
"step": 12900
},
{
"epoch": 0.44998437554251586,
"eval_covost2-en-de_loss": 3.5126776695251465,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 82.6713,
"eval_covost2-en-de_samples_per_second": 0.774,
"eval_covost2-en-de_steps_per_second": 0.048,
"step": 12960
},
{
"epoch": 0.44998437554251586,
"eval_covost2-zh-en_loss": 4.814241409301758,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 77.3656,
"eval_covost2-zh-en_samples_per_second": 0.827,
"eval_covost2-zh-en_steps_per_second": 0.052,
"step": 12960
},
{
"epoch": 0.44998437554251586,
"eval_peoplespeech-clean-transcription_loss": 4.253605365753174,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 83.1111,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.77,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.048,
"step": 12960
},
{
"epoch": 0.45137321620777054,
"grad_norm": 0.050537109375,
"learning_rate": 0.000248113935329545,
"loss": 0.0683,
"step": 13000
},
{
"epoch": 0.45484531787090726,
"grad_norm": 0.057861328125,
"learning_rate": 0.00024154201946633753,
"loss": 0.0693,
"step": 13100
},
{
"epoch": 0.458317419534044,
"grad_norm": 0.049072265625,
"learning_rate": 0.0002354419376744259,
"loss": 0.0687,
"step": 13200
},
{
"epoch": 0.46178952119718064,
"grad_norm": 0.059814453125,
"learning_rate": 0.00022981704274042592,
"loss": 0.0702,
"step": 13300
},
{
"epoch": 0.46526162286031736,
"grad_norm": 0.052001953125,
"learning_rate": 0.00022467042627410459,
"loss": 0.0696,
"step": 13400
},
{
"epoch": 0.468733724523454,
"grad_norm": 0.055908203125,
"learning_rate": 0.00022000491700914,
"loss": 0.0685,
"step": 13500
},
{
"epoch": 0.47220582618659074,
"grad_norm": 0.042724609375,
"learning_rate": 0.00021582307924836473,
"loss": 0.0693,
"step": 13600
},
{
"epoch": 0.4749835075171001,
"eval_covost2-en-de_loss": 3.5136711597442627,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 82.3494,
"eval_covost2-en-de_samples_per_second": 0.777,
"eval_covost2-en-de_steps_per_second": 0.049,
"step": 13680
},
{
"epoch": 0.4749835075171001,
"eval_covost2-zh-en_loss": 4.807031631469727,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 77.0766,
"eval_covost2-zh-en_samples_per_second": 0.83,
"eval_covost2-zh-en_steps_per_second": 0.052,
"step": 13680
},
{
"epoch": 0.4749835075171001,
"eval_peoplespeech-clean-transcription_loss": 4.247273921966553,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 79.3908,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.806,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.05,
"step": 13680
},
{
"epoch": 0.47567792784972746,
"grad_norm": 0.06787109375,
"learning_rate": 0.00021212721145434866,
"loss": 0.0696,
"step": 13700
},
{
"epoch": 0.4791500295128641,
"grad_norm": 0.06494140625,
"learning_rate": 0.00020891934498609588,
"loss": 0.0686,
"step": 13800
},
{
"epoch": 0.48262213117600083,
"grad_norm": 0.048828125,
"learning_rate": 0.00020620124298254867,
"loss": 0.0689,
"step": 13900
},
{
"epoch": 0.48609423283913755,
"grad_norm": 0.055419921875,
"learning_rate": 0.00020397439939351435,
"loss": 0.0684,
"step": 14000
},
{
"epoch": 0.4895663345022742,
"grad_norm": 0.048583984375,
"learning_rate": 0.00020224003815854565,
"loss": 0.0688,
"step": 14100
},
{
"epoch": 0.49303843616541093,
"grad_norm": 0.03955078125,
"learning_rate": 0.0002009991125342271,
"loss": 0.0701,
"step": 14200
},
{
"epoch": 0.4965105378285476,
"grad_norm": 0.06591796875,
"learning_rate": 0.0002002523045702371,
"loss": 0.0689,
"step": 14300
},
{
"epoch": 0.4999826394916843,
"grad_norm": 0.050537109375,
"learning_rate": 0.000200000024734473,
"loss": 0.0693,
"step": 14400
},
{
"epoch": 0.4999826394916843,
"eval_covost2-en-de_loss": 3.511782169342041,
"eval_covost2-en-de_model_preparation_time": 0.0589,
"eval_covost2-en-de_runtime": 80.0758,
"eval_covost2-en-de_samples_per_second": 0.799,
"eval_covost2-en-de_steps_per_second": 0.05,
"step": 14400
},
{
"epoch": 0.4999826394916843,
"eval_covost2-zh-en_loss": 4.806510925292969,
"eval_covost2-zh-en_model_preparation_time": 0.0589,
"eval_covost2-zh-en_runtime": 77.6827,
"eval_covost2-zh-en_samples_per_second": 0.824,
"eval_covost2-zh-en_steps_per_second": 0.051,
"step": 14400
},
{
"epoch": 0.4999826394916843,
"eval_peoplespeech-clean-transcription_loss": 4.244372367858887,
"eval_peoplespeech-clean-transcription_model_preparation_time": 0.0589,
"eval_peoplespeech-clean-transcription_runtime": 79.8633,
"eval_peoplespeech-clean-transcription_samples_per_second": 0.801,
"eval_peoplespeech-clean-transcription_steps_per_second": 0.05,
"step": 14400
}
],
"logging_steps": 100,
"max_steps": 14400,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 3600,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 7.624630689605878e+17,
"train_batch_size": 48,
"trial_name": null,
"trial_params": null
}