iteboshi-medium / trainer_state.json
kiritan's picture
训练结束,上传最终模型
760d4a8 verified
{
"best_metric": 81.62187647336162,
"best_model_checkpoint": "./iteboshi_student_model_temp/checkpoint-19000",
"epoch": 22.026431718061673,
"eval_steps": 1000,
"global_step": 20000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02753303964757709,
"grad_norm": 4.4843268394470215,
"learning_rate": 1.0000000000000002e-06,
"loss": 11.1906,
"step": 25
},
{
"epoch": 0.05506607929515418,
"grad_norm": 2.329829692840576,
"learning_rate": 2.0000000000000003e-06,
"loss": 10.3963,
"step": 50
},
{
"epoch": 0.08259911894273128,
"grad_norm": 2.1956491470336914,
"learning_rate": 3e-06,
"loss": 8.582,
"step": 75
},
{
"epoch": 0.11013215859030837,
"grad_norm": 1.6026716232299805,
"learning_rate": 4.000000000000001e-06,
"loss": 6.6173,
"step": 100
},
{
"epoch": 0.13766519823788545,
"grad_norm": 2.139051675796509,
"learning_rate": 5e-06,
"loss": 5.5439,
"step": 125
},
{
"epoch": 0.16519823788546256,
"grad_norm": 1.4878082275390625,
"learning_rate": 6e-06,
"loss": 4.606,
"step": 150
},
{
"epoch": 0.19273127753303965,
"grad_norm": 1.5405058860778809,
"learning_rate": 7e-06,
"loss": 4.2556,
"step": 175
},
{
"epoch": 0.22026431718061673,
"grad_norm": 1.7798066139221191,
"learning_rate": 8.000000000000001e-06,
"loss": 4.1084,
"step": 200
},
{
"epoch": 0.24779735682819384,
"grad_norm": 1.461482048034668,
"learning_rate": 9e-06,
"loss": 3.9714,
"step": 225
},
{
"epoch": 0.2753303964757709,
"grad_norm": 1.7320806980133057,
"learning_rate": 1e-05,
"loss": 3.8459,
"step": 250
},
{
"epoch": 0.30286343612334804,
"grad_norm": 1.756719708442688,
"learning_rate": 1.1000000000000001e-05,
"loss": 3.7018,
"step": 275
},
{
"epoch": 0.3303964757709251,
"grad_norm": 1.5017330646514893,
"learning_rate": 1.2e-05,
"loss": 3.5979,
"step": 300
},
{
"epoch": 0.3579295154185022,
"grad_norm": 1.5448871850967407,
"learning_rate": 1.3000000000000001e-05,
"loss": 3.495,
"step": 325
},
{
"epoch": 0.3854625550660793,
"grad_norm": 1.4887356758117676,
"learning_rate": 1.4e-05,
"loss": 3.4151,
"step": 350
},
{
"epoch": 0.4129955947136564,
"grad_norm": 1.64043390750885,
"learning_rate": 1.5000000000000002e-05,
"loss": 3.3033,
"step": 375
},
{
"epoch": 0.44052863436123346,
"grad_norm": 1.9127826690673828,
"learning_rate": 1.6000000000000003e-05,
"loss": 3.134,
"step": 400
},
{
"epoch": 0.46806167400881055,
"grad_norm": 1.5309818983078003,
"learning_rate": 1.7e-05,
"loss": 3.0402,
"step": 425
},
{
"epoch": 0.4955947136563877,
"grad_norm": 1.752582311630249,
"learning_rate": 1.8e-05,
"loss": 2.9363,
"step": 450
},
{
"epoch": 0.5231277533039648,
"grad_norm": 1.5565321445465088,
"learning_rate": 1.9e-05,
"loss": 2.7852,
"step": 475
},
{
"epoch": 0.5506607929515418,
"grad_norm": 1.5152559280395508,
"learning_rate": 2e-05,
"loss": 2.587,
"step": 500
},
{
"epoch": 0.5781938325991189,
"grad_norm": 1.79190993309021,
"learning_rate": 1.9974358974358975e-05,
"loss": 2.46,
"step": 525
},
{
"epoch": 0.6057268722466961,
"grad_norm": 1.7385997772216797,
"learning_rate": 1.994871794871795e-05,
"loss": 2.2762,
"step": 550
},
{
"epoch": 0.6332599118942731,
"grad_norm": 1.6728637218475342,
"learning_rate": 1.9923076923076926e-05,
"loss": 2.1675,
"step": 575
},
{
"epoch": 0.6607929515418502,
"grad_norm": 1.6606602668762207,
"learning_rate": 1.98974358974359e-05,
"loss": 2.0062,
"step": 600
},
{
"epoch": 0.6883259911894273,
"grad_norm": 1.6768505573272705,
"learning_rate": 1.9871794871794873e-05,
"loss": 1.9047,
"step": 625
},
{
"epoch": 0.7158590308370044,
"grad_norm": 2.253502130508423,
"learning_rate": 1.9846153846153847e-05,
"loss": 1.8625,
"step": 650
},
{
"epoch": 0.7433920704845814,
"grad_norm": 2.384761333465576,
"learning_rate": 1.9820512820512824e-05,
"loss": 1.7707,
"step": 675
},
{
"epoch": 0.7709251101321586,
"grad_norm": 1.652563452720642,
"learning_rate": 1.9794871794871798e-05,
"loss": 1.6815,
"step": 700
},
{
"epoch": 0.7984581497797357,
"grad_norm": 2.095320224761963,
"learning_rate": 1.976923076923077e-05,
"loss": 1.6741,
"step": 725
},
{
"epoch": 0.8259911894273128,
"grad_norm": 1.6573553085327148,
"learning_rate": 1.9743589743589745e-05,
"loss": 1.6342,
"step": 750
},
{
"epoch": 0.8535242290748899,
"grad_norm": 2.439465045928955,
"learning_rate": 1.9717948717948722e-05,
"loss": 1.5066,
"step": 775
},
{
"epoch": 0.8810572687224669,
"grad_norm": 2.218928575515747,
"learning_rate": 1.9692307692307696e-05,
"loss": 1.508,
"step": 800
},
{
"epoch": 0.9085903083700441,
"grad_norm": 1.4493534564971924,
"learning_rate": 1.9666666666666666e-05,
"loss": 1.3678,
"step": 825
},
{
"epoch": 0.9361233480176211,
"grad_norm": 1.3857249021530151,
"learning_rate": 1.9641025641025643e-05,
"loss": 1.372,
"step": 850
},
{
"epoch": 0.9636563876651982,
"grad_norm": 1.2312829494476318,
"learning_rate": 1.9615384615384617e-05,
"loss": 1.3583,
"step": 875
},
{
"epoch": 0.9911894273127754,
"grad_norm": 1.7964283227920532,
"learning_rate": 1.958974358974359e-05,
"loss": 1.3459,
"step": 900
},
{
"epoch": 1.0187224669603525,
"grad_norm": 1.721659779548645,
"learning_rate": 1.9564102564102564e-05,
"loss": 1.2279,
"step": 925
},
{
"epoch": 1.0462555066079295,
"grad_norm": 1.5802332162857056,
"learning_rate": 1.953846153846154e-05,
"loss": 1.1097,
"step": 950
},
{
"epoch": 1.0737885462555066,
"grad_norm": 1.7199921607971191,
"learning_rate": 1.9512820512820515e-05,
"loss": 1.1012,
"step": 975
},
{
"epoch": 1.1013215859030836,
"grad_norm": 1.5837390422821045,
"learning_rate": 1.9487179487179488e-05,
"loss": 1.0877,
"step": 1000
},
{
"epoch": 1.1013215859030836,
"eval_cer": 54.04090195224285,
"eval_loss": 1.2933422327041626,
"eval_runtime": 1360.2269,
"eval_samples_per_second": 7.779,
"eval_steps_per_second": 1.945,
"eval_wer": 98.37812352663839,
"step": 1000
},
{
"epoch": 1.1288546255506609,
"grad_norm": 1.3139415979385376,
"learning_rate": 1.9461538461538462e-05,
"loss": 1.0589,
"step": 1025
},
{
"epoch": 1.1563876651982379,
"grad_norm": 1.2565332651138306,
"learning_rate": 1.943589743589744e-05,
"loss": 1.0856,
"step": 1050
},
{
"epoch": 1.183920704845815,
"grad_norm": 1.5922749042510986,
"learning_rate": 1.9410256410256413e-05,
"loss": 1.0407,
"step": 1075
},
{
"epoch": 1.2114537444933922,
"grad_norm": 1.555267572402954,
"learning_rate": 1.9384615384615386e-05,
"loss": 1.0012,
"step": 1100
},
{
"epoch": 1.2389867841409692,
"grad_norm": 1.5344181060791016,
"learning_rate": 1.935897435897436e-05,
"loss": 1.0704,
"step": 1125
},
{
"epoch": 1.2665198237885462,
"grad_norm": 1.516708493232727,
"learning_rate": 1.9333333333333333e-05,
"loss": 0.9818,
"step": 1150
},
{
"epoch": 1.2940528634361232,
"grad_norm": 1.461151123046875,
"learning_rate": 1.930769230769231e-05,
"loss": 0.9635,
"step": 1175
},
{
"epoch": 1.3215859030837005,
"grad_norm": 1.800354242324829,
"learning_rate": 1.9282051282051284e-05,
"loss": 0.9683,
"step": 1200
},
{
"epoch": 1.3491189427312775,
"grad_norm": 1.3849306106567383,
"learning_rate": 1.9256410256410258e-05,
"loss": 0.9675,
"step": 1225
},
{
"epoch": 1.3766519823788546,
"grad_norm": 1.376198172569275,
"learning_rate": 1.923076923076923e-05,
"loss": 0.986,
"step": 1250
},
{
"epoch": 1.4041850220264318,
"grad_norm": 3.222456693649292,
"learning_rate": 1.920512820512821e-05,
"loss": 0.9326,
"step": 1275
},
{
"epoch": 1.4317180616740088,
"grad_norm": 1.2251626253128052,
"learning_rate": 1.9179487179487182e-05,
"loss": 0.915,
"step": 1300
},
{
"epoch": 1.4592511013215859,
"grad_norm": 1.2199984788894653,
"learning_rate": 1.9153846153846156e-05,
"loss": 0.9763,
"step": 1325
},
{
"epoch": 1.4867841409691631,
"grad_norm": 1.6957029104232788,
"learning_rate": 1.912820512820513e-05,
"loss": 0.9419,
"step": 1350
},
{
"epoch": 1.51431718061674,
"grad_norm": 1.2903211116790771,
"learning_rate": 1.9102564102564106e-05,
"loss": 0.9087,
"step": 1375
},
{
"epoch": 1.5418502202643172,
"grad_norm": 1.2986080646514893,
"learning_rate": 1.907692307692308e-05,
"loss": 0.9325,
"step": 1400
},
{
"epoch": 1.5693832599118944,
"grad_norm": 1.4618104696273804,
"learning_rate": 1.905128205128205e-05,
"loss": 0.9184,
"step": 1425
},
{
"epoch": 1.5969162995594712,
"grad_norm": 2.7148098945617676,
"learning_rate": 1.9025641025641027e-05,
"loss": 0.9187,
"step": 1450
},
{
"epoch": 1.6244493392070485,
"grad_norm": 1.289120078086853,
"learning_rate": 1.9e-05,
"loss": 0.8714,
"step": 1475
},
{
"epoch": 1.6519823788546255,
"grad_norm": 1.3392589092254639,
"learning_rate": 1.8974358974358975e-05,
"loss": 0.8392,
"step": 1500
},
{
"epoch": 1.6795154185022025,
"grad_norm": 1.1385656595230103,
"learning_rate": 1.894871794871795e-05,
"loss": 0.8515,
"step": 1525
},
{
"epoch": 1.7070484581497798,
"grad_norm": 1.4275093078613281,
"learning_rate": 1.8923076923076925e-05,
"loss": 0.8418,
"step": 1550
},
{
"epoch": 1.7345814977973568,
"grad_norm": 1.6075072288513184,
"learning_rate": 1.88974358974359e-05,
"loss": 0.8379,
"step": 1575
},
{
"epoch": 1.7621145374449338,
"grad_norm": 1.2246577739715576,
"learning_rate": 1.8871794871794873e-05,
"loss": 0.8321,
"step": 1600
},
{
"epoch": 1.789647577092511,
"grad_norm": 1.1842765808105469,
"learning_rate": 1.8846153846153846e-05,
"loss": 0.8636,
"step": 1625
},
{
"epoch": 1.8171806167400881,
"grad_norm": 2.2065274715423584,
"learning_rate": 1.8820512820512823e-05,
"loss": 0.8869,
"step": 1650
},
{
"epoch": 1.8447136563876652,
"grad_norm": 1.0761529207229614,
"learning_rate": 1.8794871794871797e-05,
"loss": 0.8663,
"step": 1675
},
{
"epoch": 1.8722466960352424,
"grad_norm": 1.3553169965744019,
"learning_rate": 1.876923076923077e-05,
"loss": 0.8197,
"step": 1700
},
{
"epoch": 1.8997797356828194,
"grad_norm": 1.3320152759552002,
"learning_rate": 1.8743589743589744e-05,
"loss": 0.8146,
"step": 1725
},
{
"epoch": 1.9273127753303965,
"grad_norm": 1.7702279090881348,
"learning_rate": 1.8717948717948718e-05,
"loss": 0.8255,
"step": 1750
},
{
"epoch": 1.9548458149779737,
"grad_norm": 1.497201681137085,
"learning_rate": 1.8692307692307695e-05,
"loss": 0.81,
"step": 1775
},
{
"epoch": 1.9823788546255505,
"grad_norm": 1.082921028137207,
"learning_rate": 1.866666666666667e-05,
"loss": 0.7969,
"step": 1800
},
{
"epoch": 2.0099118942731278,
"grad_norm": 0.8988214135169983,
"learning_rate": 1.8641025641025642e-05,
"loss": 0.7524,
"step": 1825
},
{
"epoch": 2.037444933920705,
"grad_norm": 1.0440396070480347,
"learning_rate": 1.8615384615384616e-05,
"loss": 0.5941,
"step": 1850
},
{
"epoch": 2.064977973568282,
"grad_norm": 1.1343231201171875,
"learning_rate": 1.8589743589743593e-05,
"loss": 0.5931,
"step": 1875
},
{
"epoch": 2.092511013215859,
"grad_norm": 0.8273594379425049,
"learning_rate": 1.8564102564102567e-05,
"loss": 0.5839,
"step": 1900
},
{
"epoch": 2.1200440528634363,
"grad_norm": 1.1608532667160034,
"learning_rate": 1.853846153846154e-05,
"loss": 0.6335,
"step": 1925
},
{
"epoch": 2.147577092511013,
"grad_norm": 1.0717401504516602,
"learning_rate": 1.8512820512820514e-05,
"loss": 0.5711,
"step": 1950
},
{
"epoch": 2.1751101321585904,
"grad_norm": 1.2120345830917358,
"learning_rate": 1.848717948717949e-05,
"loss": 0.5823,
"step": 1975
},
{
"epoch": 2.202643171806167,
"grad_norm": 0.9634266495704651,
"learning_rate": 1.8461538461538465e-05,
"loss": 0.6047,
"step": 2000
},
{
"epoch": 2.202643171806167,
"eval_cer": 34.19462817970263,
"eval_loss": 0.8662471175193787,
"eval_runtime": 1317.6626,
"eval_samples_per_second": 8.03,
"eval_steps_per_second": 2.008,
"eval_wer": 91.66430928807166,
"step": 2000
},
{
"epoch": 2.2301762114537445,
"grad_norm": 2.5108625888824463,
"learning_rate": 1.8435897435897435e-05,
"loss": 0.5885,
"step": 2025
},
{
"epoch": 2.2577092511013217,
"grad_norm": 1.1990822553634644,
"learning_rate": 1.8410256410256412e-05,
"loss": 0.5913,
"step": 2050
},
{
"epoch": 2.2852422907488985,
"grad_norm": 0.9576376080513,
"learning_rate": 1.8384615384615386e-05,
"loss": 0.5648,
"step": 2075
},
{
"epoch": 2.3127753303964758,
"grad_norm": 2.0012855529785156,
"learning_rate": 1.835897435897436e-05,
"loss": 0.5829,
"step": 2100
},
{
"epoch": 2.340308370044053,
"grad_norm": 1.3808636665344238,
"learning_rate": 1.8333333333333333e-05,
"loss": 0.5258,
"step": 2125
},
{
"epoch": 2.36784140969163,
"grad_norm": 0.9106646776199341,
"learning_rate": 1.830769230769231e-05,
"loss": 0.5541,
"step": 2150
},
{
"epoch": 2.395374449339207,
"grad_norm": 0.9727596640586853,
"learning_rate": 1.8282051282051284e-05,
"loss": 0.5608,
"step": 2175
},
{
"epoch": 2.4229074889867843,
"grad_norm": 1.2104076147079468,
"learning_rate": 1.8256410256410257e-05,
"loss": 0.5298,
"step": 2200
},
{
"epoch": 2.450440528634361,
"grad_norm": 0.9904928803443909,
"learning_rate": 1.823076923076923e-05,
"loss": 0.5236,
"step": 2225
},
{
"epoch": 2.4779735682819384,
"grad_norm": 1.0200227499008179,
"learning_rate": 1.8205128205128208e-05,
"loss": 0.556,
"step": 2250
},
{
"epoch": 2.505506607929515,
"grad_norm": 1.2327479124069214,
"learning_rate": 1.817948717948718e-05,
"loss": 0.6067,
"step": 2275
},
{
"epoch": 2.5330396475770924,
"grad_norm": 0.8597280383110046,
"learning_rate": 1.8153846153846155e-05,
"loss": 0.5704,
"step": 2300
},
{
"epoch": 2.5605726872246697,
"grad_norm": 1.0720624923706055,
"learning_rate": 1.812820512820513e-05,
"loss": 0.5499,
"step": 2325
},
{
"epoch": 2.5881057268722465,
"grad_norm": 0.8711467981338501,
"learning_rate": 1.8102564102564102e-05,
"loss": 0.5539,
"step": 2350
},
{
"epoch": 2.6156387665198237,
"grad_norm": 0.9706287384033203,
"learning_rate": 1.807692307692308e-05,
"loss": 0.5647,
"step": 2375
},
{
"epoch": 2.643171806167401,
"grad_norm": 0.9005234241485596,
"learning_rate": 1.8051282051282053e-05,
"loss": 0.5261,
"step": 2400
},
{
"epoch": 2.670704845814978,
"grad_norm": 0.9805070161819458,
"learning_rate": 1.8025641025641027e-05,
"loss": 0.5165,
"step": 2425
},
{
"epoch": 2.698237885462555,
"grad_norm": 1.09579598903656,
"learning_rate": 1.8e-05,
"loss": 0.5761,
"step": 2450
},
{
"epoch": 2.7257709251101323,
"grad_norm": 1.0162299871444702,
"learning_rate": 1.7974358974358977e-05,
"loss": 0.5228,
"step": 2475
},
{
"epoch": 2.753303964757709,
"grad_norm": 1.1346451044082642,
"learning_rate": 1.794871794871795e-05,
"loss": 0.5937,
"step": 2500
},
{
"epoch": 2.7808370044052864,
"grad_norm": 1.3441213369369507,
"learning_rate": 1.7923076923076925e-05,
"loss": 0.5553,
"step": 2525
},
{
"epoch": 2.8083700440528636,
"grad_norm": 1.1649460792541504,
"learning_rate": 1.78974358974359e-05,
"loss": 0.507,
"step": 2550
},
{
"epoch": 2.8359030837004404,
"grad_norm": 0.9251694083213806,
"learning_rate": 1.7871794871794875e-05,
"loss": 0.5293,
"step": 2575
},
{
"epoch": 2.8634361233480177,
"grad_norm": 1.4110281467437744,
"learning_rate": 1.784615384615385e-05,
"loss": 0.5654,
"step": 2600
},
{
"epoch": 2.890969162995595,
"grad_norm": 1.63406240940094,
"learning_rate": 1.7820512820512823e-05,
"loss": 0.5154,
"step": 2625
},
{
"epoch": 2.9185022026431717,
"grad_norm": 1.6731914281845093,
"learning_rate": 1.7794871794871796e-05,
"loss": 0.5209,
"step": 2650
},
{
"epoch": 2.946035242290749,
"grad_norm": 0.8957790732383728,
"learning_rate": 1.776923076923077e-05,
"loss": 0.5275,
"step": 2675
},
{
"epoch": 2.9735682819383262,
"grad_norm": 1.0544830560684204,
"learning_rate": 1.7743589743589744e-05,
"loss": 0.5149,
"step": 2700
},
{
"epoch": 3.001101321585903,
"grad_norm": 1.1973352432250977,
"learning_rate": 1.7717948717948717e-05,
"loss": 0.5632,
"step": 2725
},
{
"epoch": 3.0286343612334803,
"grad_norm": 0.9459331035614014,
"learning_rate": 1.7692307692307694e-05,
"loss": 0.3906,
"step": 2750
},
{
"epoch": 3.056167400881057,
"grad_norm": 0.7281056642532349,
"learning_rate": 1.7666666666666668e-05,
"loss": 0.3398,
"step": 2775
},
{
"epoch": 3.0837004405286343,
"grad_norm": 0.9286469221115112,
"learning_rate": 1.7641025641025642e-05,
"loss": 0.3603,
"step": 2800
},
{
"epoch": 3.1112334801762116,
"grad_norm": 0.6543312072753906,
"learning_rate": 1.7615384615384615e-05,
"loss": 0.348,
"step": 2825
},
{
"epoch": 3.1387665198237884,
"grad_norm": 0.852942168712616,
"learning_rate": 1.7589743589743592e-05,
"loss": 0.3527,
"step": 2850
},
{
"epoch": 3.1662995594713657,
"grad_norm": 1.0175626277923584,
"learning_rate": 1.7564102564102566e-05,
"loss": 0.3779,
"step": 2875
},
{
"epoch": 3.193832599118943,
"grad_norm": 0.9942947030067444,
"learning_rate": 1.753846153846154e-05,
"loss": 0.3732,
"step": 2900
},
{
"epoch": 3.2213656387665197,
"grad_norm": 0.713638186454773,
"learning_rate": 1.7512820512820513e-05,
"loss": 0.3757,
"step": 2925
},
{
"epoch": 3.248898678414097,
"grad_norm": 0.9049544334411621,
"learning_rate": 1.7487179487179487e-05,
"loss": 0.3769,
"step": 2950
},
{
"epoch": 3.2764317180616738,
"grad_norm": 0.7445310354232788,
"learning_rate": 1.7461538461538464e-05,
"loss": 0.361,
"step": 2975
},
{
"epoch": 3.303964757709251,
"grad_norm": 0.6536186933517456,
"learning_rate": 1.7435897435897438e-05,
"loss": 0.3394,
"step": 3000
},
{
"epoch": 3.303964757709251,
"eval_cer": 34.55343734595853,
"eval_loss": 0.7762519717216492,
"eval_runtime": 1343.0708,
"eval_samples_per_second": 7.878,
"eval_steps_per_second": 1.97,
"eval_wer": 88.5997171145686,
"step": 3000
},
{
"epoch": 3.3314977973568283,
"grad_norm": 0.820868194103241,
"learning_rate": 1.741025641025641e-05,
"loss": 0.3648,
"step": 3025
},
{
"epoch": 3.359030837004405,
"grad_norm": 0.8310422897338867,
"learning_rate": 1.7384615384615385e-05,
"loss": 0.3907,
"step": 3050
},
{
"epoch": 3.3865638766519823,
"grad_norm": 0.9818564057350159,
"learning_rate": 1.7358974358974362e-05,
"loss": 0.3515,
"step": 3075
},
{
"epoch": 3.4140969162995596,
"grad_norm": 0.8036455512046814,
"learning_rate": 1.7333333333333336e-05,
"loss": 0.4122,
"step": 3100
},
{
"epoch": 3.4416299559471364,
"grad_norm": 0.8108682632446289,
"learning_rate": 1.730769230769231e-05,
"loss": 0.3818,
"step": 3125
},
{
"epoch": 3.4691629955947136,
"grad_norm": 1.3358936309814453,
"learning_rate": 1.7282051282051283e-05,
"loss": 0.3533,
"step": 3150
},
{
"epoch": 3.496696035242291,
"grad_norm": 0.8960527777671814,
"learning_rate": 1.725641025641026e-05,
"loss": 0.3816,
"step": 3175
},
{
"epoch": 3.5242290748898677,
"grad_norm": 2.2398922443389893,
"learning_rate": 1.7230769230769234e-05,
"loss": 0.3855,
"step": 3200
},
{
"epoch": 3.551762114537445,
"grad_norm": 0.8993757367134094,
"learning_rate": 1.7205128205128207e-05,
"loss": 0.3503,
"step": 3225
},
{
"epoch": 3.579295154185022,
"grad_norm": 0.8796055912971497,
"learning_rate": 1.717948717948718e-05,
"loss": 0.3479,
"step": 3250
},
{
"epoch": 3.606828193832599,
"grad_norm": 0.7127211093902588,
"learning_rate": 1.7153846153846155e-05,
"loss": 0.3803,
"step": 3275
},
{
"epoch": 3.6343612334801763,
"grad_norm": 0.9272491335868835,
"learning_rate": 1.7128205128205128e-05,
"loss": 0.3819,
"step": 3300
},
{
"epoch": 3.6618942731277535,
"grad_norm": 0.6483042240142822,
"learning_rate": 1.7102564102564102e-05,
"loss": 0.3236,
"step": 3325
},
{
"epoch": 3.6894273127753303,
"grad_norm": 0.8154418468475342,
"learning_rate": 1.707692307692308e-05,
"loss": 0.3622,
"step": 3350
},
{
"epoch": 3.7169603524229076,
"grad_norm": 0.9721470475196838,
"learning_rate": 1.7051282051282053e-05,
"loss": 0.3215,
"step": 3375
},
{
"epoch": 3.744493392070485,
"grad_norm": 0.7829445004463196,
"learning_rate": 1.7025641025641026e-05,
"loss": 0.3652,
"step": 3400
},
{
"epoch": 3.7720264317180616,
"grad_norm": 0.8178221583366394,
"learning_rate": 1.7e-05,
"loss": 0.3522,
"step": 3425
},
{
"epoch": 3.799559471365639,
"grad_norm": 0.9240100979804993,
"learning_rate": 1.6974358974358977e-05,
"loss": 0.3553,
"step": 3450
},
{
"epoch": 3.827092511013216,
"grad_norm": 0.7727634906768799,
"learning_rate": 1.694871794871795e-05,
"loss": 0.3343,
"step": 3475
},
{
"epoch": 3.854625550660793,
"grad_norm": 0.8703585863113403,
"learning_rate": 1.6923076923076924e-05,
"loss": 0.4124,
"step": 3500
},
{
"epoch": 3.88215859030837,
"grad_norm": 0.6991782188415527,
"learning_rate": 1.6897435897435898e-05,
"loss": 0.3452,
"step": 3525
},
{
"epoch": 3.909691629955947,
"grad_norm": 0.8196091651916504,
"learning_rate": 1.687179487179487e-05,
"loss": 0.3681,
"step": 3550
},
{
"epoch": 3.9372246696035242,
"grad_norm": 0.7455742955207825,
"learning_rate": 1.684615384615385e-05,
"loss": 0.3963,
"step": 3575
},
{
"epoch": 3.964757709251101,
"grad_norm": 1.1615126132965088,
"learning_rate": 1.6820512820512822e-05,
"loss": 0.3811,
"step": 3600
},
{
"epoch": 3.9922907488986783,
"grad_norm": 1.1333736181259155,
"learning_rate": 1.6794871794871796e-05,
"loss": 0.3724,
"step": 3625
},
{
"epoch": 4.0198237885462555,
"grad_norm": 0.9983144998550415,
"learning_rate": 1.676923076923077e-05,
"loss": 0.2523,
"step": 3650
},
{
"epoch": 4.047356828193832,
"grad_norm": 0.8323332667350769,
"learning_rate": 1.6743589743589747e-05,
"loss": 0.24,
"step": 3675
},
{
"epoch": 4.07488986784141,
"grad_norm": 0.9976752400398254,
"learning_rate": 1.671794871794872e-05,
"loss": 0.2255,
"step": 3700
},
{
"epoch": 4.102422907488987,
"grad_norm": 0.56157386302948,
"learning_rate": 1.6692307692307694e-05,
"loss": 0.2547,
"step": 3725
},
{
"epoch": 4.129955947136564,
"grad_norm": 0.7160229086875916,
"learning_rate": 1.6666666666666667e-05,
"loss": 0.2373,
"step": 3750
},
{
"epoch": 4.157488986784141,
"grad_norm": 0.88517165184021,
"learning_rate": 1.6641025641025645e-05,
"loss": 0.2279,
"step": 3775
},
{
"epoch": 4.185022026431718,
"grad_norm": 0.5281697511672974,
"learning_rate": 1.6615384615384618e-05,
"loss": 0.1878,
"step": 3800
},
{
"epoch": 4.212555066079295,
"grad_norm": 0.5709527134895325,
"learning_rate": 1.6589743589743592e-05,
"loss": 0.2223,
"step": 3825
},
{
"epoch": 4.240088105726873,
"grad_norm": 0.8601750135421753,
"learning_rate": 1.6564102564102565e-05,
"loss": 0.2166,
"step": 3850
},
{
"epoch": 4.2676211453744495,
"grad_norm": 0.5579825043678284,
"learning_rate": 1.653846153846154e-05,
"loss": 0.251,
"step": 3875
},
{
"epoch": 4.295154185022026,
"grad_norm": 0.5097801089286804,
"learning_rate": 1.6512820512820513e-05,
"loss": 0.2342,
"step": 3900
},
{
"epoch": 4.322687224669604,
"grad_norm": 0.6469748020172119,
"learning_rate": 1.6487179487179486e-05,
"loss": 0.2465,
"step": 3925
},
{
"epoch": 4.350220264317181,
"grad_norm": 0.7211927175521851,
"learning_rate": 1.6461538461538463e-05,
"loss": 0.2358,
"step": 3950
},
{
"epoch": 4.377753303964758,
"grad_norm": 0.6948506832122803,
"learning_rate": 1.6435897435897437e-05,
"loss": 0.2603,
"step": 3975
},
{
"epoch": 4.405286343612334,
"grad_norm": 0.7800878882408142,
"learning_rate": 1.641025641025641e-05,
"loss": 0.243,
"step": 4000
},
{
"epoch": 4.405286343612334,
"eval_cer": 26.809808214781107,
"eval_loss": 0.7649882435798645,
"eval_runtime": 1301.1909,
"eval_samples_per_second": 8.132,
"eval_steps_per_second": 2.034,
"eval_wer": 86.5912305516266,
"step": 4000
},
{
"epoch": 4.432819383259912,
"grad_norm": 0.6358670592308044,
"learning_rate": 1.6384615384615384e-05,
"loss": 0.2619,
"step": 4025
},
{
"epoch": 4.460352422907489,
"grad_norm": 0.7865478992462158,
"learning_rate": 1.635897435897436e-05,
"loss": 0.2671,
"step": 4050
},
{
"epoch": 4.487885462555066,
"grad_norm": 0.890876054763794,
"learning_rate": 1.6333333333333335e-05,
"loss": 0.2592,
"step": 4075
},
{
"epoch": 4.515418502202643,
"grad_norm": 0.7817333936691284,
"learning_rate": 1.630769230769231e-05,
"loss": 0.2229,
"step": 4100
},
{
"epoch": 4.54295154185022,
"grad_norm": 1.0560652017593384,
"learning_rate": 1.6282051282051282e-05,
"loss": 0.2499,
"step": 4125
},
{
"epoch": 4.570484581497797,
"grad_norm": 0.9233481884002686,
"learning_rate": 1.625641025641026e-05,
"loss": 0.2623,
"step": 4150
},
{
"epoch": 4.598017621145375,
"grad_norm": 0.7783260941505432,
"learning_rate": 1.6230769230769233e-05,
"loss": 0.2484,
"step": 4175
},
{
"epoch": 4.6255506607929515,
"grad_norm": 0.765825629234314,
"learning_rate": 1.6205128205128207e-05,
"loss": 0.2427,
"step": 4200
},
{
"epoch": 4.653083700440528,
"grad_norm": 1.1650230884552002,
"learning_rate": 1.617948717948718e-05,
"loss": 0.2443,
"step": 4225
},
{
"epoch": 4.680616740088106,
"grad_norm": 0.8231481909751892,
"learning_rate": 1.6153846153846154e-05,
"loss": 0.2316,
"step": 4250
},
{
"epoch": 4.708149779735683,
"grad_norm": 0.5373929738998413,
"learning_rate": 1.612820512820513e-05,
"loss": 0.2419,
"step": 4275
},
{
"epoch": 4.73568281938326,
"grad_norm": 0.7449804544448853,
"learning_rate": 1.6102564102564105e-05,
"loss": 0.2484,
"step": 4300
},
{
"epoch": 4.763215859030837,
"grad_norm": 0.889363706111908,
"learning_rate": 1.607692307692308e-05,
"loss": 0.245,
"step": 4325
},
{
"epoch": 4.790748898678414,
"grad_norm": 0.8369229435920715,
"learning_rate": 1.6051282051282052e-05,
"loss": 0.2394,
"step": 4350
},
{
"epoch": 4.818281938325991,
"grad_norm": 0.9110807180404663,
"learning_rate": 1.602564102564103e-05,
"loss": 0.2509,
"step": 4375
},
{
"epoch": 4.845814977973569,
"grad_norm": 0.4285617768764496,
"learning_rate": 1.6000000000000003e-05,
"loss": 0.2244,
"step": 4400
},
{
"epoch": 4.8733480176211454,
"grad_norm": 0.6245120763778687,
"learning_rate": 1.5974358974358976e-05,
"loss": 0.2297,
"step": 4425
},
{
"epoch": 4.900881057268722,
"grad_norm": 0.5603160262107849,
"learning_rate": 1.594871794871795e-05,
"loss": 0.2383,
"step": 4450
},
{
"epoch": 4.9284140969163,
"grad_norm": 1.5242382287979126,
"learning_rate": 1.5923076923076924e-05,
"loss": 0.222,
"step": 4475
},
{
"epoch": 4.955947136563877,
"grad_norm": 0.4880862832069397,
"learning_rate": 1.5897435897435897e-05,
"loss": 0.2412,
"step": 4500
},
{
"epoch": 4.983480176211454,
"grad_norm": 0.8488364815711975,
"learning_rate": 1.587179487179487e-05,
"loss": 0.2459,
"step": 4525
},
{
"epoch": 5.011013215859031,
"grad_norm": 0.6367073059082031,
"learning_rate": 1.5846153846153848e-05,
"loss": 0.2024,
"step": 4550
},
{
"epoch": 5.038546255506608,
"grad_norm": 0.6037558913230896,
"learning_rate": 1.582051282051282e-05,
"loss": 0.183,
"step": 4575
},
{
"epoch": 5.066079295154185,
"grad_norm": 0.6983600854873657,
"learning_rate": 1.5794871794871795e-05,
"loss": 0.1353,
"step": 4600
},
{
"epoch": 5.093612334801762,
"grad_norm": 0.39246147871017456,
"learning_rate": 1.576923076923077e-05,
"loss": 0.1867,
"step": 4625
},
{
"epoch": 5.121145374449339,
"grad_norm": 0.6254175305366516,
"learning_rate": 1.5743589743589746e-05,
"loss": 0.1741,
"step": 4650
},
{
"epoch": 5.148678414096916,
"grad_norm": 0.3912099599838257,
"learning_rate": 1.571794871794872e-05,
"loss": 0.1744,
"step": 4675
},
{
"epoch": 5.176211453744493,
"grad_norm": 0.46106234192848206,
"learning_rate": 1.5692307692307693e-05,
"loss": 0.1437,
"step": 4700
},
{
"epoch": 5.203744493392071,
"grad_norm": 0.7368497252464294,
"learning_rate": 1.5666666666666667e-05,
"loss": 0.1606,
"step": 4725
},
{
"epoch": 5.2312775330396475,
"grad_norm": 0.9352337718009949,
"learning_rate": 1.5641025641025644e-05,
"loss": 0.1395,
"step": 4750
},
{
"epoch": 5.258810572687224,
"grad_norm": 0.785354733467102,
"learning_rate": 1.5615384615384618e-05,
"loss": 0.2019,
"step": 4775
},
{
"epoch": 5.286343612334802,
"grad_norm": 0.511448860168457,
"learning_rate": 1.558974358974359e-05,
"loss": 0.1484,
"step": 4800
},
{
"epoch": 5.313876651982379,
"grad_norm": 0.6866703033447266,
"learning_rate": 1.5564102564102565e-05,
"loss": 0.1413,
"step": 4825
},
{
"epoch": 5.341409691629956,
"grad_norm": 0.5879825949668884,
"learning_rate": 1.553846153846154e-05,
"loss": 0.1611,
"step": 4850
},
{
"epoch": 5.368942731277533,
"grad_norm": 0.5954814553260803,
"learning_rate": 1.5512820512820516e-05,
"loss": 0.1603,
"step": 4875
},
{
"epoch": 5.39647577092511,
"grad_norm": 0.6819676756858826,
"learning_rate": 1.548717948717949e-05,
"loss": 0.1458,
"step": 4900
},
{
"epoch": 5.424008810572687,
"grad_norm": 0.895698070526123,
"learning_rate": 1.5461538461538463e-05,
"loss": 0.2016,
"step": 4925
},
{
"epoch": 5.451541850220265,
"grad_norm": 0.6772642135620117,
"learning_rate": 1.5435897435897436e-05,
"loss": 0.1762,
"step": 4950
},
{
"epoch": 5.479074889867841,
"grad_norm": 0.49580076336860657,
"learning_rate": 1.5410256410256414e-05,
"loss": 0.1563,
"step": 4975
},
{
"epoch": 5.506607929515418,
"grad_norm": 0.5544493198394775,
"learning_rate": 1.5384615384615387e-05,
"loss": 0.1633,
"step": 5000
},
{
"epoch": 5.506607929515418,
"eval_cer": 27.411644675753443,
"eval_loss": 0.7654321789741516,
"eval_runtime": 1299.7877,
"eval_samples_per_second": 8.141,
"eval_steps_per_second": 2.036,
"eval_wer": 88.18481848184818,
"step": 5000
},
{
"epoch": 5.534140969162996,
"grad_norm": 0.5057178139686584,
"learning_rate": 1.535897435897436e-05,
"loss": 0.1553,
"step": 5025
},
{
"epoch": 5.561674008810573,
"grad_norm": 0.6811597347259521,
"learning_rate": 1.5333333333333334e-05,
"loss": 0.186,
"step": 5050
},
{
"epoch": 5.5892070484581495,
"grad_norm": 0.6146376132965088,
"learning_rate": 1.5307692307692308e-05,
"loss": 0.1658,
"step": 5075
},
{
"epoch": 5.616740088105727,
"grad_norm": 0.5837799310684204,
"learning_rate": 1.5282051282051282e-05,
"loss": 0.15,
"step": 5100
},
{
"epoch": 5.644273127753304,
"grad_norm": 0.8449403643608093,
"learning_rate": 1.5256410256410257e-05,
"loss": 0.1698,
"step": 5125
},
{
"epoch": 5.671806167400881,
"grad_norm": 0.5833331346511841,
"learning_rate": 1.523076923076923e-05,
"loss": 0.175,
"step": 5150
},
{
"epoch": 5.6993392070484585,
"grad_norm": 0.4615674912929535,
"learning_rate": 1.5205128205128206e-05,
"loss": 0.1374,
"step": 5175
},
{
"epoch": 5.726872246696035,
"grad_norm": 0.5180462598800659,
"learning_rate": 1.517948717948718e-05,
"loss": 0.2045,
"step": 5200
},
{
"epoch": 5.754405286343612,
"grad_norm": 0.4603562653064728,
"learning_rate": 1.5153846153846155e-05,
"loss": 0.1595,
"step": 5225
},
{
"epoch": 5.78193832599119,
"grad_norm": 0.7777943015098572,
"learning_rate": 1.5128205128205129e-05,
"loss": 0.1954,
"step": 5250
},
{
"epoch": 5.809471365638767,
"grad_norm": 0.4417853355407715,
"learning_rate": 1.5102564102564104e-05,
"loss": 0.1665,
"step": 5275
},
{
"epoch": 5.8370044052863435,
"grad_norm": 0.6847853660583496,
"learning_rate": 1.5076923076923078e-05,
"loss": 0.1629,
"step": 5300
},
{
"epoch": 5.864537444933921,
"grad_norm": 0.6507083177566528,
"learning_rate": 1.5051282051282053e-05,
"loss": 0.1677,
"step": 5325
},
{
"epoch": 5.892070484581498,
"grad_norm": 0.6313048005104065,
"learning_rate": 1.5025641025641027e-05,
"loss": 0.1863,
"step": 5350
},
{
"epoch": 5.919603524229075,
"grad_norm": 0.5796169638633728,
"learning_rate": 1.5000000000000002e-05,
"loss": 0.1497,
"step": 5375
},
{
"epoch": 5.9471365638766525,
"grad_norm": 0.6254355907440186,
"learning_rate": 1.4974358974358976e-05,
"loss": 0.168,
"step": 5400
},
{
"epoch": 5.974669603524229,
"grad_norm": 0.742472767829895,
"learning_rate": 1.494871794871795e-05,
"loss": 0.1474,
"step": 5425
},
{
"epoch": 6.002202643171806,
"grad_norm": 0.45657020807266235,
"learning_rate": 1.4923076923076925e-05,
"loss": 0.1376,
"step": 5450
},
{
"epoch": 6.029735682819383,
"grad_norm": 0.49746203422546387,
"learning_rate": 1.4897435897435898e-05,
"loss": 0.1143,
"step": 5475
},
{
"epoch": 6.057268722466961,
"grad_norm": 0.41658642888069153,
"learning_rate": 1.4871794871794874e-05,
"loss": 0.0941,
"step": 5500
},
{
"epoch": 6.084801762114537,
"grad_norm": 0.5831480026245117,
"learning_rate": 1.4846153846153847e-05,
"loss": 0.1195,
"step": 5525
},
{
"epoch": 6.112334801762114,
"grad_norm": 0.4936138391494751,
"learning_rate": 1.4820512820512823e-05,
"loss": 0.1244,
"step": 5550
},
{
"epoch": 6.139867841409692,
"grad_norm": 0.5034681558609009,
"learning_rate": 1.4794871794871796e-05,
"loss": 0.1133,
"step": 5575
},
{
"epoch": 6.167400881057269,
"grad_norm": 0.6369169354438782,
"learning_rate": 1.4769230769230772e-05,
"loss": 0.1292,
"step": 5600
},
{
"epoch": 6.1949339207048455,
"grad_norm": 0.3793521225452423,
"learning_rate": 1.4743589743589745e-05,
"loss": 0.0927,
"step": 5625
},
{
"epoch": 6.222466960352423,
"grad_norm": 0.6850873827934265,
"learning_rate": 1.471794871794872e-05,
"loss": 0.1249,
"step": 5650
},
{
"epoch": 6.25,
"grad_norm": 0.5474864840507507,
"learning_rate": 1.4692307692307694e-05,
"loss": 0.1206,
"step": 5675
},
{
"epoch": 6.277533039647577,
"grad_norm": 0.2814996838569641,
"learning_rate": 1.4666666666666666e-05,
"loss": 0.0832,
"step": 5700
},
{
"epoch": 6.3050660792951545,
"grad_norm": 0.3831164538860321,
"learning_rate": 1.4641025641025642e-05,
"loss": 0.1137,
"step": 5725
},
{
"epoch": 6.332599118942731,
"grad_norm": 0.48341113328933716,
"learning_rate": 1.4615384615384615e-05,
"loss": 0.096,
"step": 5750
},
{
"epoch": 6.360132158590308,
"grad_norm": 0.4037848711013794,
"learning_rate": 1.458974358974359e-05,
"loss": 0.096,
"step": 5775
},
{
"epoch": 6.387665198237886,
"grad_norm": 0.5997889637947083,
"learning_rate": 1.4564102564102564e-05,
"loss": 0.1143,
"step": 5800
},
{
"epoch": 6.415198237885463,
"grad_norm": 0.6279116272926331,
"learning_rate": 1.453846153846154e-05,
"loss": 0.102,
"step": 5825
},
{
"epoch": 6.442731277533039,
"grad_norm": 0.5346375703811646,
"learning_rate": 1.4512820512820513e-05,
"loss": 0.1336,
"step": 5850
},
{
"epoch": 6.470264317180617,
"grad_norm": 0.5246495008468628,
"learning_rate": 1.4487179487179489e-05,
"loss": 0.1357,
"step": 5875
},
{
"epoch": 6.497797356828194,
"grad_norm": 0.4948612153530121,
"learning_rate": 1.4461538461538462e-05,
"loss": 0.1229,
"step": 5900
},
{
"epoch": 6.525330396475771,
"grad_norm": 0.6798549294471741,
"learning_rate": 1.4435897435897438e-05,
"loss": 0.1319,
"step": 5925
},
{
"epoch": 6.5528634361233475,
"grad_norm": 0.26407966017723083,
"learning_rate": 1.4410256410256411e-05,
"loss": 0.1187,
"step": 5950
},
{
"epoch": 6.580396475770925,
"grad_norm": 0.7821138501167297,
"learning_rate": 1.4384615384615387e-05,
"loss": 0.1218,
"step": 5975
},
{
"epoch": 6.607929515418502,
"grad_norm": 0.5022340416908264,
"learning_rate": 1.435897435897436e-05,
"loss": 0.1113,
"step": 6000
},
{
"epoch": 6.607929515418502,
"eval_cer": 26.76395400823083,
"eval_loss": 0.7905788421630859,
"eval_runtime": 1296.6476,
"eval_samples_per_second": 8.16,
"eval_steps_per_second": 2.041,
"eval_wer": 86.42149929278642,
"step": 6000
},
{
"epoch": 6.635462555066079,
"grad_norm": 0.55087810754776,
"learning_rate": 1.4333333333333334e-05,
"loss": 0.0871,
"step": 6025
},
{
"epoch": 6.6629955947136565,
"grad_norm": 0.2666168510913849,
"learning_rate": 1.430769230769231e-05,
"loss": 0.1377,
"step": 6050
},
{
"epoch": 6.690528634361233,
"grad_norm": 0.6376879811286926,
"learning_rate": 1.4282051282051283e-05,
"loss": 0.1371,
"step": 6075
},
{
"epoch": 6.71806167400881,
"grad_norm": 0.6430523991584778,
"learning_rate": 1.4256410256410258e-05,
"loss": 0.1426,
"step": 6100
},
{
"epoch": 6.745594713656388,
"grad_norm": 0.4945014715194702,
"learning_rate": 1.4230769230769232e-05,
"loss": 0.1288,
"step": 6125
},
{
"epoch": 6.773127753303965,
"grad_norm": 0.7737772464752197,
"learning_rate": 1.4205128205128207e-05,
"loss": 0.1124,
"step": 6150
},
{
"epoch": 6.8006607929515415,
"grad_norm": 0.6132165789604187,
"learning_rate": 1.4179487179487181e-05,
"loss": 0.1234,
"step": 6175
},
{
"epoch": 6.828193832599119,
"grad_norm": 0.3980983793735504,
"learning_rate": 1.4153846153846156e-05,
"loss": 0.1079,
"step": 6200
},
{
"epoch": 6.855726872246696,
"grad_norm": 0.583961009979248,
"learning_rate": 1.412820512820513e-05,
"loss": 0.1378,
"step": 6225
},
{
"epoch": 6.883259911894273,
"grad_norm": 0.37798672914505005,
"learning_rate": 1.4102564102564105e-05,
"loss": 0.1087,
"step": 6250
},
{
"epoch": 6.9107929515418505,
"grad_norm": 0.4246102571487427,
"learning_rate": 1.4076923076923079e-05,
"loss": 0.0986,
"step": 6275
},
{
"epoch": 6.938325991189427,
"grad_norm": 0.5898611545562744,
"learning_rate": 1.405128205128205e-05,
"loss": 0.1116,
"step": 6300
},
{
"epoch": 6.965859030837004,
"grad_norm": 0.6355592608451843,
"learning_rate": 1.4025641025641026e-05,
"loss": 0.1243,
"step": 6325
},
{
"epoch": 6.993392070484582,
"grad_norm": 0.7065783739089966,
"learning_rate": 1.4e-05,
"loss": 0.1175,
"step": 6350
},
{
"epoch": 7.020925110132159,
"grad_norm": 0.2747715711593628,
"learning_rate": 1.3974358974358975e-05,
"loss": 0.0794,
"step": 6375
},
{
"epoch": 7.048458149779735,
"grad_norm": 0.4818078875541687,
"learning_rate": 1.3948717948717949e-05,
"loss": 0.0762,
"step": 6400
},
{
"epoch": 7.075991189427313,
"grad_norm": 0.32405680418014526,
"learning_rate": 1.3923076923076924e-05,
"loss": 0.0672,
"step": 6425
},
{
"epoch": 7.10352422907489,
"grad_norm": 0.43055686354637146,
"learning_rate": 1.3897435897435898e-05,
"loss": 0.0608,
"step": 6450
},
{
"epoch": 7.131057268722467,
"grad_norm": 0.18856699764728546,
"learning_rate": 1.3871794871794873e-05,
"loss": 0.0737,
"step": 6475
},
{
"epoch": 7.158590308370044,
"grad_norm": 0.3173889219760895,
"learning_rate": 1.3846153846153847e-05,
"loss": 0.082,
"step": 6500
},
{
"epoch": 7.186123348017621,
"grad_norm": 0.6276409029960632,
"learning_rate": 1.3820512820512822e-05,
"loss": 0.0711,
"step": 6525
},
{
"epoch": 7.213656387665198,
"grad_norm": 0.6348710656166077,
"learning_rate": 1.3794871794871796e-05,
"loss": 0.0702,
"step": 6550
},
{
"epoch": 7.241189427312776,
"grad_norm": 0.22085753083229065,
"learning_rate": 1.3769230769230771e-05,
"loss": 0.08,
"step": 6575
},
{
"epoch": 7.2687224669603525,
"grad_norm": 0.5453472137451172,
"learning_rate": 1.3743589743589745e-05,
"loss": 0.0747,
"step": 6600
},
{
"epoch": 7.296255506607929,
"grad_norm": 0.31948912143707275,
"learning_rate": 1.3717948717948718e-05,
"loss": 0.091,
"step": 6625
},
{
"epoch": 7.323788546255507,
"grad_norm": 0.3135238289833069,
"learning_rate": 1.3692307692307694e-05,
"loss": 0.0786,
"step": 6650
},
{
"epoch": 7.351321585903084,
"grad_norm": 0.5533322691917419,
"learning_rate": 1.3666666666666667e-05,
"loss": 0.0864,
"step": 6675
},
{
"epoch": 7.378854625550661,
"grad_norm": 0.38217464089393616,
"learning_rate": 1.3641025641025643e-05,
"loss": 0.0634,
"step": 6700
},
{
"epoch": 7.406387665198238,
"grad_norm": 0.16825051605701447,
"learning_rate": 1.3615384615384616e-05,
"loss": 0.0601,
"step": 6725
},
{
"epoch": 7.433920704845815,
"grad_norm": 0.4651043117046356,
"learning_rate": 1.3589743589743592e-05,
"loss": 0.0747,
"step": 6750
},
{
"epoch": 7.461453744493392,
"grad_norm": 0.28918442130088806,
"learning_rate": 1.3564102564102565e-05,
"loss": 0.086,
"step": 6775
},
{
"epoch": 7.48898678414097,
"grad_norm": 0.28500112891197205,
"learning_rate": 1.353846153846154e-05,
"loss": 0.0773,
"step": 6800
},
{
"epoch": 7.516519823788546,
"grad_norm": 0.4884459376335144,
"learning_rate": 1.3512820512820514e-05,
"loss": 0.0738,
"step": 6825
},
{
"epoch": 7.544052863436123,
"grad_norm": 0.436380535364151,
"learning_rate": 1.348717948717949e-05,
"loss": 0.0736,
"step": 6850
},
{
"epoch": 7.5715859030837,
"grad_norm": 0.6121215224266052,
"learning_rate": 1.3461538461538463e-05,
"loss": 0.0732,
"step": 6875
},
{
"epoch": 7.599118942731278,
"grad_norm": 0.42172011733055115,
"learning_rate": 1.3435897435897435e-05,
"loss": 0.0947,
"step": 6900
},
{
"epoch": 7.6266519823788546,
"grad_norm": 0.423505961894989,
"learning_rate": 1.341025641025641e-05,
"loss": 0.0704,
"step": 6925
},
{
"epoch": 7.654185022026431,
"grad_norm": 0.5993038415908813,
"learning_rate": 1.3384615384615384e-05,
"loss": 0.1009,
"step": 6950
},
{
"epoch": 7.681718061674009,
"grad_norm": 0.5881207585334778,
"learning_rate": 1.335897435897436e-05,
"loss": 0.0912,
"step": 6975
},
{
"epoch": 7.709251101321586,
"grad_norm": 0.5848076939582825,
"learning_rate": 1.3333333333333333e-05,
"loss": 0.0673,
"step": 7000
},
{
"epoch": 7.709251101321586,
"eval_cer": 27.13709261403368,
"eval_loss": 0.7989184260368347,
"eval_runtime": 1306.8838,
"eval_samples_per_second": 8.096,
"eval_steps_per_second": 2.025,
"eval_wer": 84.61103253182462,
"step": 7000
},
{
"epoch": 7.736784140969163,
"grad_norm": 0.8075867295265198,
"learning_rate": 1.3307692307692309e-05,
"loss": 0.0946,
"step": 7025
},
{
"epoch": 7.76431718061674,
"grad_norm": 0.6053417325019836,
"learning_rate": 1.3282051282051282e-05,
"loss": 0.0816,
"step": 7050
},
{
"epoch": 7.791850220264317,
"grad_norm": 0.6305066347122192,
"learning_rate": 1.3256410256410258e-05,
"loss": 0.079,
"step": 7075
},
{
"epoch": 7.819383259911894,
"grad_norm": 0.5551367998123169,
"learning_rate": 1.3230769230769231e-05,
"loss": 0.0743,
"step": 7100
},
{
"epoch": 7.846916299559472,
"grad_norm": 0.362006276845932,
"learning_rate": 1.3205128205128207e-05,
"loss": 0.0694,
"step": 7125
},
{
"epoch": 7.8744493392070485,
"grad_norm": 0.3232136368751526,
"learning_rate": 1.317948717948718e-05,
"loss": 0.0947,
"step": 7150
},
{
"epoch": 7.901982378854625,
"grad_norm": 0.23883138597011566,
"learning_rate": 1.3153846153846156e-05,
"loss": 0.0896,
"step": 7175
},
{
"epoch": 7.929515418502203,
"grad_norm": 0.49950769543647766,
"learning_rate": 1.312820512820513e-05,
"loss": 0.1053,
"step": 7200
},
{
"epoch": 7.95704845814978,
"grad_norm": 0.5408104658126831,
"learning_rate": 1.3102564102564103e-05,
"loss": 0.0903,
"step": 7225
},
{
"epoch": 7.984581497797357,
"grad_norm": 0.35938769578933716,
"learning_rate": 1.3076923076923078e-05,
"loss": 0.0908,
"step": 7250
},
{
"epoch": 8.012114537444933,
"grad_norm": 0.29418617486953735,
"learning_rate": 1.3051282051282052e-05,
"loss": 0.0998,
"step": 7275
},
{
"epoch": 8.039647577092511,
"grad_norm": 0.39860227704048157,
"learning_rate": 1.3025641025641027e-05,
"loss": 0.0574,
"step": 7300
},
{
"epoch": 8.067180616740089,
"grad_norm": 0.34400638937950134,
"learning_rate": 1.3000000000000001e-05,
"loss": 0.0556,
"step": 7325
},
{
"epoch": 8.094713656387665,
"grad_norm": 0.39651989936828613,
"learning_rate": 1.2974358974358976e-05,
"loss": 0.0561,
"step": 7350
},
{
"epoch": 8.122246696035242,
"grad_norm": 0.43096983432769775,
"learning_rate": 1.294871794871795e-05,
"loss": 0.0521,
"step": 7375
},
{
"epoch": 8.14977973568282,
"grad_norm": 0.2908264994621277,
"learning_rate": 1.2923076923076925e-05,
"loss": 0.052,
"step": 7400
},
{
"epoch": 8.177312775330396,
"grad_norm": 0.1850954294204712,
"learning_rate": 1.2897435897435899e-05,
"loss": 0.0537,
"step": 7425
},
{
"epoch": 8.204845814977974,
"grad_norm": 0.11924485862255096,
"learning_rate": 1.2871794871794874e-05,
"loss": 0.0436,
"step": 7450
},
{
"epoch": 8.232378854625551,
"grad_norm": 0.9369707703590393,
"learning_rate": 1.2846153846153848e-05,
"loss": 0.0616,
"step": 7475
},
{
"epoch": 8.259911894273127,
"grad_norm": 0.3720383644104004,
"learning_rate": 1.2820512820512823e-05,
"loss": 0.0483,
"step": 7500
},
{
"epoch": 8.287444933920705,
"grad_norm": 0.5974137187004089,
"learning_rate": 1.2794871794871795e-05,
"loss": 0.0469,
"step": 7525
},
{
"epoch": 8.314977973568283,
"grad_norm": 0.3387705981731415,
"learning_rate": 1.2769230769230769e-05,
"loss": 0.0512,
"step": 7550
},
{
"epoch": 8.342511013215859,
"grad_norm": 0.5228244066238403,
"learning_rate": 1.2743589743589744e-05,
"loss": 0.0542,
"step": 7575
},
{
"epoch": 8.370044052863436,
"grad_norm": 0.6965579390525818,
"learning_rate": 1.2717948717948718e-05,
"loss": 0.0499,
"step": 7600
},
{
"epoch": 8.397577092511014,
"grad_norm": 0.5114963054656982,
"learning_rate": 1.2692307692307693e-05,
"loss": 0.0641,
"step": 7625
},
{
"epoch": 8.42511013215859,
"grad_norm": 0.17714178562164307,
"learning_rate": 1.2666666666666667e-05,
"loss": 0.0474,
"step": 7650
},
{
"epoch": 8.452643171806168,
"grad_norm": 0.9684805274009705,
"learning_rate": 1.2641025641025642e-05,
"loss": 0.0557,
"step": 7675
},
{
"epoch": 8.480176211453745,
"grad_norm": 0.20210257172584534,
"learning_rate": 1.2615384615384616e-05,
"loss": 0.0548,
"step": 7700
},
{
"epoch": 8.507709251101321,
"grad_norm": 0.27927425503730774,
"learning_rate": 1.2589743589743591e-05,
"loss": 0.0523,
"step": 7725
},
{
"epoch": 8.535242290748899,
"grad_norm": 0.5390238761901855,
"learning_rate": 1.2564102564102565e-05,
"loss": 0.0693,
"step": 7750
},
{
"epoch": 8.562775330396477,
"grad_norm": 0.4304569363594055,
"learning_rate": 1.253846153846154e-05,
"loss": 0.0613,
"step": 7775
},
{
"epoch": 8.590308370044053,
"grad_norm": 0.5010227560997009,
"learning_rate": 1.2512820512820514e-05,
"loss": 0.0637,
"step": 7800
},
{
"epoch": 8.61784140969163,
"grad_norm": 0.39493897557258606,
"learning_rate": 1.2487179487179487e-05,
"loss": 0.0674,
"step": 7825
},
{
"epoch": 8.645374449339208,
"grad_norm": 0.3617560565471649,
"learning_rate": 1.2461538461538463e-05,
"loss": 0.0552,
"step": 7850
},
{
"epoch": 8.672907488986784,
"grad_norm": 0.9049015045166016,
"learning_rate": 1.2435897435897436e-05,
"loss": 0.0424,
"step": 7875
},
{
"epoch": 8.700440528634362,
"grad_norm": 0.687857449054718,
"learning_rate": 1.2410256410256412e-05,
"loss": 0.0656,
"step": 7900
},
{
"epoch": 8.72797356828194,
"grad_norm": 0.525587260723114,
"learning_rate": 1.2384615384615385e-05,
"loss": 0.0655,
"step": 7925
},
{
"epoch": 8.755506607929515,
"grad_norm": 0.4091179072856903,
"learning_rate": 1.235897435897436e-05,
"loss": 0.066,
"step": 7950
},
{
"epoch": 8.783039647577093,
"grad_norm": 0.32147249579429626,
"learning_rate": 1.2333333333333334e-05,
"loss": 0.0583,
"step": 7975
},
{
"epoch": 8.810572687224669,
"grad_norm": 0.47001683712005615,
"learning_rate": 1.230769230769231e-05,
"loss": 0.0624,
"step": 8000
},
{
"epoch": 8.810572687224669,
"eval_cer": 24.95328602707691,
"eval_loss": 0.8190127015113831,
"eval_runtime": 1298.6148,
"eval_samples_per_second": 8.148,
"eval_steps_per_second": 2.038,
"eval_wer": 84.65818010372466,
"step": 8000
},
{
"epoch": 8.838105726872246,
"grad_norm": 0.4862002730369568,
"learning_rate": 1.2282051282051283e-05,
"loss": 0.0534,
"step": 8025
},
{
"epoch": 8.865638766519824,
"grad_norm": 0.4833989143371582,
"learning_rate": 1.2256410256410259e-05,
"loss": 0.0582,
"step": 8050
},
{
"epoch": 8.8931718061674,
"grad_norm": 0.3487415611743927,
"learning_rate": 1.2230769230769232e-05,
"loss": 0.0569,
"step": 8075
},
{
"epoch": 8.920704845814978,
"grad_norm": 0.22568881511688232,
"learning_rate": 1.2205128205128208e-05,
"loss": 0.0668,
"step": 8100
},
{
"epoch": 8.948237885462555,
"grad_norm": 0.47726893424987793,
"learning_rate": 1.217948717948718e-05,
"loss": 0.0518,
"step": 8125
},
{
"epoch": 8.975770925110131,
"grad_norm": 0.4065573513507843,
"learning_rate": 1.2153846153846153e-05,
"loss": 0.0406,
"step": 8150
},
{
"epoch": 9.003303964757709,
"grad_norm": 0.2732442021369934,
"learning_rate": 1.2128205128205129e-05,
"loss": 0.0624,
"step": 8175
},
{
"epoch": 9.030837004405287,
"grad_norm": 0.4044542610645294,
"learning_rate": 1.2102564102564102e-05,
"loss": 0.0338,
"step": 8200
},
{
"epoch": 9.058370044052863,
"grad_norm": 0.12826256453990936,
"learning_rate": 1.2076923076923078e-05,
"loss": 0.0321,
"step": 8225
},
{
"epoch": 9.08590308370044,
"grad_norm": 0.3568229377269745,
"learning_rate": 1.2051282051282051e-05,
"loss": 0.0285,
"step": 8250
},
{
"epoch": 9.113436123348018,
"grad_norm": 0.37862274050712585,
"learning_rate": 1.2025641025641027e-05,
"loss": 0.0343,
"step": 8275
},
{
"epoch": 9.140969162995594,
"grad_norm": 0.2104802429676056,
"learning_rate": 1.2e-05,
"loss": 0.0384,
"step": 8300
},
{
"epoch": 9.168502202643172,
"grad_norm": 0.2630591094493866,
"learning_rate": 1.1974358974358976e-05,
"loss": 0.0266,
"step": 8325
},
{
"epoch": 9.19603524229075,
"grad_norm": 0.40195232629776,
"learning_rate": 1.194871794871795e-05,
"loss": 0.038,
"step": 8350
},
{
"epoch": 9.223568281938325,
"grad_norm": 0.23489591479301453,
"learning_rate": 1.1923076923076925e-05,
"loss": 0.0301,
"step": 8375
},
{
"epoch": 9.251101321585903,
"grad_norm": 0.24858242273330688,
"learning_rate": 1.1897435897435898e-05,
"loss": 0.0313,
"step": 8400
},
{
"epoch": 9.27863436123348,
"grad_norm": 0.3414604067802429,
"learning_rate": 1.1871794871794872e-05,
"loss": 0.0293,
"step": 8425
},
{
"epoch": 9.306167400881057,
"grad_norm": 0.2159339338541031,
"learning_rate": 1.1846153846153847e-05,
"loss": 0.0365,
"step": 8450
},
{
"epoch": 9.333700440528634,
"grad_norm": 0.2686537206172943,
"learning_rate": 1.1820512820512821e-05,
"loss": 0.0313,
"step": 8475
},
{
"epoch": 9.361233480176212,
"grad_norm": 0.2898625433444977,
"learning_rate": 1.1794871794871796e-05,
"loss": 0.0414,
"step": 8500
},
{
"epoch": 9.388766519823788,
"grad_norm": 0.5631348490715027,
"learning_rate": 1.176923076923077e-05,
"loss": 0.0354,
"step": 8525
},
{
"epoch": 9.416299559471366,
"grad_norm": 0.19131043553352356,
"learning_rate": 1.1743589743589745e-05,
"loss": 0.0409,
"step": 8550
},
{
"epoch": 9.443832599118943,
"grad_norm": 0.37251201272010803,
"learning_rate": 1.1717948717948719e-05,
"loss": 0.0342,
"step": 8575
},
{
"epoch": 9.47136563876652,
"grad_norm": 0.3231589198112488,
"learning_rate": 1.1692307692307694e-05,
"loss": 0.0336,
"step": 8600
},
{
"epoch": 9.498898678414097,
"grad_norm": 0.6271162033081055,
"learning_rate": 1.1666666666666668e-05,
"loss": 0.0459,
"step": 8625
},
{
"epoch": 9.526431718061675,
"grad_norm": 0.11155181378126144,
"learning_rate": 1.1641025641025643e-05,
"loss": 0.0446,
"step": 8650
},
{
"epoch": 9.55396475770925,
"grad_norm": 0.2658681869506836,
"learning_rate": 1.1615384615384617e-05,
"loss": 0.0352,
"step": 8675
},
{
"epoch": 9.581497797356828,
"grad_norm": 0.3016481399536133,
"learning_rate": 1.1589743589743592e-05,
"loss": 0.0267,
"step": 8700
},
{
"epoch": 9.609030837004406,
"grad_norm": 0.1538410484790802,
"learning_rate": 1.1564102564102566e-05,
"loss": 0.0431,
"step": 8725
},
{
"epoch": 9.636563876651982,
"grad_norm": 0.28318852186203003,
"learning_rate": 1.1538461538461538e-05,
"loss": 0.0439,
"step": 8750
},
{
"epoch": 9.66409691629956,
"grad_norm": 0.43803560733795166,
"learning_rate": 1.1512820512820513e-05,
"loss": 0.0329,
"step": 8775
},
{
"epoch": 9.691629955947137,
"grad_norm": 0.2602671980857849,
"learning_rate": 1.1487179487179487e-05,
"loss": 0.0293,
"step": 8800
},
{
"epoch": 9.719162995594713,
"grad_norm": 0.3581433892250061,
"learning_rate": 1.1461538461538462e-05,
"loss": 0.0343,
"step": 8825
},
{
"epoch": 9.746696035242291,
"grad_norm": 0.37504616379737854,
"learning_rate": 1.1435897435897436e-05,
"loss": 0.0344,
"step": 8850
},
{
"epoch": 9.774229074889869,
"grad_norm": 0.2697007954120636,
"learning_rate": 1.1410256410256411e-05,
"loss": 0.0416,
"step": 8875
},
{
"epoch": 9.801762114537445,
"grad_norm": 0.4063343107700348,
"learning_rate": 1.1384615384615385e-05,
"loss": 0.0331,
"step": 8900
},
{
"epoch": 9.829295154185022,
"grad_norm": 0.24234430491924286,
"learning_rate": 1.135897435897436e-05,
"loss": 0.0519,
"step": 8925
},
{
"epoch": 9.8568281938326,
"grad_norm": 0.539232075214386,
"learning_rate": 1.1333333333333334e-05,
"loss": 0.0475,
"step": 8950
},
{
"epoch": 9.884361233480176,
"grad_norm": 0.37743422389030457,
"learning_rate": 1.1307692307692309e-05,
"loss": 0.0328,
"step": 8975
},
{
"epoch": 9.911894273127754,
"grad_norm": 0.32006239891052246,
"learning_rate": 1.1282051282051283e-05,
"loss": 0.0355,
"step": 9000
},
{
"epoch": 9.911894273127754,
"eval_cer": 24.863870324303875,
"eval_loss": 0.8439357280731201,
"eval_runtime": 1305.3411,
"eval_samples_per_second": 8.106,
"eval_steps_per_second": 2.027,
"eval_wer": 84.0924092409241,
"step": 9000
},
{
"epoch": 9.939427312775331,
"grad_norm": 0.22994668781757355,
"learning_rate": 1.1256410256410258e-05,
"loss": 0.0382,
"step": 9025
},
{
"epoch": 9.966960352422907,
"grad_norm": 0.6525270938873291,
"learning_rate": 1.1230769230769232e-05,
"loss": 0.0414,
"step": 9050
},
{
"epoch": 9.994493392070485,
"grad_norm": 0.37982580065727234,
"learning_rate": 1.1205128205128205e-05,
"loss": 0.0378,
"step": 9075
},
{
"epoch": 10.022026431718063,
"grad_norm": 0.4457830786705017,
"learning_rate": 1.117948717948718e-05,
"loss": 0.026,
"step": 9100
},
{
"epoch": 10.049559471365638,
"grad_norm": 0.36874255537986755,
"learning_rate": 1.1153846153846154e-05,
"loss": 0.0235,
"step": 9125
},
{
"epoch": 10.077092511013216,
"grad_norm": 0.33007436990737915,
"learning_rate": 1.112820512820513e-05,
"loss": 0.0276,
"step": 9150
},
{
"epoch": 10.104625550660794,
"grad_norm": 0.6235056519508362,
"learning_rate": 1.1102564102564103e-05,
"loss": 0.0198,
"step": 9175
},
{
"epoch": 10.13215859030837,
"grad_norm": 0.3338548243045807,
"learning_rate": 1.1076923076923079e-05,
"loss": 0.0219,
"step": 9200
},
{
"epoch": 10.159691629955947,
"grad_norm": 0.21556895971298218,
"learning_rate": 1.1051282051282052e-05,
"loss": 0.021,
"step": 9225
},
{
"epoch": 10.187224669603523,
"grad_norm": 0.22250282764434814,
"learning_rate": 1.1025641025641028e-05,
"loss": 0.029,
"step": 9250
},
{
"epoch": 10.214757709251101,
"grad_norm": 0.17351406812667847,
"learning_rate": 1.1000000000000001e-05,
"loss": 0.0171,
"step": 9275
},
{
"epoch": 10.242290748898679,
"grad_norm": 0.396767795085907,
"learning_rate": 1.0974358974358977e-05,
"loss": 0.0178,
"step": 9300
},
{
"epoch": 10.269823788546255,
"grad_norm": 0.3102997839450836,
"learning_rate": 1.094871794871795e-05,
"loss": 0.0224,
"step": 9325
},
{
"epoch": 10.297356828193832,
"grad_norm": 0.27433833479881287,
"learning_rate": 1.0923076923076922e-05,
"loss": 0.0279,
"step": 9350
},
{
"epoch": 10.32488986784141,
"grad_norm": 0.4114525020122528,
"learning_rate": 1.0897435897435898e-05,
"loss": 0.0281,
"step": 9375
},
{
"epoch": 10.352422907488986,
"grad_norm": 0.2669256627559662,
"learning_rate": 1.0871794871794871e-05,
"loss": 0.0223,
"step": 9400
},
{
"epoch": 10.379955947136564,
"grad_norm": 0.1441548466682434,
"learning_rate": 1.0846153846153847e-05,
"loss": 0.0275,
"step": 9425
},
{
"epoch": 10.407488986784141,
"grad_norm": 0.25263088941574097,
"learning_rate": 1.082051282051282e-05,
"loss": 0.0287,
"step": 9450
},
{
"epoch": 10.435022026431717,
"grad_norm": 0.4364434480667114,
"learning_rate": 1.0794871794871796e-05,
"loss": 0.0257,
"step": 9475
},
{
"epoch": 10.462555066079295,
"grad_norm": 0.287691593170166,
"learning_rate": 1.076923076923077e-05,
"loss": 0.0273,
"step": 9500
},
{
"epoch": 10.490088105726873,
"grad_norm": 0.22964105010032654,
"learning_rate": 1.0743589743589745e-05,
"loss": 0.0319,
"step": 9525
},
{
"epoch": 10.517621145374449,
"grad_norm": 0.3216736316680908,
"learning_rate": 1.0717948717948718e-05,
"loss": 0.0259,
"step": 9550
},
{
"epoch": 10.545154185022026,
"grad_norm": 0.3481162190437317,
"learning_rate": 1.0692307692307694e-05,
"loss": 0.0273,
"step": 9575
},
{
"epoch": 10.572687224669604,
"grad_norm": 0.4183979630470276,
"learning_rate": 1.0666666666666667e-05,
"loss": 0.0329,
"step": 9600
},
{
"epoch": 10.60022026431718,
"grad_norm": 0.45229265093803406,
"learning_rate": 1.0641025641025643e-05,
"loss": 0.0351,
"step": 9625
},
{
"epoch": 10.627753303964758,
"grad_norm": 0.20271629095077515,
"learning_rate": 1.0615384615384616e-05,
"loss": 0.033,
"step": 9650
},
{
"epoch": 10.655286343612335,
"grad_norm": 0.2974882125854492,
"learning_rate": 1.058974358974359e-05,
"loss": 0.0336,
"step": 9675
},
{
"epoch": 10.682819383259911,
"grad_norm": 0.31492355465888977,
"learning_rate": 1.0564102564102565e-05,
"loss": 0.0318,
"step": 9700
},
{
"epoch": 10.710352422907489,
"grad_norm": 0.3608117699623108,
"learning_rate": 1.0538461538461539e-05,
"loss": 0.0329,
"step": 9725
},
{
"epoch": 10.737885462555067,
"grad_norm": 0.20358450710773468,
"learning_rate": 1.0512820512820514e-05,
"loss": 0.0322,
"step": 9750
},
{
"epoch": 10.765418502202643,
"grad_norm": 0.26170098781585693,
"learning_rate": 1.0487179487179488e-05,
"loss": 0.0279,
"step": 9775
},
{
"epoch": 10.79295154185022,
"grad_norm": 0.45118996500968933,
"learning_rate": 1.0461538461538463e-05,
"loss": 0.0304,
"step": 9800
},
{
"epoch": 10.820484581497798,
"grad_norm": 0.3625517785549164,
"learning_rate": 1.0435897435897437e-05,
"loss": 0.0314,
"step": 9825
},
{
"epoch": 10.848017621145374,
"grad_norm": 0.15762348473072052,
"learning_rate": 1.0410256410256412e-05,
"loss": 0.0235,
"step": 9850
},
{
"epoch": 10.875550660792952,
"grad_norm": 0.24998855590820312,
"learning_rate": 1.0384615384615386e-05,
"loss": 0.0278,
"step": 9875
},
{
"epoch": 10.90308370044053,
"grad_norm": 0.3237763047218323,
"learning_rate": 1.0358974358974361e-05,
"loss": 0.024,
"step": 9900
},
{
"epoch": 10.930616740088105,
"grad_norm": 0.3094446063041687,
"learning_rate": 1.0333333333333335e-05,
"loss": 0.0202,
"step": 9925
},
{
"epoch": 10.958149779735683,
"grad_norm": 0.20104403793811798,
"learning_rate": 1.0307692307692307e-05,
"loss": 0.0266,
"step": 9950
},
{
"epoch": 10.98568281938326,
"grad_norm": 0.7808817028999329,
"learning_rate": 1.0282051282051282e-05,
"loss": 0.0315,
"step": 9975
},
{
"epoch": 11.013215859030836,
"grad_norm": 0.19259458780288696,
"learning_rate": 1.0256410256410256e-05,
"loss": 0.0262,
"step": 10000
},
{
"epoch": 11.013215859030836,
"eval_cer": 25.51901230039091,
"eval_loss": 0.8545929193496704,
"eval_runtime": 1300.8852,
"eval_samples_per_second": 8.134,
"eval_steps_per_second": 2.034,
"eval_wer": 84.75247524752476,
"step": 10000
},
{
"epoch": 11.040748898678414,
"grad_norm": 0.31786996126174927,
"learning_rate": 1.0230769230769231e-05,
"loss": 0.0195,
"step": 10025
},
{
"epoch": 11.068281938325992,
"grad_norm": 0.22015278041362762,
"learning_rate": 1.0205128205128205e-05,
"loss": 0.02,
"step": 10050
},
{
"epoch": 11.095814977973568,
"grad_norm": 0.06974627077579498,
"learning_rate": 1.017948717948718e-05,
"loss": 0.018,
"step": 10075
},
{
"epoch": 11.123348017621145,
"grad_norm": 0.2470981329679489,
"learning_rate": 1.0153846153846154e-05,
"loss": 0.0177,
"step": 10100
},
{
"epoch": 11.150881057268723,
"grad_norm": 0.15128523111343384,
"learning_rate": 1.012820512820513e-05,
"loss": 0.0165,
"step": 10125
},
{
"epoch": 11.178414096916299,
"grad_norm": 0.19330032169818878,
"learning_rate": 1.0102564102564103e-05,
"loss": 0.0194,
"step": 10150
},
{
"epoch": 11.205947136563877,
"grad_norm": 0.6133860349655151,
"learning_rate": 1.0076923076923078e-05,
"loss": 0.0208,
"step": 10175
},
{
"epoch": 11.233480176211454,
"grad_norm": 0.1094980537891388,
"learning_rate": 1.0051282051282052e-05,
"loss": 0.0253,
"step": 10200
},
{
"epoch": 11.26101321585903,
"grad_norm": 0.21437473595142365,
"learning_rate": 1.0025641025641027e-05,
"loss": 0.0185,
"step": 10225
},
{
"epoch": 11.288546255506608,
"grad_norm": 0.12496840208768845,
"learning_rate": 1e-05,
"loss": 0.0174,
"step": 10250
},
{
"epoch": 11.316079295154186,
"grad_norm": 0.08179640024900436,
"learning_rate": 9.974358974358974e-06,
"loss": 0.0175,
"step": 10275
},
{
"epoch": 11.343612334801762,
"grad_norm": 0.19107000529766083,
"learning_rate": 9.94871794871795e-06,
"loss": 0.0192,
"step": 10300
},
{
"epoch": 11.37114537444934,
"grad_norm": 0.07791900634765625,
"learning_rate": 9.923076923076923e-06,
"loss": 0.0183,
"step": 10325
},
{
"epoch": 11.398678414096917,
"grad_norm": 0.1208205595612526,
"learning_rate": 9.897435897435899e-06,
"loss": 0.0194,
"step": 10350
},
{
"epoch": 11.426211453744493,
"grad_norm": 0.17363104224205017,
"learning_rate": 9.871794871794872e-06,
"loss": 0.0182,
"step": 10375
},
{
"epoch": 11.45374449339207,
"grad_norm": 0.15507516264915466,
"learning_rate": 9.846153846153848e-06,
"loss": 0.0211,
"step": 10400
},
{
"epoch": 11.481277533039648,
"grad_norm": 0.42732682824134827,
"learning_rate": 9.820512820512821e-06,
"loss": 0.0196,
"step": 10425
},
{
"epoch": 11.508810572687224,
"grad_norm": 0.2086770087480545,
"learning_rate": 9.794871794871795e-06,
"loss": 0.0174,
"step": 10450
},
{
"epoch": 11.536343612334802,
"grad_norm": 0.13412240147590637,
"learning_rate": 9.76923076923077e-06,
"loss": 0.0115,
"step": 10475
},
{
"epoch": 11.56387665198238,
"grad_norm": 0.5641820430755615,
"learning_rate": 9.743589743589744e-06,
"loss": 0.0213,
"step": 10500
},
{
"epoch": 11.591409691629956,
"grad_norm": 0.2280423492193222,
"learning_rate": 9.71794871794872e-06,
"loss": 0.0208,
"step": 10525
},
{
"epoch": 11.618942731277533,
"grad_norm": 0.34723925590515137,
"learning_rate": 9.692307692307693e-06,
"loss": 0.0145,
"step": 10550
},
{
"epoch": 11.646475770925111,
"grad_norm": 0.42695844173431396,
"learning_rate": 9.666666666666667e-06,
"loss": 0.0191,
"step": 10575
},
{
"epoch": 11.674008810572687,
"grad_norm": 0.179496169090271,
"learning_rate": 9.641025641025642e-06,
"loss": 0.0159,
"step": 10600
},
{
"epoch": 11.701541850220265,
"grad_norm": 0.20549049973487854,
"learning_rate": 9.615384615384616e-06,
"loss": 0.0148,
"step": 10625
},
{
"epoch": 11.729074889867842,
"grad_norm": 0.07314030826091766,
"learning_rate": 9.589743589743591e-06,
"loss": 0.0166,
"step": 10650
},
{
"epoch": 11.756607929515418,
"grad_norm": 0.1925014853477478,
"learning_rate": 9.564102564102565e-06,
"loss": 0.0156,
"step": 10675
},
{
"epoch": 11.784140969162996,
"grad_norm": 0.3135490417480469,
"learning_rate": 9.53846153846154e-06,
"loss": 0.0186,
"step": 10700
},
{
"epoch": 11.811674008810574,
"grad_norm": 0.4319511651992798,
"learning_rate": 9.512820512820514e-06,
"loss": 0.0172,
"step": 10725
},
{
"epoch": 11.83920704845815,
"grad_norm": 0.7985556125640869,
"learning_rate": 9.487179487179487e-06,
"loss": 0.0254,
"step": 10750
},
{
"epoch": 11.866740088105727,
"grad_norm": 0.0852452889084816,
"learning_rate": 9.461538461538463e-06,
"loss": 0.0207,
"step": 10775
},
{
"epoch": 11.894273127753303,
"grad_norm": 0.1884285807609558,
"learning_rate": 9.435897435897436e-06,
"loss": 0.0165,
"step": 10800
},
{
"epoch": 11.92180616740088,
"grad_norm": 0.1979898363351822,
"learning_rate": 9.410256410256412e-06,
"loss": 0.0174,
"step": 10825
},
{
"epoch": 11.949339207048459,
"grad_norm": 0.48160043358802795,
"learning_rate": 9.384615384615385e-06,
"loss": 0.0208,
"step": 10850
},
{
"epoch": 11.976872246696034,
"grad_norm": 0.3505367040634155,
"learning_rate": 9.358974358974359e-06,
"loss": 0.0234,
"step": 10875
},
{
"epoch": 12.004405286343612,
"grad_norm": 0.057797495275735855,
"learning_rate": 9.333333333333334e-06,
"loss": 0.0196,
"step": 10900
},
{
"epoch": 12.03193832599119,
"grad_norm": 0.0727895125746727,
"learning_rate": 9.307692307692308e-06,
"loss": 0.0155,
"step": 10925
},
{
"epoch": 12.059471365638766,
"grad_norm": 0.15412183105945587,
"learning_rate": 9.282051282051283e-06,
"loss": 0.0136,
"step": 10950
},
{
"epoch": 12.087004405286343,
"grad_norm": 0.11031453311443329,
"learning_rate": 9.256410256410257e-06,
"loss": 0.0097,
"step": 10975
},
{
"epoch": 12.114537444933921,
"grad_norm": 0.08627741038799286,
"learning_rate": 9.230769230769232e-06,
"loss": 0.0146,
"step": 11000
},
{
"epoch": 12.114537444933921,
"eval_cer": 23.871126752490458,
"eval_loss": 0.8571091890335083,
"eval_runtime": 1299.2634,
"eval_samples_per_second": 8.144,
"eval_steps_per_second": 2.037,
"eval_wer": 83.58321546440358,
"step": 11000
},
{
"epoch": 12.142070484581497,
"grad_norm": 0.17661835253238678,
"learning_rate": 9.205128205128206e-06,
"loss": 0.0164,
"step": 11025
},
{
"epoch": 12.169603524229075,
"grad_norm": 0.4166176915168762,
"learning_rate": 9.17948717948718e-06,
"loss": 0.0125,
"step": 11050
},
{
"epoch": 12.197136563876652,
"grad_norm": 0.08312999457120895,
"learning_rate": 9.153846153846155e-06,
"loss": 0.0164,
"step": 11075
},
{
"epoch": 12.224669603524228,
"grad_norm": 0.24172239005565643,
"learning_rate": 9.128205128205129e-06,
"loss": 0.0147,
"step": 11100
},
{
"epoch": 12.252202643171806,
"grad_norm": 0.054716769605875015,
"learning_rate": 9.102564102564104e-06,
"loss": 0.012,
"step": 11125
},
{
"epoch": 12.279735682819384,
"grad_norm": 0.2059546858072281,
"learning_rate": 9.076923076923078e-06,
"loss": 0.0126,
"step": 11150
},
{
"epoch": 12.30726872246696,
"grad_norm": 0.12244844436645508,
"learning_rate": 9.051282051282051e-06,
"loss": 0.0121,
"step": 11175
},
{
"epoch": 12.334801762114537,
"grad_norm": 0.10340509563684464,
"learning_rate": 9.025641025641027e-06,
"loss": 0.0124,
"step": 11200
},
{
"epoch": 12.362334801762115,
"grad_norm": 0.19950826466083527,
"learning_rate": 9e-06,
"loss": 0.0127,
"step": 11225
},
{
"epoch": 12.389867841409691,
"grad_norm": 0.17073887586593628,
"learning_rate": 8.974358974358976e-06,
"loss": 0.0109,
"step": 11250
},
{
"epoch": 12.417400881057269,
"grad_norm": 1.0863208770751953,
"learning_rate": 8.94871794871795e-06,
"loss": 0.0121,
"step": 11275
},
{
"epoch": 12.444933920704846,
"grad_norm": 0.08929897099733353,
"learning_rate": 8.923076923076925e-06,
"loss": 0.0164,
"step": 11300
},
{
"epoch": 12.472466960352422,
"grad_norm": 0.18044264614582062,
"learning_rate": 8.897435897435898e-06,
"loss": 0.0123,
"step": 11325
},
{
"epoch": 12.5,
"grad_norm": 0.1298847645521164,
"learning_rate": 8.871794871794872e-06,
"loss": 0.0167,
"step": 11350
},
{
"epoch": 12.527533039647578,
"grad_norm": 0.2778928577899933,
"learning_rate": 8.846153846153847e-06,
"loss": 0.0149,
"step": 11375
},
{
"epoch": 12.555066079295154,
"grad_norm": 0.17325860261917114,
"learning_rate": 8.820512820512821e-06,
"loss": 0.0181,
"step": 11400
},
{
"epoch": 12.582599118942731,
"grad_norm": 0.1085548922419548,
"learning_rate": 8.794871794871796e-06,
"loss": 0.0178,
"step": 11425
},
{
"epoch": 12.610132158590309,
"grad_norm": 0.1976926326751709,
"learning_rate": 8.76923076923077e-06,
"loss": 0.0163,
"step": 11450
},
{
"epoch": 12.637665198237885,
"grad_norm": 0.38330408930778503,
"learning_rate": 8.743589743589743e-06,
"loss": 0.0202,
"step": 11475
},
{
"epoch": 12.665198237885463,
"grad_norm": 0.24536395072937012,
"learning_rate": 8.717948717948719e-06,
"loss": 0.0167,
"step": 11500
},
{
"epoch": 12.69273127753304,
"grad_norm": 0.27022993564605713,
"learning_rate": 8.692307692307692e-06,
"loss": 0.0116,
"step": 11525
},
{
"epoch": 12.720264317180616,
"grad_norm": 0.33529072999954224,
"learning_rate": 8.666666666666668e-06,
"loss": 0.0116,
"step": 11550
},
{
"epoch": 12.747797356828194,
"grad_norm": 0.2724566161632538,
"learning_rate": 8.641025641025641e-06,
"loss": 0.0128,
"step": 11575
},
{
"epoch": 12.775330396475772,
"grad_norm": 0.42723408341407776,
"learning_rate": 8.615384615384617e-06,
"loss": 0.0136,
"step": 11600
},
{
"epoch": 12.802863436123348,
"grad_norm": 0.07522483170032501,
"learning_rate": 8.58974358974359e-06,
"loss": 0.0115,
"step": 11625
},
{
"epoch": 12.830396475770925,
"grad_norm": 0.3467361330986023,
"learning_rate": 8.564102564102564e-06,
"loss": 0.0136,
"step": 11650
},
{
"epoch": 12.857929515418503,
"grad_norm": 0.22382023930549622,
"learning_rate": 8.53846153846154e-06,
"loss": 0.014,
"step": 11675
},
{
"epoch": 12.885462555066079,
"grad_norm": 0.32350465655326843,
"learning_rate": 8.512820512820513e-06,
"loss": 0.0142,
"step": 11700
},
{
"epoch": 12.912995594713657,
"grad_norm": 0.09188594669103622,
"learning_rate": 8.487179487179488e-06,
"loss": 0.0137,
"step": 11725
},
{
"epoch": 12.940528634361234,
"grad_norm": 0.28958576917648315,
"learning_rate": 8.461538461538462e-06,
"loss": 0.013,
"step": 11750
},
{
"epoch": 12.96806167400881,
"grad_norm": 0.2282794862985611,
"learning_rate": 8.435897435897436e-06,
"loss": 0.0157,
"step": 11775
},
{
"epoch": 12.995594713656388,
"grad_norm": 0.4104384183883667,
"learning_rate": 8.410256410256411e-06,
"loss": 0.0166,
"step": 11800
},
{
"epoch": 13.023127753303966,
"grad_norm": 0.1922859251499176,
"learning_rate": 8.384615384615385e-06,
"loss": 0.0142,
"step": 11825
},
{
"epoch": 13.050660792951541,
"grad_norm": 0.13582871854305267,
"learning_rate": 8.35897435897436e-06,
"loss": 0.0084,
"step": 11850
},
{
"epoch": 13.07819383259912,
"grad_norm": 0.06614464521408081,
"learning_rate": 8.333333333333334e-06,
"loss": 0.0096,
"step": 11875
},
{
"epoch": 13.105726872246697,
"grad_norm": 0.05340244993567467,
"learning_rate": 8.307692307692309e-06,
"loss": 0.0101,
"step": 11900
},
{
"epoch": 13.133259911894273,
"grad_norm": 0.08093658089637756,
"learning_rate": 8.282051282051283e-06,
"loss": 0.0106,
"step": 11925
},
{
"epoch": 13.16079295154185,
"grad_norm": 0.06269955635070801,
"learning_rate": 8.256410256410256e-06,
"loss": 0.0088,
"step": 11950
},
{
"epoch": 13.188325991189428,
"grad_norm": 0.09279240667819977,
"learning_rate": 8.230769230769232e-06,
"loss": 0.0087,
"step": 11975
},
{
"epoch": 13.215859030837004,
"grad_norm": 0.7125237584114075,
"learning_rate": 8.205128205128205e-06,
"loss": 0.0089,
"step": 12000
},
{
"epoch": 13.215859030837004,
"eval_cer": 23.365584125273692,
"eval_loss": 0.8545632362365723,
"eval_runtime": 1300.1108,
"eval_samples_per_second": 8.139,
"eval_steps_per_second": 2.035,
"eval_wer": 82.86657237152286,
"step": 12000
},
{
"epoch": 13.243392070484582,
"grad_norm": 0.03670433908700943,
"learning_rate": 8.17948717948718e-06,
"loss": 0.0089,
"step": 12025
},
{
"epoch": 13.270925110132158,
"grad_norm": 0.047082215547561646,
"learning_rate": 8.153846153846154e-06,
"loss": 0.0082,
"step": 12050
},
{
"epoch": 13.298458149779735,
"grad_norm": 0.10568676888942719,
"learning_rate": 8.12820512820513e-06,
"loss": 0.0124,
"step": 12075
},
{
"epoch": 13.325991189427313,
"grad_norm": 0.08048704266548157,
"learning_rate": 8.102564102564103e-06,
"loss": 0.0096,
"step": 12100
},
{
"epoch": 13.353524229074889,
"grad_norm": 0.2693760395050049,
"learning_rate": 8.076923076923077e-06,
"loss": 0.0113,
"step": 12125
},
{
"epoch": 13.381057268722467,
"grad_norm": 0.10042113065719604,
"learning_rate": 8.051282051282052e-06,
"loss": 0.0098,
"step": 12150
},
{
"epoch": 13.408590308370044,
"grad_norm": 0.09456273168325424,
"learning_rate": 8.025641025641026e-06,
"loss": 0.0089,
"step": 12175
},
{
"epoch": 13.43612334801762,
"grad_norm": 0.0927928239107132,
"learning_rate": 8.000000000000001e-06,
"loss": 0.0085,
"step": 12200
},
{
"epoch": 13.463656387665198,
"grad_norm": 0.08245880156755447,
"learning_rate": 7.974358974358975e-06,
"loss": 0.0081,
"step": 12225
},
{
"epoch": 13.491189427312776,
"grad_norm": 0.11756213009357452,
"learning_rate": 7.948717948717949e-06,
"loss": 0.0084,
"step": 12250
},
{
"epoch": 13.518722466960352,
"grad_norm": 0.10488010197877884,
"learning_rate": 7.923076923076924e-06,
"loss": 0.0085,
"step": 12275
},
{
"epoch": 13.54625550660793,
"grad_norm": 0.26627275347709656,
"learning_rate": 7.897435897435898e-06,
"loss": 0.0104,
"step": 12300
},
{
"epoch": 13.573788546255507,
"grad_norm": 0.07299219816923141,
"learning_rate": 7.871794871794873e-06,
"loss": 0.0097,
"step": 12325
},
{
"epoch": 13.601321585903083,
"grad_norm": 0.12166466563940048,
"learning_rate": 7.846153846153847e-06,
"loss": 0.0098,
"step": 12350
},
{
"epoch": 13.62885462555066,
"grad_norm": 0.05905827134847641,
"learning_rate": 7.820512820512822e-06,
"loss": 0.0139,
"step": 12375
},
{
"epoch": 13.656387665198238,
"grad_norm": 0.2588983476161957,
"learning_rate": 7.794871794871796e-06,
"loss": 0.0122,
"step": 12400
},
{
"epoch": 13.683920704845814,
"grad_norm": 0.04871825873851776,
"learning_rate": 7.76923076923077e-06,
"loss": 0.0099,
"step": 12425
},
{
"epoch": 13.711453744493392,
"grad_norm": 0.14013290405273438,
"learning_rate": 7.743589743589745e-06,
"loss": 0.0084,
"step": 12450
},
{
"epoch": 13.73898678414097,
"grad_norm": 0.22295306622982025,
"learning_rate": 7.717948717948718e-06,
"loss": 0.0098,
"step": 12475
},
{
"epoch": 13.766519823788546,
"grad_norm": 0.1525130271911621,
"learning_rate": 7.692307692307694e-06,
"loss": 0.0094,
"step": 12500
},
{
"epoch": 13.794052863436123,
"grad_norm": 0.062416739761829376,
"learning_rate": 7.666666666666667e-06,
"loss": 0.0083,
"step": 12525
},
{
"epoch": 13.821585903083701,
"grad_norm": 0.09009592980146408,
"learning_rate": 7.641025641025641e-06,
"loss": 0.0094,
"step": 12550
},
{
"epoch": 13.849118942731277,
"grad_norm": 0.33683300018310547,
"learning_rate": 7.615384615384615e-06,
"loss": 0.0087,
"step": 12575
},
{
"epoch": 13.876651982378855,
"grad_norm": 0.029187936335802078,
"learning_rate": 7.58974358974359e-06,
"loss": 0.0103,
"step": 12600
},
{
"epoch": 13.904185022026432,
"grad_norm": 0.24236032366752625,
"learning_rate": 7.564102564102564e-06,
"loss": 0.0119,
"step": 12625
},
{
"epoch": 13.931718061674008,
"grad_norm": 0.042555954307317734,
"learning_rate": 7.538461538461539e-06,
"loss": 0.0107,
"step": 12650
},
{
"epoch": 13.959251101321586,
"grad_norm": 0.05459924414753914,
"learning_rate": 7.512820512820513e-06,
"loss": 0.0096,
"step": 12675
},
{
"epoch": 13.986784140969164,
"grad_norm": 0.3051317632198334,
"learning_rate": 7.487179487179488e-06,
"loss": 0.0115,
"step": 12700
},
{
"epoch": 14.01431718061674,
"grad_norm": 0.07105377316474915,
"learning_rate": 7.461538461538462e-06,
"loss": 0.0118,
"step": 12725
},
{
"epoch": 14.041850220264317,
"grad_norm": 0.32920604944229126,
"learning_rate": 7.435897435897437e-06,
"loss": 0.0136,
"step": 12750
},
{
"epoch": 14.069383259911895,
"grad_norm": 0.7461898922920227,
"learning_rate": 7.410256410256411e-06,
"loss": 0.0123,
"step": 12775
},
{
"epoch": 14.09691629955947,
"grad_norm": 0.08446882665157318,
"learning_rate": 7.384615384615386e-06,
"loss": 0.0086,
"step": 12800
},
{
"epoch": 14.124449339207048,
"grad_norm": 0.275642991065979,
"learning_rate": 7.35897435897436e-06,
"loss": 0.0108,
"step": 12825
},
{
"epoch": 14.151982378854626,
"grad_norm": 0.06443957984447479,
"learning_rate": 7.333333333333333e-06,
"loss": 0.0091,
"step": 12850
},
{
"epoch": 14.179515418502202,
"grad_norm": 0.11947249621152878,
"learning_rate": 7.307692307692308e-06,
"loss": 0.0093,
"step": 12875
},
{
"epoch": 14.20704845814978,
"grad_norm": 0.03607160225510597,
"learning_rate": 7.282051282051282e-06,
"loss": 0.0091,
"step": 12900
},
{
"epoch": 14.234581497797357,
"grad_norm": 0.16806413233280182,
"learning_rate": 7.256410256410257e-06,
"loss": 0.007,
"step": 12925
},
{
"epoch": 14.262114537444933,
"grad_norm": 0.08241688460111618,
"learning_rate": 7.230769230769231e-06,
"loss": 0.0068,
"step": 12950
},
{
"epoch": 14.289647577092511,
"grad_norm": 0.0358104445040226,
"learning_rate": 7.205128205128206e-06,
"loss": 0.0079,
"step": 12975
},
{
"epoch": 14.317180616740089,
"grad_norm": 0.2213512361049652,
"learning_rate": 7.17948717948718e-06,
"loss": 0.0098,
"step": 13000
},
{
"epoch": 14.317180616740089,
"eval_cer": 23.705478431327595,
"eval_loss": 0.8760920763015747,
"eval_runtime": 1295.1667,
"eval_samples_per_second": 8.17,
"eval_steps_per_second": 2.043,
"eval_wer": 83.88495992456389,
"step": 13000
},
{
"epoch": 14.344713656387665,
"grad_norm": 0.3628979027271271,
"learning_rate": 7.153846153846155e-06,
"loss": 0.0078,
"step": 13025
},
{
"epoch": 14.372246696035242,
"grad_norm": 0.0924341008067131,
"learning_rate": 7.128205128205129e-06,
"loss": 0.0088,
"step": 13050
},
{
"epoch": 14.39977973568282,
"grad_norm": 0.6887943148612976,
"learning_rate": 7.102564102564104e-06,
"loss": 0.0088,
"step": 13075
},
{
"epoch": 14.427312775330396,
"grad_norm": 0.3483204245567322,
"learning_rate": 7.076923076923078e-06,
"loss": 0.0085,
"step": 13100
},
{
"epoch": 14.454845814977974,
"grad_norm": 0.0400351881980896,
"learning_rate": 7.051282051282053e-06,
"loss": 0.0094,
"step": 13125
},
{
"epoch": 14.482378854625551,
"grad_norm": 0.15237635374069214,
"learning_rate": 7.025641025641025e-06,
"loss": 0.0092,
"step": 13150
},
{
"epoch": 14.509911894273127,
"grad_norm": 0.1783585399389267,
"learning_rate": 7e-06,
"loss": 0.0068,
"step": 13175
},
{
"epoch": 14.537444933920705,
"grad_norm": 0.21304604411125183,
"learning_rate": 6.974358974358974e-06,
"loss": 0.0084,
"step": 13200
},
{
"epoch": 14.564977973568283,
"grad_norm": 0.01639872044324875,
"learning_rate": 6.948717948717949e-06,
"loss": 0.0065,
"step": 13225
},
{
"epoch": 14.592511013215859,
"grad_norm": 0.19896923005580902,
"learning_rate": 6.923076923076923e-06,
"loss": 0.0062,
"step": 13250
},
{
"epoch": 14.620044052863436,
"grad_norm": 0.03251450136303902,
"learning_rate": 6.897435897435898e-06,
"loss": 0.0076,
"step": 13275
},
{
"epoch": 14.647577092511014,
"grad_norm": 0.06453995406627655,
"learning_rate": 6.871794871794872e-06,
"loss": 0.0075,
"step": 13300
},
{
"epoch": 14.67511013215859,
"grad_norm": 0.24187321960926056,
"learning_rate": 6.846153846153847e-06,
"loss": 0.0063,
"step": 13325
},
{
"epoch": 14.702643171806168,
"grad_norm": 0.3047979772090912,
"learning_rate": 6.820512820512821e-06,
"loss": 0.0091,
"step": 13350
},
{
"epoch": 14.730176211453745,
"grad_norm": 0.06145111471414566,
"learning_rate": 6.794871794871796e-06,
"loss": 0.0066,
"step": 13375
},
{
"epoch": 14.757709251101321,
"grad_norm": 0.2679343819618225,
"learning_rate": 6.76923076923077e-06,
"loss": 0.0098,
"step": 13400
},
{
"epoch": 14.785242290748899,
"grad_norm": 0.030670566484332085,
"learning_rate": 6.743589743589745e-06,
"loss": 0.0085,
"step": 13425
},
{
"epoch": 14.812775330396477,
"grad_norm": 0.332878440618515,
"learning_rate": 6.717948717948718e-06,
"loss": 0.01,
"step": 13450
},
{
"epoch": 14.840308370044053,
"grad_norm": 0.2726658582687378,
"learning_rate": 6.692307692307692e-06,
"loss": 0.009,
"step": 13475
},
{
"epoch": 14.86784140969163,
"grad_norm": 0.13588112592697144,
"learning_rate": 6.666666666666667e-06,
"loss": 0.0068,
"step": 13500
},
{
"epoch": 14.895374449339208,
"grad_norm": 0.09434377402067184,
"learning_rate": 6.641025641025641e-06,
"loss": 0.0083,
"step": 13525
},
{
"epoch": 14.922907488986784,
"grad_norm": 0.18095168471336365,
"learning_rate": 6.615384615384616e-06,
"loss": 0.0078,
"step": 13550
},
{
"epoch": 14.950440528634362,
"grad_norm": 0.05562664195895195,
"learning_rate": 6.58974358974359e-06,
"loss": 0.0072,
"step": 13575
},
{
"epoch": 14.97797356828194,
"grad_norm": 0.15413448214530945,
"learning_rate": 6.564102564102565e-06,
"loss": 0.0112,
"step": 13600
},
{
"epoch": 15.005506607929515,
"grad_norm": 0.12938186526298523,
"learning_rate": 6.538461538461539e-06,
"loss": 0.0079,
"step": 13625
},
{
"epoch": 15.033039647577093,
"grad_norm": 0.07106704264879227,
"learning_rate": 6.512820512820514e-06,
"loss": 0.0078,
"step": 13650
},
{
"epoch": 15.060572687224669,
"grad_norm": 0.3770430386066437,
"learning_rate": 6.487179487179488e-06,
"loss": 0.0085,
"step": 13675
},
{
"epoch": 15.088105726872246,
"grad_norm": 0.03784358873963356,
"learning_rate": 6.461538461538463e-06,
"loss": 0.007,
"step": 13700
},
{
"epoch": 15.115638766519824,
"grad_norm": 0.10024132579565048,
"learning_rate": 6.435897435897437e-06,
"loss": 0.0087,
"step": 13725
},
{
"epoch": 15.1431718061674,
"grad_norm": 0.2917975187301636,
"learning_rate": 6.410256410256412e-06,
"loss": 0.0085,
"step": 13750
},
{
"epoch": 15.170704845814978,
"grad_norm": 0.4482714533805847,
"learning_rate": 6.384615384615384e-06,
"loss": 0.0062,
"step": 13775
},
{
"epoch": 15.198237885462555,
"grad_norm": 0.04433772340416908,
"learning_rate": 6.358974358974359e-06,
"loss": 0.0052,
"step": 13800
},
{
"epoch": 15.225770925110131,
"grad_norm": 0.1545044481754303,
"learning_rate": 6.333333333333333e-06,
"loss": 0.0075,
"step": 13825
},
{
"epoch": 15.253303964757709,
"grad_norm": 0.2501870095729828,
"learning_rate": 6.307692307692308e-06,
"loss": 0.0066,
"step": 13850
},
{
"epoch": 15.280837004405287,
"grad_norm": 0.14185580611228943,
"learning_rate": 6.282051282051282e-06,
"loss": 0.0073,
"step": 13875
},
{
"epoch": 15.308370044052863,
"grad_norm": 0.166921004652977,
"learning_rate": 6.256410256410257e-06,
"loss": 0.0105,
"step": 13900
},
{
"epoch": 15.33590308370044,
"grad_norm": 0.055341098457574844,
"learning_rate": 6.230769230769231e-06,
"loss": 0.0086,
"step": 13925
},
{
"epoch": 15.363436123348018,
"grad_norm": 0.28348565101623535,
"learning_rate": 6.205128205128206e-06,
"loss": 0.0089,
"step": 13950
},
{
"epoch": 15.390969162995594,
"grad_norm": 0.10073395073413849,
"learning_rate": 6.17948717948718e-06,
"loss": 0.0084,
"step": 13975
},
{
"epoch": 15.418502202643172,
"grad_norm": 0.05402739346027374,
"learning_rate": 6.153846153846155e-06,
"loss": 0.0076,
"step": 14000
},
{
"epoch": 15.418502202643172,
"eval_cer": 23.665356000596105,
"eval_loss": 0.8774716854095459,
"eval_runtime": 1296.7729,
"eval_samples_per_second": 8.159,
"eval_steps_per_second": 2.04,
"eval_wer": 83.18717586044319,
"step": 14000
},
{
"epoch": 15.44603524229075,
"grad_norm": 0.09686881303787231,
"learning_rate": 6.128205128205129e-06,
"loss": 0.0078,
"step": 14025
},
{
"epoch": 15.473568281938325,
"grad_norm": 0.10279793292284012,
"learning_rate": 6.102564102564104e-06,
"loss": 0.0096,
"step": 14050
},
{
"epoch": 15.501101321585903,
"grad_norm": 0.1425463706254959,
"learning_rate": 6.076923076923077e-06,
"loss": 0.0077,
"step": 14075
},
{
"epoch": 15.52863436123348,
"grad_norm": 0.1172114908695221,
"learning_rate": 6.051282051282051e-06,
"loss": 0.0046,
"step": 14100
},
{
"epoch": 15.556167400881057,
"grad_norm": 0.03888353705406189,
"learning_rate": 6.025641025641026e-06,
"loss": 0.0052,
"step": 14125
},
{
"epoch": 15.583700440528634,
"grad_norm": 0.2044220268726349,
"learning_rate": 6e-06,
"loss": 0.0065,
"step": 14150
},
{
"epoch": 15.611233480176212,
"grad_norm": 0.18413954973220825,
"learning_rate": 5.974358974358975e-06,
"loss": 0.0051,
"step": 14175
},
{
"epoch": 15.638766519823788,
"grad_norm": 0.16837269067764282,
"learning_rate": 5.948717948717949e-06,
"loss": 0.0109,
"step": 14200
},
{
"epoch": 15.666299559471366,
"grad_norm": 0.06798551231622696,
"learning_rate": 5.923076923076924e-06,
"loss": 0.0048,
"step": 14225
},
{
"epoch": 15.693832599118943,
"grad_norm": 0.1540171355009079,
"learning_rate": 5.897435897435898e-06,
"loss": 0.0055,
"step": 14250
},
{
"epoch": 15.72136563876652,
"grad_norm": 0.028167154639959335,
"learning_rate": 5.871794871794873e-06,
"loss": 0.0045,
"step": 14275
},
{
"epoch": 15.748898678414097,
"grad_norm": 0.12515689432621002,
"learning_rate": 5.846153846153847e-06,
"loss": 0.0047,
"step": 14300
},
{
"epoch": 15.776431718061675,
"grad_norm": 0.04906751960515976,
"learning_rate": 5.820512820512822e-06,
"loss": 0.0079,
"step": 14325
},
{
"epoch": 15.80396475770925,
"grad_norm": 0.05166306719183922,
"learning_rate": 5.794871794871796e-06,
"loss": 0.0102,
"step": 14350
},
{
"epoch": 15.831497797356828,
"grad_norm": 0.11004568636417389,
"learning_rate": 5.769230769230769e-06,
"loss": 0.0067,
"step": 14375
},
{
"epoch": 15.859030837004406,
"grad_norm": 0.1999882608652115,
"learning_rate": 5.743589743589743e-06,
"loss": 0.007,
"step": 14400
},
{
"epoch": 15.886563876651982,
"grad_norm": 0.34716781973838806,
"learning_rate": 5.717948717948718e-06,
"loss": 0.0076,
"step": 14425
},
{
"epoch": 15.91409691629956,
"grad_norm": 0.31926947832107544,
"learning_rate": 5.692307692307692e-06,
"loss": 0.0073,
"step": 14450
},
{
"epoch": 15.941629955947137,
"grad_norm": 0.1822180300951004,
"learning_rate": 5.666666666666667e-06,
"loss": 0.0061,
"step": 14475
},
{
"epoch": 15.969162995594713,
"grad_norm": 0.029914982616901398,
"learning_rate": 5.641025641025641e-06,
"loss": 0.0079,
"step": 14500
},
{
"epoch": 15.996696035242291,
"grad_norm": 0.045799557119607925,
"learning_rate": 5.615384615384616e-06,
"loss": 0.0075,
"step": 14525
},
{
"epoch": 16.024229074889867,
"grad_norm": 0.2783111333847046,
"learning_rate": 5.58974358974359e-06,
"loss": 0.0081,
"step": 14550
},
{
"epoch": 16.051762114537446,
"grad_norm": 0.059823885560035706,
"learning_rate": 5.564102564102565e-06,
"loss": 0.0063,
"step": 14575
},
{
"epoch": 16.079295154185022,
"grad_norm": 0.15732045471668243,
"learning_rate": 5.538461538461539e-06,
"loss": 0.0077,
"step": 14600
},
{
"epoch": 16.106828193832598,
"grad_norm": 0.028589140623807907,
"learning_rate": 5.512820512820514e-06,
"loss": 0.0054,
"step": 14625
},
{
"epoch": 16.134361233480178,
"grad_norm": 0.10421264916658401,
"learning_rate": 5.487179487179488e-06,
"loss": 0.0048,
"step": 14650
},
{
"epoch": 16.161894273127754,
"grad_norm": 0.0383140854537487,
"learning_rate": 5.461538461538461e-06,
"loss": 0.0055,
"step": 14675
},
{
"epoch": 16.18942731277533,
"grad_norm": 0.10105977952480316,
"learning_rate": 5.435897435897436e-06,
"loss": 0.0044,
"step": 14700
},
{
"epoch": 16.21696035242291,
"grad_norm": 0.027039742097258568,
"learning_rate": 5.41025641025641e-06,
"loss": 0.0058,
"step": 14725
},
{
"epoch": 16.244493392070485,
"grad_norm": 0.037745460867881775,
"learning_rate": 5.384615384615385e-06,
"loss": 0.0048,
"step": 14750
},
{
"epoch": 16.27202643171806,
"grad_norm": 0.09512659162282944,
"learning_rate": 5.358974358974359e-06,
"loss": 0.0073,
"step": 14775
},
{
"epoch": 16.29955947136564,
"grad_norm": 0.4294408857822418,
"learning_rate": 5.333333333333334e-06,
"loss": 0.0054,
"step": 14800
},
{
"epoch": 16.327092511013216,
"grad_norm": 0.06851428002119064,
"learning_rate": 5.307692307692308e-06,
"loss": 0.0046,
"step": 14825
},
{
"epoch": 16.354625550660792,
"grad_norm": 0.021705007180571556,
"learning_rate": 5.282051282051283e-06,
"loss": 0.005,
"step": 14850
},
{
"epoch": 16.38215859030837,
"grad_norm": 0.49824318289756775,
"learning_rate": 5.256410256410257e-06,
"loss": 0.0061,
"step": 14875
},
{
"epoch": 16.409691629955947,
"grad_norm": 0.3271447718143463,
"learning_rate": 5.230769230769232e-06,
"loss": 0.0051,
"step": 14900
},
{
"epoch": 16.437224669603523,
"grad_norm": 0.09235868602991104,
"learning_rate": 5.205128205128206e-06,
"loss": 0.0037,
"step": 14925
},
{
"epoch": 16.464757709251103,
"grad_norm": 0.4005078971385956,
"learning_rate": 5.179487179487181e-06,
"loss": 0.0087,
"step": 14950
},
{
"epoch": 16.49229074889868,
"grad_norm": 0.2035885453224182,
"learning_rate": 5.1538461538461534e-06,
"loss": 0.0116,
"step": 14975
},
{
"epoch": 16.519823788546255,
"grad_norm": 0.13736362755298615,
"learning_rate": 5.128205128205128e-06,
"loss": 0.0044,
"step": 15000
},
{
"epoch": 16.519823788546255,
"eval_cer": 23.58683067187876,
"eval_loss": 0.8781108856201172,
"eval_runtime": 1297.7364,
"eval_samples_per_second": 8.153,
"eval_steps_per_second": 2.039,
"eval_wer": 83.07402168788308,
"step": 15000
},
{
"epoch": 16.547356828193834,
"grad_norm": 0.06933388859033585,
"learning_rate": 5.1025641025641024e-06,
"loss": 0.0075,
"step": 15025
},
{
"epoch": 16.57488986784141,
"grad_norm": 0.13286051154136658,
"learning_rate": 5.076923076923077e-06,
"loss": 0.0051,
"step": 15050
},
{
"epoch": 16.602422907488986,
"grad_norm": 0.026176370680332184,
"learning_rate": 5.051282051282051e-06,
"loss": 0.0033,
"step": 15075
},
{
"epoch": 16.629955947136565,
"grad_norm": 0.29031485319137573,
"learning_rate": 5.025641025641026e-06,
"loss": 0.0079,
"step": 15100
},
{
"epoch": 16.65748898678414,
"grad_norm": 0.020291967317461967,
"learning_rate": 5e-06,
"loss": 0.005,
"step": 15125
},
{
"epoch": 16.685022026431717,
"grad_norm": 0.14216738939285278,
"learning_rate": 4.974358974358975e-06,
"loss": 0.0033,
"step": 15150
},
{
"epoch": 16.712555066079297,
"grad_norm": 0.012860552407801151,
"learning_rate": 4.948717948717949e-06,
"loss": 0.0051,
"step": 15175
},
{
"epoch": 16.740088105726873,
"grad_norm": 0.10258791595697403,
"learning_rate": 4.923076923076924e-06,
"loss": 0.0058,
"step": 15200
},
{
"epoch": 16.76762114537445,
"grad_norm": 0.057229600846767426,
"learning_rate": 4.8974358974358975e-06,
"loss": 0.0063,
"step": 15225
},
{
"epoch": 16.795154185022028,
"grad_norm": 0.04477262496948242,
"learning_rate": 4.871794871794872e-06,
"loss": 0.0064,
"step": 15250
},
{
"epoch": 16.822687224669604,
"grad_norm": 0.05280523747205734,
"learning_rate": 4.8461538461538465e-06,
"loss": 0.005,
"step": 15275
},
{
"epoch": 16.85022026431718,
"grad_norm": 0.2496304214000702,
"learning_rate": 4.820512820512821e-06,
"loss": 0.0046,
"step": 15300
},
{
"epoch": 16.87775330396476,
"grad_norm": 0.05990584194660187,
"learning_rate": 4.7948717948717955e-06,
"loss": 0.0034,
"step": 15325
},
{
"epoch": 16.905286343612335,
"grad_norm": 0.03158105909824371,
"learning_rate": 4.76923076923077e-06,
"loss": 0.0039,
"step": 15350
},
{
"epoch": 16.93281938325991,
"grad_norm": 0.04606786370277405,
"learning_rate": 4.743589743589744e-06,
"loss": 0.0023,
"step": 15375
},
{
"epoch": 16.96035242290749,
"grad_norm": 0.06241815909743309,
"learning_rate": 4.717948717948718e-06,
"loss": 0.0043,
"step": 15400
},
{
"epoch": 16.987885462555067,
"grad_norm": 0.10293944925069809,
"learning_rate": 4.692307692307693e-06,
"loss": 0.0049,
"step": 15425
},
{
"epoch": 17.015418502202643,
"grad_norm": 0.10291476547718048,
"learning_rate": 4.666666666666667e-06,
"loss": 0.0033,
"step": 15450
},
{
"epoch": 17.042951541850222,
"grad_norm": 0.021487778052687645,
"learning_rate": 4.641025641025642e-06,
"loss": 0.0039,
"step": 15475
},
{
"epoch": 17.070484581497798,
"grad_norm": 0.0317191444337368,
"learning_rate": 4.615384615384616e-06,
"loss": 0.0041,
"step": 15500
},
{
"epoch": 17.098017621145374,
"grad_norm": 0.010435504838824272,
"learning_rate": 4.58974358974359e-06,
"loss": 0.0028,
"step": 15525
},
{
"epoch": 17.125550660792953,
"grad_norm": 0.03608101233839989,
"learning_rate": 4.564102564102564e-06,
"loss": 0.0036,
"step": 15550
},
{
"epoch": 17.15308370044053,
"grad_norm": 0.04067333787679672,
"learning_rate": 4.538461538461539e-06,
"loss": 0.0037,
"step": 15575
},
{
"epoch": 17.180616740088105,
"grad_norm": 0.011465054005384445,
"learning_rate": 4.512820512820513e-06,
"loss": 0.0029,
"step": 15600
},
{
"epoch": 17.208149779735685,
"grad_norm": 0.016354389488697052,
"learning_rate": 4.487179487179488e-06,
"loss": 0.0025,
"step": 15625
},
{
"epoch": 17.23568281938326,
"grad_norm": 0.023780396208167076,
"learning_rate": 4.461538461538462e-06,
"loss": 0.0027,
"step": 15650
},
{
"epoch": 17.263215859030836,
"grad_norm": 0.12193647772073746,
"learning_rate": 4.435897435897436e-06,
"loss": 0.0038,
"step": 15675
},
{
"epoch": 17.290748898678412,
"grad_norm": 0.20270489156246185,
"learning_rate": 4.4102564102564104e-06,
"loss": 0.0024,
"step": 15700
},
{
"epoch": 17.318281938325992,
"grad_norm": 0.21343407034873962,
"learning_rate": 4.384615384615385e-06,
"loss": 0.0033,
"step": 15725
},
{
"epoch": 17.345814977973568,
"grad_norm": 0.17167609930038452,
"learning_rate": 4.358974358974359e-06,
"loss": 0.0038,
"step": 15750
},
{
"epoch": 17.373348017621144,
"grad_norm": 0.07102500647306442,
"learning_rate": 4.333333333333334e-06,
"loss": 0.0027,
"step": 15775
},
{
"epoch": 17.400881057268723,
"grad_norm": 0.01935497298836708,
"learning_rate": 4.307692307692308e-06,
"loss": 0.003,
"step": 15800
},
{
"epoch": 17.4284140969163,
"grad_norm": 0.013304700143635273,
"learning_rate": 4.282051282051282e-06,
"loss": 0.0041,
"step": 15825
},
{
"epoch": 17.455947136563875,
"grad_norm": 0.01759173534810543,
"learning_rate": 4.2564102564102566e-06,
"loss": 0.0037,
"step": 15850
},
{
"epoch": 17.483480176211454,
"grad_norm": 0.22442099452018738,
"learning_rate": 4.230769230769231e-06,
"loss": 0.0041,
"step": 15875
},
{
"epoch": 17.51101321585903,
"grad_norm": 0.08358743786811829,
"learning_rate": 4.2051282051282055e-06,
"loss": 0.0031,
"step": 15900
},
{
"epoch": 17.538546255506606,
"grad_norm": 0.020445672795176506,
"learning_rate": 4.17948717948718e-06,
"loss": 0.0036,
"step": 15925
},
{
"epoch": 17.566079295154186,
"grad_norm": 0.27063530683517456,
"learning_rate": 4.1538461538461545e-06,
"loss": 0.0045,
"step": 15950
},
{
"epoch": 17.59361233480176,
"grad_norm": 0.16294091939926147,
"learning_rate": 4.128205128205128e-06,
"loss": 0.0043,
"step": 15975
},
{
"epoch": 17.621145374449338,
"grad_norm": 0.012531392276287079,
"learning_rate": 4.102564102564103e-06,
"loss": 0.005,
"step": 16000
},
{
"epoch": 17.621145374449338,
"eval_cer": 22.748845047172512,
"eval_loss": 0.8773518204689026,
"eval_runtime": 1297.9232,
"eval_samples_per_second": 8.152,
"eval_steps_per_second": 2.039,
"eval_wer": 82.3102310231023,
"step": 16000
},
{
"epoch": 17.648678414096917,
"grad_norm": 0.01578434184193611,
"learning_rate": 4.076923076923077e-06,
"loss": 0.0031,
"step": 16025
},
{
"epoch": 17.676211453744493,
"grad_norm": 0.21798987686634064,
"learning_rate": 4.051282051282052e-06,
"loss": 0.0029,
"step": 16050
},
{
"epoch": 17.70374449339207,
"grad_norm": 0.015405426733195782,
"learning_rate": 4.025641025641026e-06,
"loss": 0.0031,
"step": 16075
},
{
"epoch": 17.73127753303965,
"grad_norm": 0.0639905333518982,
"learning_rate": 4.000000000000001e-06,
"loss": 0.0045,
"step": 16100
},
{
"epoch": 17.758810572687224,
"grad_norm": 0.03249204158782959,
"learning_rate": 3.974358974358974e-06,
"loss": 0.0025,
"step": 16125
},
{
"epoch": 17.7863436123348,
"grad_norm": 0.007253581192344427,
"learning_rate": 3.948717948717949e-06,
"loss": 0.0026,
"step": 16150
},
{
"epoch": 17.81387665198238,
"grad_norm": 0.052789948880672455,
"learning_rate": 3.923076923076923e-06,
"loss": 0.0023,
"step": 16175
},
{
"epoch": 17.841409691629956,
"grad_norm": 0.028969332575798035,
"learning_rate": 3.897435897435898e-06,
"loss": 0.0022,
"step": 16200
},
{
"epoch": 17.86894273127753,
"grad_norm": 0.040875017642974854,
"learning_rate": 3.871794871794872e-06,
"loss": 0.0032,
"step": 16225
},
{
"epoch": 17.89647577092511,
"grad_norm": 0.021907728165388107,
"learning_rate": 3.846153846153847e-06,
"loss": 0.0048,
"step": 16250
},
{
"epoch": 17.924008810572687,
"grad_norm": 0.03670613095164299,
"learning_rate": 3.8205128205128204e-06,
"loss": 0.004,
"step": 16275
},
{
"epoch": 17.951541850220263,
"grad_norm": 0.03614376112818718,
"learning_rate": 3.794871794871795e-06,
"loss": 0.0025,
"step": 16300
},
{
"epoch": 17.979074889867842,
"grad_norm": 0.028139924630522728,
"learning_rate": 3.7692307692307694e-06,
"loss": 0.0019,
"step": 16325
},
{
"epoch": 18.006607929515418,
"grad_norm": 0.013178820721805096,
"learning_rate": 3.743589743589744e-06,
"loss": 0.0016,
"step": 16350
},
{
"epoch": 18.034140969162994,
"grad_norm": 0.014315299689769745,
"learning_rate": 3.7179487179487184e-06,
"loss": 0.0021,
"step": 16375
},
{
"epoch": 18.061674008810574,
"grad_norm": 0.15686722099781036,
"learning_rate": 3.692307692307693e-06,
"loss": 0.0019,
"step": 16400
},
{
"epoch": 18.08920704845815,
"grad_norm": 0.07895169407129288,
"learning_rate": 3.6666666666666666e-06,
"loss": 0.0032,
"step": 16425
},
{
"epoch": 18.116740088105725,
"grad_norm": 0.01955692283809185,
"learning_rate": 3.641025641025641e-06,
"loss": 0.0031,
"step": 16450
},
{
"epoch": 18.144273127753305,
"grad_norm": 0.07983898371458054,
"learning_rate": 3.6153846153846156e-06,
"loss": 0.0024,
"step": 16475
},
{
"epoch": 18.17180616740088,
"grad_norm": 0.012052379548549652,
"learning_rate": 3.58974358974359e-06,
"loss": 0.0013,
"step": 16500
},
{
"epoch": 18.199339207048457,
"grad_norm": 0.00846129097044468,
"learning_rate": 3.5641025641025646e-06,
"loss": 0.0024,
"step": 16525
},
{
"epoch": 18.226872246696036,
"grad_norm": 0.015683434903621674,
"learning_rate": 3.538461538461539e-06,
"loss": 0.0028,
"step": 16550
},
{
"epoch": 18.254405286343612,
"grad_norm": 0.0070321750827133656,
"learning_rate": 3.5128205128205127e-06,
"loss": 0.0021,
"step": 16575
},
{
"epoch": 18.281938325991188,
"grad_norm": 0.037714555859565735,
"learning_rate": 3.487179487179487e-06,
"loss": 0.0012,
"step": 16600
},
{
"epoch": 18.309471365638768,
"grad_norm": 0.010739946737885475,
"learning_rate": 3.4615384615384617e-06,
"loss": 0.0011,
"step": 16625
},
{
"epoch": 18.337004405286343,
"grad_norm": 0.009480941109359264,
"learning_rate": 3.435897435897436e-06,
"loss": 0.0018,
"step": 16650
},
{
"epoch": 18.36453744493392,
"grad_norm": 0.011291285045444965,
"learning_rate": 3.4102564102564107e-06,
"loss": 0.0023,
"step": 16675
},
{
"epoch": 18.3920704845815,
"grad_norm": 0.014357962645590305,
"learning_rate": 3.384615384615385e-06,
"loss": 0.002,
"step": 16700
},
{
"epoch": 18.419603524229075,
"grad_norm": 0.005591503344476223,
"learning_rate": 3.358974358974359e-06,
"loss": 0.0015,
"step": 16725
},
{
"epoch": 18.44713656387665,
"grad_norm": 0.014310234226286411,
"learning_rate": 3.3333333333333333e-06,
"loss": 0.0027,
"step": 16750
},
{
"epoch": 18.47466960352423,
"grad_norm": 0.026868004351854324,
"learning_rate": 3.307692307692308e-06,
"loss": 0.0018,
"step": 16775
},
{
"epoch": 18.502202643171806,
"grad_norm": 0.011405620723962784,
"learning_rate": 3.2820512820512823e-06,
"loss": 0.0021,
"step": 16800
},
{
"epoch": 18.529735682819382,
"grad_norm": 0.007671520579606295,
"learning_rate": 3.256410256410257e-06,
"loss": 0.0021,
"step": 16825
},
{
"epoch": 18.55726872246696,
"grad_norm": 0.09985347092151642,
"learning_rate": 3.2307692307692313e-06,
"loss": 0.0045,
"step": 16850
},
{
"epoch": 18.584801762114537,
"grad_norm": 0.0127689428627491,
"learning_rate": 3.205128205128206e-06,
"loss": 0.0031,
"step": 16875
},
{
"epoch": 18.612334801762113,
"grad_norm": 0.12681765854358673,
"learning_rate": 3.1794871794871795e-06,
"loss": 0.0013,
"step": 16900
},
{
"epoch": 18.639867841409693,
"grad_norm": 0.008006567135453224,
"learning_rate": 3.153846153846154e-06,
"loss": 0.0015,
"step": 16925
},
{
"epoch": 18.66740088105727,
"grad_norm": 0.020465383306145668,
"learning_rate": 3.1282051282051284e-06,
"loss": 0.0016,
"step": 16950
},
{
"epoch": 18.694933920704845,
"grad_norm": 0.00684625469148159,
"learning_rate": 3.102564102564103e-06,
"loss": 0.0013,
"step": 16975
},
{
"epoch": 18.722466960352424,
"grad_norm": 0.011861232109367847,
"learning_rate": 3.0769230769230774e-06,
"loss": 0.0026,
"step": 17000
},
{
"epoch": 18.722466960352424,
"eval_cer": 22.531610743640595,
"eval_loss": 0.8913528919219971,
"eval_runtime": 1295.7696,
"eval_samples_per_second": 8.166,
"eval_steps_per_second": 2.042,
"eval_wer": 82.14992927864215,
"step": 17000
},
{
"epoch": 18.75,
"grad_norm": 0.017606543377041817,
"learning_rate": 3.051282051282052e-06,
"loss": 0.0032,
"step": 17025
},
{
"epoch": 18.777533039647576,
"grad_norm": 0.009970780462026596,
"learning_rate": 3.0256410256410256e-06,
"loss": 0.0018,
"step": 17050
},
{
"epoch": 18.805066079295155,
"grad_norm": 0.07100367546081543,
"learning_rate": 3e-06,
"loss": 0.0015,
"step": 17075
},
{
"epoch": 18.83259911894273,
"grad_norm": 0.010995105840265751,
"learning_rate": 2.9743589743589746e-06,
"loss": 0.0024,
"step": 17100
},
{
"epoch": 18.860132158590307,
"grad_norm": 0.20596833527088165,
"learning_rate": 2.948717948717949e-06,
"loss": 0.0027,
"step": 17125
},
{
"epoch": 18.887665198237887,
"grad_norm": 0.014417466707527637,
"learning_rate": 2.9230769230769236e-06,
"loss": 0.0018,
"step": 17150
},
{
"epoch": 18.915198237885463,
"grad_norm": 0.025823669508099556,
"learning_rate": 2.897435897435898e-06,
"loss": 0.0014,
"step": 17175
},
{
"epoch": 18.94273127753304,
"grad_norm": 0.009362194687128067,
"learning_rate": 2.8717948717948717e-06,
"loss": 0.0027,
"step": 17200
},
{
"epoch": 18.970264317180618,
"grad_norm": 0.07939445227384567,
"learning_rate": 2.846153846153846e-06,
"loss": 0.0017,
"step": 17225
},
{
"epoch": 18.997797356828194,
"grad_norm": 0.011641057208180428,
"learning_rate": 2.8205128205128207e-06,
"loss": 0.0013,
"step": 17250
},
{
"epoch": 19.02533039647577,
"grad_norm": 0.006191305350512266,
"learning_rate": 2.794871794871795e-06,
"loss": 0.0018,
"step": 17275
},
{
"epoch": 19.05286343612335,
"grad_norm": 0.007426435127854347,
"learning_rate": 2.7692307692307697e-06,
"loss": 0.001,
"step": 17300
},
{
"epoch": 19.080396475770925,
"grad_norm": 0.04474742338061333,
"learning_rate": 2.743589743589744e-06,
"loss": 0.003,
"step": 17325
},
{
"epoch": 19.1079295154185,
"grad_norm": 0.026934364810585976,
"learning_rate": 2.717948717948718e-06,
"loss": 0.0019,
"step": 17350
},
{
"epoch": 19.13546255506608,
"grad_norm": 0.008537307381629944,
"learning_rate": 2.6923076923076923e-06,
"loss": 0.001,
"step": 17375
},
{
"epoch": 19.162995594713657,
"grad_norm": 0.06050959601998329,
"learning_rate": 2.666666666666667e-06,
"loss": 0.0016,
"step": 17400
},
{
"epoch": 19.190528634361232,
"grad_norm": 0.010981615632772446,
"learning_rate": 2.6410256410256413e-06,
"loss": 0.0012,
"step": 17425
},
{
"epoch": 19.218061674008812,
"grad_norm": 0.01226646639406681,
"learning_rate": 2.615384615384616e-06,
"loss": 0.0012,
"step": 17450
},
{
"epoch": 19.245594713656388,
"grad_norm": 0.00811366643756628,
"learning_rate": 2.5897435897435903e-06,
"loss": 0.0013,
"step": 17475
},
{
"epoch": 19.273127753303964,
"grad_norm": 0.009269953705370426,
"learning_rate": 2.564102564102564e-06,
"loss": 0.0011,
"step": 17500
},
{
"epoch": 19.300660792951543,
"grad_norm": 0.00818623322993517,
"learning_rate": 2.5384615384615385e-06,
"loss": 0.0024,
"step": 17525
},
{
"epoch": 19.32819383259912,
"grad_norm": 0.011427664197981358,
"learning_rate": 2.512820512820513e-06,
"loss": 0.001,
"step": 17550
},
{
"epoch": 19.355726872246695,
"grad_norm": 0.009334642440080643,
"learning_rate": 2.4871794871794875e-06,
"loss": 0.0018,
"step": 17575
},
{
"epoch": 19.383259911894275,
"grad_norm": 0.0788646936416626,
"learning_rate": 2.461538461538462e-06,
"loss": 0.0014,
"step": 17600
},
{
"epoch": 19.41079295154185,
"grad_norm": 0.07220646739006042,
"learning_rate": 2.435897435897436e-06,
"loss": 0.0015,
"step": 17625
},
{
"epoch": 19.438325991189426,
"grad_norm": 0.01058241818100214,
"learning_rate": 2.4102564102564105e-06,
"loss": 0.0023,
"step": 17650
},
{
"epoch": 19.465859030837006,
"grad_norm": 0.10551930218935013,
"learning_rate": 2.384615384615385e-06,
"loss": 0.0017,
"step": 17675
},
{
"epoch": 19.493392070484582,
"grad_norm": 0.009333525784313679,
"learning_rate": 2.358974358974359e-06,
"loss": 0.0015,
"step": 17700
},
{
"epoch": 19.520925110132158,
"grad_norm": 0.011308999732136726,
"learning_rate": 2.3333333333333336e-06,
"loss": 0.0025,
"step": 17725
},
{
"epoch": 19.548458149779737,
"grad_norm": 0.009809375740587711,
"learning_rate": 2.307692307692308e-06,
"loss": 0.0014,
"step": 17750
},
{
"epoch": 19.575991189427313,
"grad_norm": 0.01724208891391754,
"learning_rate": 2.282051282051282e-06,
"loss": 0.0016,
"step": 17775
},
{
"epoch": 19.60352422907489,
"grad_norm": 0.014492113143205643,
"learning_rate": 2.2564102564102566e-06,
"loss": 0.0009,
"step": 17800
},
{
"epoch": 19.63105726872247,
"grad_norm": 0.010606609284877777,
"learning_rate": 2.230769230769231e-06,
"loss": 0.0014,
"step": 17825
},
{
"epoch": 19.658590308370044,
"grad_norm": 0.009576292708516121,
"learning_rate": 2.2051282051282052e-06,
"loss": 0.0008,
"step": 17850
},
{
"epoch": 19.68612334801762,
"grad_norm": 0.005126140080392361,
"learning_rate": 2.1794871794871797e-06,
"loss": 0.0012,
"step": 17875
},
{
"epoch": 19.7136563876652,
"grad_norm": 0.006038044113665819,
"learning_rate": 2.153846153846154e-06,
"loss": 0.0017,
"step": 17900
},
{
"epoch": 19.741189427312776,
"grad_norm": 0.15704374015331268,
"learning_rate": 2.1282051282051283e-06,
"loss": 0.0017,
"step": 17925
},
{
"epoch": 19.76872246696035,
"grad_norm": 0.008588683791458607,
"learning_rate": 2.1025641025641028e-06,
"loss": 0.0009,
"step": 17950
},
{
"epoch": 19.79625550660793,
"grad_norm": 0.023944512009620667,
"learning_rate": 2.0769230769230773e-06,
"loss": 0.0011,
"step": 17975
},
{
"epoch": 19.823788546255507,
"grad_norm": 0.013375409878790379,
"learning_rate": 2.0512820512820513e-06,
"loss": 0.0015,
"step": 18000
},
{
"epoch": 19.823788546255507,
"eval_cer": 22.31323008494492,
"eval_loss": 0.8889961838722229,
"eval_runtime": 1299.2034,
"eval_samples_per_second": 8.144,
"eval_steps_per_second": 2.037,
"eval_wer": 81.98962753418199,
"step": 18000
},
{
"epoch": 19.851321585903083,
"grad_norm": 0.0055061751045286655,
"learning_rate": 2.025641025641026e-06,
"loss": 0.002,
"step": 18025
},
{
"epoch": 19.878854625550662,
"grad_norm": 0.00620003929361701,
"learning_rate": 2.0000000000000003e-06,
"loss": 0.001,
"step": 18050
},
{
"epoch": 19.90638766519824,
"grad_norm": 0.012960962951183319,
"learning_rate": 1.9743589743589744e-06,
"loss": 0.001,
"step": 18075
},
{
"epoch": 19.933920704845814,
"grad_norm": 0.00596979632973671,
"learning_rate": 1.948717948717949e-06,
"loss": 0.0016,
"step": 18100
},
{
"epoch": 19.961453744493394,
"grad_norm": 0.00921118725091219,
"learning_rate": 1.9230769230769234e-06,
"loss": 0.0015,
"step": 18125
},
{
"epoch": 19.98898678414097,
"grad_norm": 0.008434862829744816,
"learning_rate": 1.8974358974358975e-06,
"loss": 0.0012,
"step": 18150
},
{
"epoch": 20.016519823788546,
"grad_norm": 0.004028468858450651,
"learning_rate": 1.871794871794872e-06,
"loss": 0.0012,
"step": 18175
},
{
"epoch": 20.044052863436125,
"grad_norm": 0.005844887346029282,
"learning_rate": 1.8461538461538465e-06,
"loss": 0.0009,
"step": 18200
},
{
"epoch": 20.0715859030837,
"grad_norm": 0.0038028398994356394,
"learning_rate": 1.8205128205128205e-06,
"loss": 0.0008,
"step": 18225
},
{
"epoch": 20.099118942731277,
"grad_norm": 0.0037237314973026514,
"learning_rate": 1.794871794871795e-06,
"loss": 0.0009,
"step": 18250
},
{
"epoch": 20.126651982378856,
"grad_norm": 0.003872903762385249,
"learning_rate": 1.7692307692307695e-06,
"loss": 0.0008,
"step": 18275
},
{
"epoch": 20.154185022026432,
"grad_norm": 0.007000759243965149,
"learning_rate": 1.7435897435897436e-06,
"loss": 0.0007,
"step": 18300
},
{
"epoch": 20.181718061674008,
"grad_norm": 0.12840576469898224,
"learning_rate": 1.717948717948718e-06,
"loss": 0.0009,
"step": 18325
},
{
"epoch": 20.209251101321588,
"grad_norm": 0.004936088342219591,
"learning_rate": 1.6923076923076926e-06,
"loss": 0.001,
"step": 18350
},
{
"epoch": 20.236784140969164,
"grad_norm": 0.008810814470052719,
"learning_rate": 1.6666666666666667e-06,
"loss": 0.0013,
"step": 18375
},
{
"epoch": 20.26431718061674,
"grad_norm": 0.008568311110138893,
"learning_rate": 1.6410256410256412e-06,
"loss": 0.0011,
"step": 18400
},
{
"epoch": 20.291850220264315,
"grad_norm": 0.006905156187713146,
"learning_rate": 1.6153846153846157e-06,
"loss": 0.001,
"step": 18425
},
{
"epoch": 20.319383259911895,
"grad_norm": 0.010452316142618656,
"learning_rate": 1.5897435897435897e-06,
"loss": 0.0013,
"step": 18450
},
{
"epoch": 20.34691629955947,
"grad_norm": 0.1557583063840866,
"learning_rate": 1.5641025641025642e-06,
"loss": 0.001,
"step": 18475
},
{
"epoch": 20.374449339207047,
"grad_norm": 0.006585159804672003,
"learning_rate": 1.5384615384615387e-06,
"loss": 0.0012,
"step": 18500
},
{
"epoch": 20.401982378854626,
"grad_norm": 0.006628986913710833,
"learning_rate": 1.5128205128205128e-06,
"loss": 0.0018,
"step": 18525
},
{
"epoch": 20.429515418502202,
"grad_norm": 0.007554872892796993,
"learning_rate": 1.4871794871794873e-06,
"loss": 0.0008,
"step": 18550
},
{
"epoch": 20.457048458149778,
"grad_norm": 0.005659104790538549,
"learning_rate": 1.4615384615384618e-06,
"loss": 0.0011,
"step": 18575
},
{
"epoch": 20.484581497797357,
"grad_norm": 0.00462260702624917,
"learning_rate": 1.4358974358974359e-06,
"loss": 0.0007,
"step": 18600
},
{
"epoch": 20.512114537444933,
"grad_norm": 0.0087654460221529,
"learning_rate": 1.4102564102564104e-06,
"loss": 0.001,
"step": 18625
},
{
"epoch": 20.53964757709251,
"grad_norm": 0.006052209064364433,
"learning_rate": 1.3846153846153848e-06,
"loss": 0.001,
"step": 18650
},
{
"epoch": 20.56718061674009,
"grad_norm": 0.004353426396846771,
"learning_rate": 1.358974358974359e-06,
"loss": 0.0008,
"step": 18675
},
{
"epoch": 20.594713656387665,
"grad_norm": 0.08834078162908554,
"learning_rate": 1.3333333333333334e-06,
"loss": 0.0011,
"step": 18700
},
{
"epoch": 20.62224669603524,
"grad_norm": 0.006872022990137339,
"learning_rate": 1.307692307692308e-06,
"loss": 0.0008,
"step": 18725
},
{
"epoch": 20.64977973568282,
"grad_norm": 0.004435420036315918,
"learning_rate": 1.282051282051282e-06,
"loss": 0.0006,
"step": 18750
},
{
"epoch": 20.677312775330396,
"grad_norm": 0.00624303100630641,
"learning_rate": 1.2564102564102565e-06,
"loss": 0.0007,
"step": 18775
},
{
"epoch": 20.704845814977972,
"grad_norm": 0.00411389023065567,
"learning_rate": 1.230769230769231e-06,
"loss": 0.0008,
"step": 18800
},
{
"epoch": 20.73237885462555,
"grad_norm": 0.09965424239635468,
"learning_rate": 1.2051282051282053e-06,
"loss": 0.0011,
"step": 18825
},
{
"epoch": 20.759911894273127,
"grad_norm": 0.004442107398062944,
"learning_rate": 1.1794871794871795e-06,
"loss": 0.0007,
"step": 18850
},
{
"epoch": 20.787444933920703,
"grad_norm": 0.004814086947590113,
"learning_rate": 1.153846153846154e-06,
"loss": 0.0007,
"step": 18875
},
{
"epoch": 20.814977973568283,
"grad_norm": 0.006195340771228075,
"learning_rate": 1.1282051282051283e-06,
"loss": 0.0008,
"step": 18900
},
{
"epoch": 20.84251101321586,
"grad_norm": 0.006218872033059597,
"learning_rate": 1.1025641025641026e-06,
"loss": 0.001,
"step": 18925
},
{
"epoch": 20.870044052863435,
"grad_norm": 0.0083702951669693,
"learning_rate": 1.076923076923077e-06,
"loss": 0.0008,
"step": 18950
},
{
"epoch": 20.897577092511014,
"grad_norm": 0.004893292207270861,
"learning_rate": 1.0512820512820514e-06,
"loss": 0.0011,
"step": 18975
},
{
"epoch": 20.92511013215859,
"grad_norm": 0.011980378068983555,
"learning_rate": 1.0256410256410257e-06,
"loss": 0.0011,
"step": 19000
},
{
"epoch": 20.92511013215859,
"eval_cer": 22.251326906102047,
"eval_loss": 0.892787754535675,
"eval_runtime": 1296.9753,
"eval_samples_per_second": 8.158,
"eval_steps_per_second": 2.04,
"eval_wer": 81.62187647336162,
"step": 19000
},
{
"epoch": 20.952643171806166,
"grad_norm": 0.0054518114775419235,
"learning_rate": 1.0000000000000002e-06,
"loss": 0.0008,
"step": 19025
},
{
"epoch": 20.980176211453745,
"grad_norm": 0.0044789970852434635,
"learning_rate": 9.743589743589745e-07,
"loss": 0.0016,
"step": 19050
},
{
"epoch": 21.00770925110132,
"grad_norm": 0.007212365977466106,
"learning_rate": 9.487179487179487e-07,
"loss": 0.0007,
"step": 19075
},
{
"epoch": 21.035242290748897,
"grad_norm": 0.0058553945273160934,
"learning_rate": 9.230769230769232e-07,
"loss": 0.0006,
"step": 19100
},
{
"epoch": 21.062775330396477,
"grad_norm": 0.0037513140123337507,
"learning_rate": 8.974358974358975e-07,
"loss": 0.0007,
"step": 19125
},
{
"epoch": 21.090308370044053,
"grad_norm": 0.0065911454148590565,
"learning_rate": 8.717948717948718e-07,
"loss": 0.0006,
"step": 19150
},
{
"epoch": 21.11784140969163,
"grad_norm": 0.00529465964064002,
"learning_rate": 8.461538461538463e-07,
"loss": 0.0007,
"step": 19175
},
{
"epoch": 21.145374449339208,
"grad_norm": 0.004264230374246836,
"learning_rate": 8.205128205128206e-07,
"loss": 0.0006,
"step": 19200
},
{
"epoch": 21.172907488986784,
"grad_norm": 0.0038558640517294407,
"learning_rate": 7.948717948717949e-07,
"loss": 0.0007,
"step": 19225
},
{
"epoch": 21.20044052863436,
"grad_norm": 0.0034747957251966,
"learning_rate": 7.692307692307694e-07,
"loss": 0.0006,
"step": 19250
},
{
"epoch": 21.22797356828194,
"grad_norm": 0.004676154814660549,
"learning_rate": 7.435897435897436e-07,
"loss": 0.0006,
"step": 19275
},
{
"epoch": 21.255506607929515,
"grad_norm": 0.005274807568639517,
"learning_rate": 7.179487179487179e-07,
"loss": 0.0006,
"step": 19300
},
{
"epoch": 21.28303964757709,
"grad_norm": 0.005552313756197691,
"learning_rate": 6.923076923076924e-07,
"loss": 0.0006,
"step": 19325
},
{
"epoch": 21.31057268722467,
"grad_norm": 0.0034383879974484444,
"learning_rate": 6.666666666666667e-07,
"loss": 0.0006,
"step": 19350
},
{
"epoch": 21.338105726872246,
"grad_norm": 0.00359016889706254,
"learning_rate": 6.41025641025641e-07,
"loss": 0.0006,
"step": 19375
},
{
"epoch": 21.365638766519822,
"grad_norm": 0.007120490074157715,
"learning_rate": 6.153846153846155e-07,
"loss": 0.0006,
"step": 19400
},
{
"epoch": 21.393171806167402,
"grad_norm": 0.005752895027399063,
"learning_rate": 5.897435897435898e-07,
"loss": 0.0007,
"step": 19425
},
{
"epoch": 21.420704845814978,
"grad_norm": 0.003924291115254164,
"learning_rate": 5.641025641025642e-07,
"loss": 0.0006,
"step": 19450
},
{
"epoch": 21.448237885462554,
"grad_norm": 0.0068072304129600525,
"learning_rate": 5.384615384615386e-07,
"loss": 0.0017,
"step": 19475
},
{
"epoch": 21.475770925110133,
"grad_norm": 0.003380249021574855,
"learning_rate": 5.128205128205128e-07,
"loss": 0.0008,
"step": 19500
},
{
"epoch": 21.50330396475771,
"grad_norm": 0.003210252383723855,
"learning_rate": 4.871794871794872e-07,
"loss": 0.0006,
"step": 19525
},
{
"epoch": 21.530837004405285,
"grad_norm": 0.12751127779483795,
"learning_rate": 4.615384615384616e-07,
"loss": 0.0006,
"step": 19550
},
{
"epoch": 21.558370044052865,
"grad_norm": 0.0041765086352825165,
"learning_rate": 4.358974358974359e-07,
"loss": 0.0006,
"step": 19575
},
{
"epoch": 21.58590308370044,
"grad_norm": 0.17753562331199646,
"learning_rate": 4.102564102564103e-07,
"loss": 0.0007,
"step": 19600
},
{
"epoch": 21.613436123348016,
"grad_norm": 0.003080737078562379,
"learning_rate": 3.846153846153847e-07,
"loss": 0.0007,
"step": 19625
},
{
"epoch": 21.640969162995596,
"grad_norm": 0.004255061503499746,
"learning_rate": 3.5897435897435896e-07,
"loss": 0.0008,
"step": 19650
},
{
"epoch": 21.66850220264317,
"grad_norm": 0.003112249309197068,
"learning_rate": 3.3333333333333335e-07,
"loss": 0.0006,
"step": 19675
},
{
"epoch": 21.696035242290748,
"grad_norm": 0.0038673074450343847,
"learning_rate": 3.0769230769230774e-07,
"loss": 0.0013,
"step": 19700
},
{
"epoch": 21.723568281938327,
"grad_norm": 0.004677412565797567,
"learning_rate": 2.820512820512821e-07,
"loss": 0.0009,
"step": 19725
},
{
"epoch": 21.751101321585903,
"grad_norm": 0.0037716683000326157,
"learning_rate": 2.564102564102564e-07,
"loss": 0.0006,
"step": 19750
},
{
"epoch": 21.77863436123348,
"grad_norm": 0.004941494669765234,
"learning_rate": 2.307692307692308e-07,
"loss": 0.0006,
"step": 19775
},
{
"epoch": 21.80616740088106,
"grad_norm": 0.007174940314143896,
"learning_rate": 2.0512820512820514e-07,
"loss": 0.0006,
"step": 19800
},
{
"epoch": 21.833700440528634,
"grad_norm": 0.004461018834263086,
"learning_rate": 1.7948717948717948e-07,
"loss": 0.0006,
"step": 19825
},
{
"epoch": 21.86123348017621,
"grad_norm": 0.0061052520759403706,
"learning_rate": 1.5384615384615387e-07,
"loss": 0.0006,
"step": 19850
},
{
"epoch": 21.88876651982379,
"grad_norm": 0.00467773899435997,
"learning_rate": 1.282051282051282e-07,
"loss": 0.0006,
"step": 19875
},
{
"epoch": 21.916299559471366,
"grad_norm": 0.003872169181704521,
"learning_rate": 1.0256410256410257e-07,
"loss": 0.0006,
"step": 19900
},
{
"epoch": 21.94383259911894,
"grad_norm": 0.0046131303533911705,
"learning_rate": 7.692307692307694e-08,
"loss": 0.0006,
"step": 19925
},
{
"epoch": 21.97136563876652,
"grad_norm": 0.007839877158403397,
"learning_rate": 5.1282051282051286e-08,
"loss": 0.0007,
"step": 19950
},
{
"epoch": 21.998898678414097,
"grad_norm": 0.004426113795489073,
"learning_rate": 2.5641025641025643e-08,
"loss": 0.0007,
"step": 19975
},
{
"epoch": 22.026431718061673,
"grad_norm": 0.00471852533519268,
"learning_rate": 0.0,
"loss": 0.0006,
"step": 20000
},
{
"epoch": 22.026431718061673,
"eval_cer": 22.213497185698074,
"eval_loss": 0.8962610960006714,
"eval_runtime": 1296.5309,
"eval_samples_per_second": 8.161,
"eval_steps_per_second": 2.041,
"eval_wer": 81.71617161716172,
"step": 20000
},
{
"epoch": 22.026431718061673,
"step": 20000,
"total_flos": 3.4362863729801953e+20,
"train_loss": 0.2747690211042762,
"train_runtime": 126669.5289,
"train_samples_per_second": 5.053,
"train_steps_per_second": 0.158
}
],
"logging_steps": 25,
"max_steps": 20000,
"num_input_tokens_seen": 0,
"num_train_epochs": 23,
"save_steps": 1000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 3.4362863729801953e+20,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}