xls-r-uzbek-cv8 / trainer_state.json
lucio's picture
End of training
cdb3b32
raw
history blame
27.9 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 99.99837925445705,
"global_step": 15400,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.65,
"learning_rate": 5.940000000000001e-06,
"loss": 11.9235,
"step": 100
},
{
"epoch": 1.3,
"learning_rate": 1.1940000000000001e-05,
"loss": 5.1849,
"step": 200
},
{
"epoch": 1.95,
"learning_rate": 1.794e-05,
"loss": 3.7405,
"step": 300
},
{
"epoch": 2.6,
"learning_rate": 2.394e-05,
"loss": 3.3052,
"step": 400
},
{
"epoch": 3.25,
"learning_rate": 2.994e-05,
"loss": 3.1401,
"step": 500
},
{
"epoch": 3.25,
"eval_cer": 1.0,
"eval_loss": 3.114637851715088,
"eval_runtime": 219.3333,
"eval_samples_per_second": 24.734,
"eval_steps_per_second": 3.096,
"eval_wer": 1.0,
"step": 500
},
{
"epoch": 3.89,
"learning_rate": 2.98006711409396e-05,
"loss": 3.0799,
"step": 600
},
{
"epoch": 4.54,
"learning_rate": 2.9599328859060405e-05,
"loss": 3.0534,
"step": 700
},
{
"epoch": 5.19,
"learning_rate": 2.9397986577181207e-05,
"loss": 2.994,
"step": 800
},
{
"epoch": 5.84,
"learning_rate": 2.9196644295302013e-05,
"loss": 2.9428,
"step": 900
},
{
"epoch": 6.49,
"learning_rate": 2.8995302013422818e-05,
"loss": 2.7484,
"step": 1000
},
{
"epoch": 6.49,
"eval_cer": 0.706927100586246,
"eval_loss": 2.284235954284668,
"eval_runtime": 222.6226,
"eval_samples_per_second": 24.369,
"eval_steps_per_second": 3.05,
"eval_wer": 1.0065119583234667,
"step": 1000
},
{
"epoch": 7.14,
"learning_rate": 2.8793959731543624e-05,
"loss": 2.0772,
"step": 1100
},
{
"epoch": 7.79,
"learning_rate": 2.859261744966443e-05,
"loss": 1.4967,
"step": 1200
},
{
"epoch": 8.44,
"learning_rate": 2.8391275167785235e-05,
"loss": 1.2731,
"step": 1300
},
{
"epoch": 9.09,
"learning_rate": 2.818993288590604e-05,
"loss": 1.1742,
"step": 1400
},
{
"epoch": 9.74,
"learning_rate": 2.7988590604026846e-05,
"loss": 1.0899,
"step": 1500
},
{
"epoch": 9.74,
"eval_cer": 0.13506382454221202,
"eval_loss": 0.5414122343063354,
"eval_runtime": 217.7702,
"eval_samples_per_second": 24.912,
"eval_steps_per_second": 3.118,
"eval_wer": 0.6124792801326071,
"step": 1500
},
{
"epoch": 10.39,
"learning_rate": 2.778724832214765e-05,
"loss": 1.0544,
"step": 1600
},
{
"epoch": 11.04,
"learning_rate": 2.7585906040268457e-05,
"loss": 1.0284,
"step": 1700
},
{
"epoch": 11.69,
"learning_rate": 2.7384563758389263e-05,
"loss": 0.9865,
"step": 1800
},
{
"epoch": 12.34,
"learning_rate": 2.7183221476510065e-05,
"loss": 0.9705,
"step": 1900
},
{
"epoch": 12.99,
"learning_rate": 2.698187919463087e-05,
"loss": 0.9465,
"step": 2000
},
{
"epoch": 12.99,
"eval_cer": 0.12229891609980818,
"eval_loss": 0.4565887749195099,
"eval_runtime": 219.4838,
"eval_samples_per_second": 24.717,
"eval_steps_per_second": 3.094,
"eval_wer": 0.5634619938432394,
"step": 2000
},
{
"epoch": 13.64,
"learning_rate": 2.6780536912751676e-05,
"loss": 0.9395,
"step": 2100
},
{
"epoch": 14.29,
"learning_rate": 2.6579194630872482e-05,
"loss": 0.9216,
"step": 2200
},
{
"epoch": 14.93,
"learning_rate": 2.6377852348993287e-05,
"loss": 0.9011,
"step": 2300
},
{
"epoch": 15.58,
"learning_rate": 2.6176510067114093e-05,
"loss": 0.9011,
"step": 2400
},
{
"epoch": 16.23,
"learning_rate": 2.59751677852349e-05,
"loss": 0.8771,
"step": 2500
},
{
"epoch": 16.23,
"eval_cer": 0.11608208983968754,
"eval_loss": 0.42124298214912415,
"eval_runtime": 218.6532,
"eval_samples_per_second": 24.811,
"eval_steps_per_second": 3.105,
"eval_wer": 0.5365557660430973,
"step": 2500
},
{
"epoch": 16.88,
"learning_rate": 2.5773825503355704e-05,
"loss": 0.8719,
"step": 2600
},
{
"epoch": 17.53,
"learning_rate": 2.557248322147651e-05,
"loss": 0.8616,
"step": 2700
},
{
"epoch": 18.18,
"learning_rate": 2.5371140939597315e-05,
"loss": 0.8573,
"step": 2800
},
{
"epoch": 18.83,
"learning_rate": 2.5169798657718124e-05,
"loss": 0.8467,
"step": 2900
},
{
"epoch": 19.48,
"learning_rate": 2.496845637583893e-05,
"loss": 0.8346,
"step": 3000
},
{
"epoch": 19.48,
"eval_cer": 0.11022733400611667,
"eval_loss": 0.3994467258453369,
"eval_runtime": 213.3023,
"eval_samples_per_second": 25.433,
"eval_steps_per_second": 3.183,
"eval_wer": 0.5143855079327492,
"step": 3000
},
{
"epoch": 20.13,
"learning_rate": 2.4767114093959732e-05,
"loss": 0.8386,
"step": 3100
},
{
"epoch": 20.78,
"learning_rate": 2.4565771812080538e-05,
"loss": 0.8283,
"step": 3200
},
{
"epoch": 21.43,
"learning_rate": 2.4364429530201343e-05,
"loss": 0.8198,
"step": 3300
},
{
"epoch": 22.08,
"learning_rate": 2.416308724832215e-05,
"loss": 0.8164,
"step": 3400
},
{
"epoch": 22.73,
"learning_rate": 2.3961744966442955e-05,
"loss": 0.8127,
"step": 3500
},
{
"epoch": 22.73,
"eval_cer": 0.10510827446479058,
"eval_loss": 0.3818908929824829,
"eval_runtime": 218.4137,
"eval_samples_per_second": 24.838,
"eval_steps_per_second": 3.109,
"eval_wer": 0.49437603599336966,
"step": 3500
},
{
"epoch": 23.38,
"learning_rate": 2.376040268456376e-05,
"loss": 0.8081,
"step": 3600
},
{
"epoch": 24.03,
"learning_rate": 2.3559060402684566e-05,
"loss": 0.7988,
"step": 3700
},
{
"epoch": 24.67,
"learning_rate": 2.335771812080537e-05,
"loss": 0.7903,
"step": 3800
},
{
"epoch": 25.32,
"learning_rate": 2.3156375838926177e-05,
"loss": 0.788,
"step": 3900
},
{
"epoch": 25.97,
"learning_rate": 2.2955033557046982e-05,
"loss": 0.7833,
"step": 4000
},
{
"epoch": 25.97,
"eval_cer": 0.10109853708140422,
"eval_loss": 0.3704679310321808,
"eval_runtime": 216.2288,
"eval_samples_per_second": 25.089,
"eval_steps_per_second": 3.14,
"eval_wer": 0.4797537295761307,
"step": 4000
},
{
"epoch": 26.62,
"learning_rate": 2.2753691275167788e-05,
"loss": 0.7724,
"step": 4100
},
{
"epoch": 27.27,
"learning_rate": 2.255234899328859e-05,
"loss": 0.7695,
"step": 4200
},
{
"epoch": 27.92,
"learning_rate": 2.2351006711409396e-05,
"loss": 0.7653,
"step": 4300
},
{
"epoch": 28.57,
"learning_rate": 2.2151677852348994e-05,
"loss": 0.7728,
"step": 4400
},
{
"epoch": 29.22,
"learning_rate": 2.19503355704698e-05,
"loss": 0.7603,
"step": 4500
},
{
"epoch": 29.22,
"eval_cer": 0.0992381113790261,
"eval_loss": 0.36611661314964294,
"eval_runtime": 217.7866,
"eval_samples_per_second": 24.91,
"eval_steps_per_second": 3.118,
"eval_wer": 0.4704001894387876,
"step": 4500
},
{
"epoch": 29.87,
"learning_rate": 2.1748993288590605e-05,
"loss": 0.7563,
"step": 4600
},
{
"epoch": 30.52,
"learning_rate": 2.154765100671141e-05,
"loss": 0.7618,
"step": 4700
},
{
"epoch": 31.17,
"learning_rate": 2.1346308724832217e-05,
"loss": 0.7467,
"step": 4800
},
{
"epoch": 31.82,
"learning_rate": 2.114697986577181e-05,
"loss": 0.7514,
"step": 4900
},
{
"epoch": 32.47,
"learning_rate": 2.0945637583892617e-05,
"loss": 0.7424,
"step": 5000
},
{
"epoch": 32.47,
"eval_cer": 0.09569444337449638,
"eval_loss": 0.3528956174850464,
"eval_runtime": 218.9072,
"eval_samples_per_second": 24.782,
"eval_steps_per_second": 3.102,
"eval_wer": 0.4577314705185887,
"step": 5000
},
{
"epoch": 33.12,
"learning_rate": 2.0744295302013423e-05,
"loss": 0.748,
"step": 5100
},
{
"epoch": 33.76,
"learning_rate": 2.054295302013423e-05,
"loss": 0.7357,
"step": 5200
},
{
"epoch": 34.41,
"learning_rate": 2.0341610738255034e-05,
"loss": 0.7357,
"step": 5300
},
{
"epoch": 35.06,
"learning_rate": 2.014026845637584e-05,
"loss": 0.735,
"step": 5400
},
{
"epoch": 35.71,
"learning_rate": 1.9938926174496645e-05,
"loss": 0.7251,
"step": 5500
},
{
"epoch": 35.71,
"eval_cer": 0.09283254627953377,
"eval_loss": 0.34103500843048096,
"eval_runtime": 218.8353,
"eval_samples_per_second": 24.79,
"eval_steps_per_second": 3.103,
"eval_wer": 0.4472827373904807,
"step": 5500
},
{
"epoch": 36.36,
"learning_rate": 1.973758389261745e-05,
"loss": 0.7306,
"step": 5600
},
{
"epoch": 37.01,
"learning_rate": 1.9536241610738256e-05,
"loss": 0.7185,
"step": 5700
},
{
"epoch": 37.66,
"learning_rate": 1.9334899328859062e-05,
"loss": 0.7135,
"step": 5800
},
{
"epoch": 38.31,
"learning_rate": 1.9133557046979864e-05,
"loss": 0.726,
"step": 5900
},
{
"epoch": 38.96,
"learning_rate": 1.893221476510067e-05,
"loss": 0.7106,
"step": 6000
},
{
"epoch": 38.96,
"eval_cer": 0.09190425933486893,
"eval_loss": 0.3401394486427307,
"eval_runtime": 218.6772,
"eval_samples_per_second": 24.808,
"eval_steps_per_second": 3.105,
"eval_wer": 0.4427539663746152,
"step": 6000
},
{
"epoch": 39.61,
"learning_rate": 1.8730872483221475e-05,
"loss": 0.7157,
"step": 6100
},
{
"epoch": 40.26,
"learning_rate": 1.852953020134228e-05,
"loss": 0.7055,
"step": 6200
},
{
"epoch": 40.91,
"learning_rate": 1.8328187919463086e-05,
"loss": 0.7108,
"step": 6300
},
{
"epoch": 41.56,
"learning_rate": 1.8126845637583892e-05,
"loss": 0.705,
"step": 6400
},
{
"epoch": 42.21,
"learning_rate": 1.7925503355704698e-05,
"loss": 0.7027,
"step": 6500
},
{
"epoch": 42.21,
"eval_cer": 0.09045597762866982,
"eval_loss": 0.3354834318161011,
"eval_runtime": 219.2927,
"eval_samples_per_second": 24.739,
"eval_steps_per_second": 3.096,
"eval_wer": 0.4352652143026285,
"step": 6500
},
{
"epoch": 42.86,
"learning_rate": 1.7724161073825503e-05,
"loss": 0.7103,
"step": 6600
},
{
"epoch": 43.51,
"learning_rate": 1.752281879194631e-05,
"loss": 0.7011,
"step": 6700
},
{
"epoch": 44.16,
"learning_rate": 1.7321476510067114e-05,
"loss": 0.7004,
"step": 6800
},
{
"epoch": 44.8,
"learning_rate": 1.712013422818792e-05,
"loss": 0.702,
"step": 6900
},
{
"epoch": 45.45,
"learning_rate": 1.6918791946308722e-05,
"loss": 0.6927,
"step": 7000
},
{
"epoch": 45.45,
"eval_cer": 0.08853392291751727,
"eval_loss": 0.33077025413513184,
"eval_runtime": 215.9851,
"eval_samples_per_second": 25.117,
"eval_steps_per_second": 3.144,
"eval_wer": 0.4296412502959981,
"step": 7000
},
{
"epoch": 46.1,
"learning_rate": 1.6717449664429528e-05,
"loss": 0.691,
"step": 7100
},
{
"epoch": 46.75,
"learning_rate": 1.6516107382550333e-05,
"loss": 0.6833,
"step": 7200
},
{
"epoch": 47.4,
"learning_rate": 1.631476510067114e-05,
"loss": 0.692,
"step": 7300
},
{
"epoch": 48.05,
"learning_rate": 1.6113422818791948e-05,
"loss": 0.6871,
"step": 7400
},
{
"epoch": 48.7,
"learning_rate": 1.5914093959731546e-05,
"loss": 0.6828,
"step": 7500
},
{
"epoch": 48.7,
"eval_cer": 0.08625364959286336,
"eval_loss": 0.324627548456192,
"eval_runtime": 221.033,
"eval_samples_per_second": 24.544,
"eval_steps_per_second": 3.072,
"eval_wer": 0.42043570921146106,
"step": 7500
},
{
"epoch": 49.35,
"learning_rate": 1.5712751677852352e-05,
"loss": 0.6789,
"step": 7600
},
{
"epoch": 50.0,
"learning_rate": 1.5511409395973158e-05,
"loss": 0.6811,
"step": 7700
},
{
"epoch": 50.65,
"learning_rate": 1.531006711409396e-05,
"loss": 0.683,
"step": 7800
},
{
"epoch": 51.3,
"learning_rate": 1.5108724832214764e-05,
"loss": 0.6765,
"step": 7900
},
{
"epoch": 51.95,
"learning_rate": 1.4907382550335571e-05,
"loss": 0.6706,
"step": 8000
},
{
"epoch": 51.95,
"eval_cer": 0.08681216248488163,
"eval_loss": 0.32503771781921387,
"eval_runtime": 215.1259,
"eval_samples_per_second": 25.218,
"eval_steps_per_second": 3.156,
"eval_wer": 0.42327729102533745,
"step": 8000
},
{
"epoch": 52.6,
"learning_rate": 1.4706040268456375e-05,
"loss": 0.6777,
"step": 8100
},
{
"epoch": 53.25,
"learning_rate": 1.450469798657718e-05,
"loss": 0.6754,
"step": 8200
},
{
"epoch": 53.89,
"learning_rate": 1.4303355704697986e-05,
"loss": 0.6675,
"step": 8300
},
{
"epoch": 54.54,
"learning_rate": 1.4102013422818792e-05,
"loss": 0.6627,
"step": 8400
},
{
"epoch": 55.19,
"learning_rate": 1.3900671140939599e-05,
"loss": 0.6629,
"step": 8500
},
{
"epoch": 55.19,
"eval_cer": 0.0849055150259227,
"eval_loss": 0.3263927102088928,
"eval_runtime": 213.0987,
"eval_samples_per_second": 25.458,
"eval_steps_per_second": 3.186,
"eval_wer": 0.4158773383850343,
"step": 8500
},
{
"epoch": 55.84,
"learning_rate": 1.3699328859060405e-05,
"loss": 0.6632,
"step": 8600
},
{
"epoch": 56.49,
"learning_rate": 1.3497986577181208e-05,
"loss": 0.6558,
"step": 8700
},
{
"epoch": 57.14,
"learning_rate": 1.3296644295302014e-05,
"loss": 0.6691,
"step": 8800
},
{
"epoch": 57.79,
"learning_rate": 1.309530201342282e-05,
"loss": 0.6633,
"step": 8900
},
{
"epoch": 58.44,
"learning_rate": 1.2893959731543625e-05,
"loss": 0.6556,
"step": 9000
},
{
"epoch": 58.44,
"eval_cer": 0.08353812139374003,
"eval_loss": 0.3212815821170807,
"eval_runtime": 220.228,
"eval_samples_per_second": 24.634,
"eval_steps_per_second": 3.083,
"eval_wer": 0.40995737627279183,
"step": 9000
},
{
"epoch": 59.09,
"learning_rate": 1.269261744966443e-05,
"loss": 0.6584,
"step": 9100
},
{
"epoch": 59.74,
"learning_rate": 1.2491275167785236e-05,
"loss": 0.6537,
"step": 9200
},
{
"epoch": 60.39,
"learning_rate": 1.228993288590604e-05,
"loss": 0.6633,
"step": 9300
},
{
"epoch": 61.04,
"learning_rate": 1.2088590604026846e-05,
"loss": 0.6474,
"step": 9400
},
{
"epoch": 61.69,
"learning_rate": 1.1887248322147651e-05,
"loss": 0.6484,
"step": 9500
},
{
"epoch": 61.69,
"eval_cer": 0.0836613794112889,
"eval_loss": 0.31816166639328003,
"eval_runtime": 216.5133,
"eval_samples_per_second": 25.056,
"eval_steps_per_second": 3.136,
"eval_wer": 0.41241416054937247,
"step": 9500
},
{
"epoch": 62.34,
"learning_rate": 1.1687919463087248e-05,
"loss": 0.6547,
"step": 9600
},
{
"epoch": 62.99,
"learning_rate": 1.1486577181208054e-05,
"loss": 0.6481,
"step": 9700
},
{
"epoch": 63.64,
"learning_rate": 1.128523489932886e-05,
"loss": 0.648,
"step": 9800
},
{
"epoch": 64.29,
"learning_rate": 1.1083892617449665e-05,
"loss": 0.6471,
"step": 9900
},
{
"epoch": 64.93,
"learning_rate": 1.088255033557047e-05,
"loss": 0.6407,
"step": 10000
},
{
"epoch": 64.93,
"eval_cer": 0.08248272461847792,
"eval_loss": 0.3171332776546478,
"eval_runtime": 217.1179,
"eval_samples_per_second": 24.986,
"eval_steps_per_second": 3.127,
"eval_wer": 0.4050142079090694,
"step": 10000
},
{
"epoch": 65.58,
"learning_rate": 1.0681208053691274e-05,
"loss": 0.6446,
"step": 10100
},
{
"epoch": 66.23,
"learning_rate": 1.047986577181208e-05,
"loss": 0.6383,
"step": 10200
},
{
"epoch": 66.88,
"learning_rate": 1.0278523489932886e-05,
"loss": 0.6413,
"step": 10300
},
{
"epoch": 67.53,
"learning_rate": 1.0077181208053691e-05,
"loss": 0.6494,
"step": 10400
},
{
"epoch": 68.18,
"learning_rate": 9.875838926174497e-06,
"loss": 0.6375,
"step": 10500
},
{
"epoch": 68.18,
"eval_cer": 0.0822362085833802,
"eval_loss": 0.31498104333877563,
"eval_runtime": 218.6829,
"eval_samples_per_second": 24.808,
"eval_steps_per_second": 3.105,
"eval_wer": 0.4038598152971821,
"step": 10500
},
{
"epoch": 68.83,
"learning_rate": 9.6744966442953e-06,
"loss": 0.6359,
"step": 10600
},
{
"epoch": 69.48,
"learning_rate": 9.473154362416108e-06,
"loss": 0.638,
"step": 10700
},
{
"epoch": 70.13,
"learning_rate": 9.271812080536914e-06,
"loss": 0.6405,
"step": 10800
},
{
"epoch": 70.78,
"learning_rate": 9.070469798657719e-06,
"loss": 0.6388,
"step": 10900
},
{
"epoch": 71.43,
"learning_rate": 8.869127516778525e-06,
"loss": 0.6363,
"step": 11000
},
{
"epoch": 71.43,
"eval_cer": 0.08095355483826237,
"eval_loss": 0.3129253089427948,
"eval_runtime": 218.9143,
"eval_samples_per_second": 24.781,
"eval_steps_per_second": 3.102,
"eval_wer": 0.3991238456073881,
"step": 11000
},
{
"epoch": 72.08,
"learning_rate": 8.66778523489933e-06,
"loss": 0.6369,
"step": 11100
},
{
"epoch": 72.73,
"learning_rate": 8.466442953020134e-06,
"loss": 0.635,
"step": 11200
},
{
"epoch": 73.38,
"learning_rate": 8.26510067114094e-06,
"loss": 0.6337,
"step": 11300
},
{
"epoch": 74.03,
"learning_rate": 8.063758389261745e-06,
"loss": 0.6308,
"step": 11400
},
{
"epoch": 74.67,
"learning_rate": 7.862416107382551e-06,
"loss": 0.6307,
"step": 11500
},
{
"epoch": 74.67,
"eval_cer": 0.08074170512060026,
"eval_loss": 0.3114279508590698,
"eval_runtime": 219.6542,
"eval_samples_per_second": 24.698,
"eval_steps_per_second": 3.091,
"eval_wer": 0.3986206488278475,
"step": 11500
},
{
"epoch": 75.32,
"learning_rate": 7.661073825503357e-06,
"loss": 0.6335,
"step": 11600
},
{
"epoch": 75.97,
"learning_rate": 7.459731543624161e-06,
"loss": 0.628,
"step": 11700
},
{
"epoch": 76.62,
"learning_rate": 7.260402684563759e-06,
"loss": 0.6324,
"step": 11800
},
{
"epoch": 77.27,
"learning_rate": 7.059060402684564e-06,
"loss": 0.6317,
"step": 11900
},
{
"epoch": 77.92,
"learning_rate": 6.857718120805369e-06,
"loss": 0.6232,
"step": 12000
},
{
"epoch": 77.92,
"eval_cer": 0.07899683380967422,
"eval_loss": 0.31030353903770447,
"eval_runtime": 220.4971,
"eval_samples_per_second": 24.603,
"eval_steps_per_second": 3.079,
"eval_wer": 0.3895335069855553,
"step": 12000
},
{
"epoch": 78.57,
"learning_rate": 6.656375838926175e-06,
"loss": 0.6295,
"step": 12100
},
{
"epoch": 79.22,
"learning_rate": 6.4550335570469795e-06,
"loss": 0.6234,
"step": 12200
},
{
"epoch": 79.87,
"learning_rate": 6.255704697986578e-06,
"loss": 0.6172,
"step": 12300
},
{
"epoch": 80.52,
"learning_rate": 6.054362416107383e-06,
"loss": 0.6203,
"step": 12400
},
{
"epoch": 81.17,
"learning_rate": 5.853020134228188e-06,
"loss": 0.6216,
"step": 12500
},
{
"epoch": 81.17,
"eval_cer": 0.0789506120530934,
"eval_loss": 0.30863967537879944,
"eval_runtime": 218.4142,
"eval_samples_per_second": 24.838,
"eval_steps_per_second": 3.109,
"eval_wer": 0.3891191096376983,
"step": 12500
},
{
"epoch": 81.82,
"learning_rate": 5.651677852348994e-06,
"loss": 0.6203,
"step": 12600
},
{
"epoch": 82.47,
"learning_rate": 5.4503355704697986e-06,
"loss": 0.6209,
"step": 12700
},
{
"epoch": 83.12,
"learning_rate": 5.248993288590604e-06,
"loss": 0.6257,
"step": 12800
},
{
"epoch": 83.76,
"learning_rate": 5.04765100671141e-06,
"loss": 0.6245,
"step": 12900
},
{
"epoch": 84.41,
"learning_rate": 4.8463087248322145e-06,
"loss": 0.6174,
"step": 13000
},
{
"epoch": 84.41,
"eval_cer": 0.07851150536557558,
"eval_loss": 0.3082079291343689,
"eval_runtime": 215.0269,
"eval_samples_per_second": 25.229,
"eval_steps_per_second": 3.158,
"eval_wer": 0.3880535164574947,
"step": 13000
},
{
"epoch": 85.06,
"learning_rate": 4.64496644295302e-06,
"loss": 0.6222,
"step": 13100
},
{
"epoch": 85.71,
"learning_rate": 4.443624161073826e-06,
"loss": 0.6113,
"step": 13200
},
{
"epoch": 86.36,
"learning_rate": 4.2422818791946304e-06,
"loss": 0.6238,
"step": 13300
},
{
"epoch": 87.01,
"learning_rate": 4.040939597315437e-06,
"loss": 0.618,
"step": 13400
},
{
"epoch": 87.66,
"learning_rate": 3.8395973154362425e-06,
"loss": 0.6196,
"step": 13500
},
{
"epoch": 87.66,
"eval_cer": 0.07821876757389704,
"eval_loss": 0.30590009689331055,
"eval_runtime": 213.2382,
"eval_samples_per_second": 25.441,
"eval_steps_per_second": 3.184,
"eval_wer": 0.3874911200568316,
"step": 13500
},
{
"epoch": 88.31,
"learning_rate": 3.6382550335570468e-06,
"loss": 0.6174,
"step": 13600
},
{
"epoch": 88.96,
"learning_rate": 3.4369127516778524e-06,
"loss": 0.6128,
"step": 13700
},
{
"epoch": 89.61,
"learning_rate": 3.235570469798658e-06,
"loss": 0.6246,
"step": 13800
},
{
"epoch": 90.26,
"learning_rate": 3.034228187919463e-06,
"loss": 0.6097,
"step": 13900
},
{
"epoch": 90.91,
"learning_rate": 2.8328859060402687e-06,
"loss": 0.6174,
"step": 14000
},
{
"epoch": 90.91,
"eval_cer": 0.07799151060404132,
"eval_loss": 0.30842480063438416,
"eval_runtime": 212.6251,
"eval_samples_per_second": 25.514,
"eval_steps_per_second": 3.193,
"eval_wer": 0.3862479280132607,
"step": 14000
},
{
"epoch": 91.56,
"learning_rate": 2.631543624161074e-06,
"loss": 0.6194,
"step": 14100
},
{
"epoch": 92.21,
"learning_rate": 2.430201342281879e-06,
"loss": 0.6167,
"step": 14200
},
{
"epoch": 92.86,
"learning_rate": 2.2288590604026842e-06,
"loss": 0.614,
"step": 14300
},
{
"epoch": 93.51,
"learning_rate": 2.0275167785234902e-06,
"loss": 0.615,
"step": 14400
},
{
"epoch": 94.16,
"learning_rate": 1.8261744966442954e-06,
"loss": 0.6169,
"step": 14500
},
{
"epoch": 94.16,
"eval_cer": 0.07787595621258926,
"eval_loss": 0.30701127648353577,
"eval_runtime": 215.1709,
"eval_samples_per_second": 25.213,
"eval_steps_per_second": 3.156,
"eval_wer": 0.3859519299076486,
"step": 14500
},
{
"epoch": 94.8,
"learning_rate": 1.6248322147651008e-06,
"loss": 0.6123,
"step": 14600
},
{
"epoch": 95.45,
"learning_rate": 1.423489932885906e-06,
"loss": 0.6133,
"step": 14700
},
{
"epoch": 96.1,
"learning_rate": 1.2221476510067115e-06,
"loss": 0.6068,
"step": 14800
},
{
"epoch": 96.75,
"learning_rate": 1.020805369127517e-06,
"loss": 0.6135,
"step": 14900
},
{
"epoch": 97.4,
"learning_rate": 8.194630872483221e-07,
"loss": 0.6166,
"step": 15000
},
{
"epoch": 97.4,
"eval_cer": 0.07776810544723402,
"eval_loss": 0.30662447214126587,
"eval_runtime": 214.9469,
"eval_samples_per_second": 25.239,
"eval_steps_per_second": 3.159,
"eval_wer": 0.3855079327492304,
"step": 15000
},
{
"epoch": 98.05,
"learning_rate": 6.181208053691276e-07,
"loss": 0.6189,
"step": 15100
},
{
"epoch": 98.7,
"learning_rate": 4.167785234899329e-07,
"loss": 0.6085,
"step": 15200
},
{
"epoch": 99.35,
"learning_rate": 2.1543624161073826e-07,
"loss": 0.6135,
"step": 15300
},
{
"epoch": 100.0,
"learning_rate": 1.4093959731543625e-08,
"loss": 0.6047,
"step": 15400
},
{
"epoch": 100.0,
"step": 15400,
"total_flos": 2.652008062738907e+20,
"train_loss": 0.9527478711016767,
"train_runtime": 101833.8069,
"train_samples_per_second": 19.371,
"train_steps_per_second": 0.151
}
],
"max_steps": 15400,
"num_train_epochs": 100,
"total_flos": 2.652008062738907e+20,
"trial_name": null,
"trial_params": null
}