| { | |
| "best_global_step": 4000, | |
| "best_metric": 81.67647178155556, | |
| "best_model_checkpoint": "./working_area/output_model/checkpoint-4000", | |
| "epoch": 0.6469351447517386, | |
| "eval_steps": 1000, | |
| "global_step": 4000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004043344654698367, | |
| "grad_norm": 10.67656421661377, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 0.2728, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.008086689309396733, | |
| "grad_norm": 0.34478363394737244, | |
| "learning_rate": 9.800000000000001e-07, | |
| "loss": 0.0358, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.012130033964095099, | |
| "grad_norm": 0.26953190565109253, | |
| "learning_rate": 1.48e-06, | |
| "loss": 0.0261, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.016173378618793467, | |
| "grad_norm": 0.2419331669807434, | |
| "learning_rate": 1.98e-06, | |
| "loss": 0.0251, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.020216723273491832, | |
| "grad_norm": 0.29993513226509094, | |
| "learning_rate": 2.4800000000000004e-06, | |
| "loss": 0.0247, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.024260067928190198, | |
| "grad_norm": 0.2362569123506546, | |
| "learning_rate": 2.9800000000000003e-06, | |
| "loss": 0.0227, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.028303412582888564, | |
| "grad_norm": 0.2520463466644287, | |
| "learning_rate": 3.48e-06, | |
| "loss": 0.0246, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.03234675723758693, | |
| "grad_norm": 0.2729567885398865, | |
| "learning_rate": 3.980000000000001e-06, | |
| "loss": 0.0236, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.036390101892285295, | |
| "grad_norm": 0.21665456891059875, | |
| "learning_rate": 4.48e-06, | |
| "loss": 0.0242, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.040433446546983665, | |
| "grad_norm": 0.27291786670684814, | |
| "learning_rate": 4.980000000000001e-06, | |
| "loss": 0.0242, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.044476791201682034, | |
| "grad_norm": 0.20698551833629608, | |
| "learning_rate": 5.480000000000001e-06, | |
| "loss": 0.0229, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.048520135856380396, | |
| "grad_norm": 0.2506471276283264, | |
| "learning_rate": 5.98e-06, | |
| "loss": 0.0225, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.052563480511078765, | |
| "grad_norm": 0.24540211260318756, | |
| "learning_rate": 6.480000000000001e-06, | |
| "loss": 0.0231, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.05660682516577713, | |
| "grad_norm": 0.2298947423696518, | |
| "learning_rate": 6.98e-06, | |
| "loss": 0.0223, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.0606501698204755, | |
| "grad_norm": 0.19595475494861603, | |
| "learning_rate": 7.48e-06, | |
| "loss": 0.022, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.06469351447517387, | |
| "grad_norm": 0.2605821490287781, | |
| "learning_rate": 7.980000000000002e-06, | |
| "loss": 0.022, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.06873685912987224, | |
| "grad_norm": 0.25346869230270386, | |
| "learning_rate": 8.48e-06, | |
| "loss": 0.0226, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.07278020378457059, | |
| "grad_norm": 0.255500465631485, | |
| "learning_rate": 8.98e-06, | |
| "loss": 0.0216, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.07682354843926896, | |
| "grad_norm": 0.24462725222110748, | |
| "learning_rate": 9.48e-06, | |
| "loss": 0.0234, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.08086689309396733, | |
| "grad_norm": 0.23454348742961884, | |
| "learning_rate": 9.980000000000001e-06, | |
| "loss": 0.024, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.0849102377486657, | |
| "grad_norm": 0.2627425193786621, | |
| "learning_rate": 9.946666666666667e-06, | |
| "loss": 0.0226, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.08895358240336407, | |
| "grad_norm": 0.22754116356372833, | |
| "learning_rate": 9.891111111111113e-06, | |
| "loss": 0.023, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.09299692705806242, | |
| "grad_norm": 0.26447415351867676, | |
| "learning_rate": 9.835555555555556e-06, | |
| "loss": 0.0215, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.09704027171276079, | |
| "grad_norm": 0.27743828296661377, | |
| "learning_rate": 9.780000000000001e-06, | |
| "loss": 0.0215, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.10108361636745916, | |
| "grad_norm": 0.23157595098018646, | |
| "learning_rate": 9.724444444444445e-06, | |
| "loss": 0.023, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.10512696102215753, | |
| "grad_norm": 0.2561393082141876, | |
| "learning_rate": 9.66888888888889e-06, | |
| "loss": 0.0209, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.1091703056768559, | |
| "grad_norm": 0.2518314719200134, | |
| "learning_rate": 9.613333333333335e-06, | |
| "loss": 0.0235, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.11321365033155426, | |
| "grad_norm": 0.2346208244562149, | |
| "learning_rate": 9.557777777777777e-06, | |
| "loss": 0.0222, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.11725699498625262, | |
| "grad_norm": 0.23254457116127014, | |
| "learning_rate": 9.502222222222223e-06, | |
| "loss": 0.0216, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.121300339640951, | |
| "grad_norm": 0.2728439271450043, | |
| "learning_rate": 9.446666666666667e-06, | |
| "loss": 0.0208, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.12534368429564935, | |
| "grad_norm": 0.24401770532131195, | |
| "learning_rate": 9.391111111111111e-06, | |
| "loss": 0.0222, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.12938702895034773, | |
| "grad_norm": 0.2546014189720154, | |
| "learning_rate": 9.335555555555557e-06, | |
| "loss": 0.0223, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.1334303736050461, | |
| "grad_norm": 0.20592975616455078, | |
| "learning_rate": 9.280000000000001e-06, | |
| "loss": 0.0214, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.13747371825974447, | |
| "grad_norm": 0.2854577898979187, | |
| "learning_rate": 9.224444444444445e-06, | |
| "loss": 0.021, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.14151706291444283, | |
| "grad_norm": 0.2555174231529236, | |
| "learning_rate": 9.168888888888889e-06, | |
| "loss": 0.0216, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.14556040756914118, | |
| "grad_norm": 0.22724099457263947, | |
| "learning_rate": 9.113333333333335e-06, | |
| "loss": 0.0226, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.14960375222383956, | |
| "grad_norm": 0.24628663063049316, | |
| "learning_rate": 9.057777777777779e-06, | |
| "loss": 0.021, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.15364709687853792, | |
| "grad_norm": 0.22158333659172058, | |
| "learning_rate": 9.002222222222223e-06, | |
| "loss": 0.0221, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.1576904415332363, | |
| "grad_norm": 0.23985975980758667, | |
| "learning_rate": 8.946666666666669e-06, | |
| "loss": 0.0227, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.16173378618793466, | |
| "grad_norm": 0.24263052642345428, | |
| "learning_rate": 8.891111111111111e-06, | |
| "loss": 0.021, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.16173378618793466, | |
| "eval_loss": 0.026818539947271347, | |
| "eval_runtime": 15594.6052, | |
| "eval_samples_per_second": 7.9, | |
| "eval_steps_per_second": 0.494, | |
| "eval_wer": 84.01915040370025, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.165777130842633, | |
| "grad_norm": 0.24622474610805511, | |
| "learning_rate": 8.835555555555557e-06, | |
| "loss": 0.0215, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.1698204754973314, | |
| "grad_norm": 0.218623086810112, | |
| "learning_rate": 8.78e-06, | |
| "loss": 0.0209, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.17386382015202975, | |
| "grad_norm": 0.2270510345697403, | |
| "learning_rate": 8.724444444444445e-06, | |
| "loss": 0.0216, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.17790716480672814, | |
| "grad_norm": 0.2207878828048706, | |
| "learning_rate": 8.66888888888889e-06, | |
| "loss": 0.022, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.1819505094614265, | |
| "grad_norm": 0.18590456247329712, | |
| "learning_rate": 8.613333333333333e-06, | |
| "loss": 0.022, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.18599385411612485, | |
| "grad_norm": 0.1838596612215042, | |
| "learning_rate": 8.557777777777778e-06, | |
| "loss": 0.0215, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.19003719877082323, | |
| "grad_norm": 0.31669357419013977, | |
| "learning_rate": 8.502222222222223e-06, | |
| "loss": 0.0223, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.19408054342552158, | |
| "grad_norm": 0.21895891427993774, | |
| "learning_rate": 8.446666666666668e-06, | |
| "loss": 0.0209, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.19812388808021997, | |
| "grad_norm": 0.20323023200035095, | |
| "learning_rate": 8.391111111111112e-06, | |
| "loss": 0.021, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.20216723273491832, | |
| "grad_norm": 0.25468146800994873, | |
| "learning_rate": 8.335555555555556e-06, | |
| "loss": 0.0224, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.20621057738961668, | |
| "grad_norm": 0.20973573625087738, | |
| "learning_rate": 8.28e-06, | |
| "loss": 0.0211, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.21025392204431506, | |
| "grad_norm": 0.26390501856803894, | |
| "learning_rate": 8.224444444444444e-06, | |
| "loss": 0.0208, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.21429726669901342, | |
| "grad_norm": 0.2301158905029297, | |
| "learning_rate": 8.16888888888889e-06, | |
| "loss": 0.0205, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.2183406113537118, | |
| "grad_norm": 0.24158801138401031, | |
| "learning_rate": 8.113333333333334e-06, | |
| "loss": 0.0206, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.22238395600841016, | |
| "grad_norm": 0.21493862569332123, | |
| "learning_rate": 8.057777777777778e-06, | |
| "loss": 0.0208, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.2264273006631085, | |
| "grad_norm": 0.2373913675546646, | |
| "learning_rate": 8.002222222222222e-06, | |
| "loss": 0.0198, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.2304706453178069, | |
| "grad_norm": 0.2382635623216629, | |
| "learning_rate": 7.946666666666666e-06, | |
| "loss": 0.0198, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.23451398997250525, | |
| "grad_norm": 0.256558895111084, | |
| "learning_rate": 7.891111111111112e-06, | |
| "loss": 0.0206, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.23855733462720363, | |
| "grad_norm": 0.23155981302261353, | |
| "learning_rate": 7.835555555555556e-06, | |
| "loss": 0.0182, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.242600679281902, | |
| "grad_norm": 0.24832689762115479, | |
| "learning_rate": 7.78e-06, | |
| "loss": 0.02, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.24664402393660034, | |
| "grad_norm": 0.22871288657188416, | |
| "learning_rate": 7.724444444444446e-06, | |
| "loss": 0.0193, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.2506873685912987, | |
| "grad_norm": 0.2521936297416687, | |
| "learning_rate": 7.66888888888889e-06, | |
| "loss": 0.0197, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.2547307132459971, | |
| "grad_norm": 0.24429504573345184, | |
| "learning_rate": 7.613333333333334e-06, | |
| "loss": 0.0221, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.25877405790069546, | |
| "grad_norm": 0.23084107041358948, | |
| "learning_rate": 7.557777777777779e-06, | |
| "loss": 0.02, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.2628174025553938, | |
| "grad_norm": 0.2697436809539795, | |
| "learning_rate": 7.502222222222223e-06, | |
| "loss": 0.0211, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.2668607472100922, | |
| "grad_norm": 0.23871037364006042, | |
| "learning_rate": 7.446666666666668e-06, | |
| "loss": 0.0214, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.27090409186479053, | |
| "grad_norm": 0.2677021920681, | |
| "learning_rate": 7.3911111111111125e-06, | |
| "loss": 0.0203, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.27494743651948894, | |
| "grad_norm": 0.1928839236497879, | |
| "learning_rate": 7.335555555555556e-06, | |
| "loss": 0.0203, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.2789907811741873, | |
| "grad_norm": 0.26213887333869934, | |
| "learning_rate": 7.280000000000001e-06, | |
| "loss": 0.0205, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.28303412582888565, | |
| "grad_norm": 0.23492328822612762, | |
| "learning_rate": 7.224444444444445e-06, | |
| "loss": 0.0201, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.287077470483584, | |
| "grad_norm": 0.23946373164653778, | |
| "learning_rate": 7.1688888888888895e-06, | |
| "loss": 0.0195, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.29112081513828236, | |
| "grad_norm": 0.20817314088344574, | |
| "learning_rate": 7.113333333333334e-06, | |
| "loss": 0.0211, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.2951641597929808, | |
| "grad_norm": 0.22937080264091492, | |
| "learning_rate": 7.057777777777778e-06, | |
| "loss": 0.0203, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.29920750444767913, | |
| "grad_norm": 0.23442518711090088, | |
| "learning_rate": 7.0022222222222225e-06, | |
| "loss": 0.0197, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.3032508491023775, | |
| "grad_norm": 0.24289068579673767, | |
| "learning_rate": 6.946666666666667e-06, | |
| "loss": 0.0198, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.30729419375707584, | |
| "grad_norm": 0.24261179566383362, | |
| "learning_rate": 6.891111111111111e-06, | |
| "loss": 0.0211, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.3113375384117742, | |
| "grad_norm": 0.21574310958385468, | |
| "learning_rate": 6.835555555555556e-06, | |
| "loss": 0.0196, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.3153808830664726, | |
| "grad_norm": 0.2648760676383972, | |
| "learning_rate": 6.780000000000001e-06, | |
| "loss": 0.021, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.31942422772117096, | |
| "grad_norm": 0.22739122807979584, | |
| "learning_rate": 6.724444444444444e-06, | |
| "loss": 0.019, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.3234675723758693, | |
| "grad_norm": 0.2167060226202011, | |
| "learning_rate": 6.668888888888889e-06, | |
| "loss": 0.0208, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.3234675723758693, | |
| "eval_loss": 0.025321291759610176, | |
| "eval_runtime": 14275.0619, | |
| "eval_samples_per_second": 8.63, | |
| "eval_steps_per_second": 0.539, | |
| "eval_wer": 82.5098389256299, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.32751091703056767, | |
| "grad_norm": 0.2261808067560196, | |
| "learning_rate": 6.613333333333334e-06, | |
| "loss": 0.0192, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.331554261685266, | |
| "grad_norm": 0.26066452264785767, | |
| "learning_rate": 6.557777777777778e-06, | |
| "loss": 0.0222, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.33559760633996444, | |
| "grad_norm": 0.22417497634887695, | |
| "learning_rate": 6.502222222222223e-06, | |
| "loss": 0.0204, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.3396409509946628, | |
| "grad_norm": 0.20988379418849945, | |
| "learning_rate": 6.446666666666668e-06, | |
| "loss": 0.0191, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.34368429564936115, | |
| "grad_norm": 0.22671236097812653, | |
| "learning_rate": 6.391111111111111e-06, | |
| "loss": 0.0211, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.3477276403040595, | |
| "grad_norm": 0.2695440351963043, | |
| "learning_rate": 6.335555555555556e-06, | |
| "loss": 0.0197, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.35177098495875786, | |
| "grad_norm": 0.20911026000976562, | |
| "learning_rate": 6.280000000000001e-06, | |
| "loss": 0.0191, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.35581432961345627, | |
| "grad_norm": 0.21661245822906494, | |
| "learning_rate": 6.224444444444445e-06, | |
| "loss": 0.0199, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.3598576742681546, | |
| "grad_norm": 0.222182035446167, | |
| "learning_rate": 6.16888888888889e-06, | |
| "loss": 0.0195, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.363901018922853, | |
| "grad_norm": 0.2006169855594635, | |
| "learning_rate": 6.113333333333333e-06, | |
| "loss": 0.0199, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.36794436357755134, | |
| "grad_norm": 0.22298727929592133, | |
| "learning_rate": 6.057777777777778e-06, | |
| "loss": 0.02, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.3719877082322497, | |
| "grad_norm": 0.19001857936382294, | |
| "learning_rate": 6.002222222222223e-06, | |
| "loss": 0.0197, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.3760310528869481, | |
| "grad_norm": 0.2434571385383606, | |
| "learning_rate": 5.946666666666668e-06, | |
| "loss": 0.021, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.38007439754164646, | |
| "grad_norm": 0.21516510844230652, | |
| "learning_rate": 5.891111111111112e-06, | |
| "loss": 0.0193, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.3841177421963448, | |
| "grad_norm": 0.24581994116306305, | |
| "learning_rate": 5.8355555555555565e-06, | |
| "loss": 0.0199, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.38816108685104317, | |
| "grad_norm": 0.23675447702407837, | |
| "learning_rate": 5.78e-06, | |
| "loss": 0.0195, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.3922044315057415, | |
| "grad_norm": 0.21948282420635223, | |
| "learning_rate": 5.724444444444445e-06, | |
| "loss": 0.0194, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.39624777616043994, | |
| "grad_norm": 0.21700553596019745, | |
| "learning_rate": 5.6688888888888895e-06, | |
| "loss": 0.0188, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.4002911208151383, | |
| "grad_norm": 0.24019096791744232, | |
| "learning_rate": 5.613333333333334e-06, | |
| "loss": 0.0211, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.40433446546983665, | |
| "grad_norm": 0.21174991130828857, | |
| "learning_rate": 5.557777777777778e-06, | |
| "loss": 0.0207, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.408377810124535, | |
| "grad_norm": 0.2548556625843048, | |
| "learning_rate": 5.5022222222222224e-06, | |
| "loss": 0.0202, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.41242115477923336, | |
| "grad_norm": 0.2106471061706543, | |
| "learning_rate": 5.4466666666666665e-06, | |
| "loss": 0.0206, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.41646449943393177, | |
| "grad_norm": 0.20076259970664978, | |
| "learning_rate": 5.391111111111111e-06, | |
| "loss": 0.0188, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.4205078440886301, | |
| "grad_norm": 0.20281557738780975, | |
| "learning_rate": 5.335555555555556e-06, | |
| "loss": 0.0189, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.4245511887433285, | |
| "grad_norm": 0.23585236072540283, | |
| "learning_rate": 5.28e-06, | |
| "loss": 0.0193, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.42859453339802683, | |
| "grad_norm": 0.20076635479927063, | |
| "learning_rate": 5.224444444444445e-06, | |
| "loss": 0.0188, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.4326378780527252, | |
| "grad_norm": 0.20363172888755798, | |
| "learning_rate": 5.168888888888889e-06, | |
| "loss": 0.0188, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.4366812227074236, | |
| "grad_norm": 0.20968079566955566, | |
| "learning_rate": 5.113333333333333e-06, | |
| "loss": 0.0198, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.44072456736212196, | |
| "grad_norm": 0.2075379192829132, | |
| "learning_rate": 5.057777777777778e-06, | |
| "loss": 0.0192, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.4447679120168203, | |
| "grad_norm": 0.19762763381004333, | |
| "learning_rate": 5.002222222222223e-06, | |
| "loss": 0.0192, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.44881125667151867, | |
| "grad_norm": 0.2090006172657013, | |
| "learning_rate": 4.946666666666667e-06, | |
| "loss": 0.0196, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.452854601326217, | |
| "grad_norm": 0.20064634084701538, | |
| "learning_rate": 4.891111111111111e-06, | |
| "loss": 0.0187, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.45689794598091543, | |
| "grad_norm": 0.20843106508255005, | |
| "learning_rate": 4.835555555555556e-06, | |
| "loss": 0.0192, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.4609412906356138, | |
| "grad_norm": 0.22503884136676788, | |
| "learning_rate": 4.78e-06, | |
| "loss": 0.0195, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.46498463529031214, | |
| "grad_norm": 0.22823262214660645, | |
| "learning_rate": 4.724444444444445e-06, | |
| "loss": 0.0199, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.4690279799450105, | |
| "grad_norm": 0.24186566472053528, | |
| "learning_rate": 4.66888888888889e-06, | |
| "loss": 0.0194, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.47307132459970885, | |
| "grad_norm": 0.27571889758110046, | |
| "learning_rate": 4.613333333333334e-06, | |
| "loss": 0.0194, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.47711466925440726, | |
| "grad_norm": 0.22955967485904694, | |
| "learning_rate": 4.557777777777778e-06, | |
| "loss": 0.0201, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.4811580139091056, | |
| "grad_norm": 0.21334044635295868, | |
| "learning_rate": 4.502222222222223e-06, | |
| "loss": 0.0204, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.485201358563804, | |
| "grad_norm": 0.20334798097610474, | |
| "learning_rate": 4.446666666666667e-06, | |
| "loss": 0.0207, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.485201358563804, | |
| "eval_loss": 0.024814918637275696, | |
| "eval_runtime": 14244.4286, | |
| "eval_samples_per_second": 8.649, | |
| "eval_steps_per_second": 0.541, | |
| "eval_wer": 81.96210492149146, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.48924470321850233, | |
| "grad_norm": 0.23993884027004242, | |
| "learning_rate": 4.391111111111112e-06, | |
| "loss": 0.0192, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.4932880478732007, | |
| "grad_norm": 0.19718846678733826, | |
| "learning_rate": 4.3355555555555565e-06, | |
| "loss": 0.0202, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.4973313925278991, | |
| "grad_norm": 0.21749120950698853, | |
| "learning_rate": 4.2800000000000005e-06, | |
| "loss": 0.0191, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.5013747371825974, | |
| "grad_norm": 0.22544993460178375, | |
| "learning_rate": 4.2244444444444446e-06, | |
| "loss": 0.0189, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.5054180818372959, | |
| "grad_norm": 0.20528769493103027, | |
| "learning_rate": 4.168888888888889e-06, | |
| "loss": 0.0195, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.5094614264919942, | |
| "grad_norm": 0.2464672178030014, | |
| "learning_rate": 4.1133333333333335e-06, | |
| "loss": 0.0184, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.5135047711466926, | |
| "grad_norm": 0.22547754645347595, | |
| "learning_rate": 4.057777777777778e-06, | |
| "loss": 0.0202, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.5175481158013909, | |
| "grad_norm": 0.1981554925441742, | |
| "learning_rate": 4.002222222222222e-06, | |
| "loss": 0.0196, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.5215914604560893, | |
| "grad_norm": 0.23502561450004578, | |
| "learning_rate": 3.946666666666667e-06, | |
| "loss": 0.0188, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.5256348051107876, | |
| "grad_norm": 0.23244182765483856, | |
| "learning_rate": 3.891111111111111e-06, | |
| "loss": 0.0194, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.529678149765486, | |
| "grad_norm": 0.23334389925003052, | |
| "learning_rate": 3.835555555555555e-06, | |
| "loss": 0.019, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.5337214944201844, | |
| "grad_norm": 0.24255788326263428, | |
| "learning_rate": 3.7800000000000002e-06, | |
| "loss": 0.019, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.5377648390748827, | |
| "grad_norm": 0.19664430618286133, | |
| "learning_rate": 3.724444444444445e-06, | |
| "loss": 0.0186, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.5418081837295811, | |
| "grad_norm": 0.23343808948993683, | |
| "learning_rate": 3.668888888888889e-06, | |
| "loss": 0.0185, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.5458515283842795, | |
| "grad_norm": 0.22261999547481537, | |
| "learning_rate": 3.6133333333333336e-06, | |
| "loss": 0.0185, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.5498948730389779, | |
| "grad_norm": 0.26387473940849304, | |
| "learning_rate": 3.5577777777777785e-06, | |
| "loss": 0.0191, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.5539382176936762, | |
| "grad_norm": 0.2524121105670929, | |
| "learning_rate": 3.5022222222222225e-06, | |
| "loss": 0.019, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.5579815623483746, | |
| "grad_norm": 0.21356618404388428, | |
| "learning_rate": 3.446666666666667e-06, | |
| "loss": 0.0202, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.562024907003073, | |
| "grad_norm": 0.20150095224380493, | |
| "learning_rate": 3.391111111111111e-06, | |
| "loss": 0.0199, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.5660682516577713, | |
| "grad_norm": 0.20904186367988586, | |
| "learning_rate": 3.335555555555556e-06, | |
| "loss": 0.0187, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.5701115963124697, | |
| "grad_norm": 0.2027483582496643, | |
| "learning_rate": 3.2800000000000004e-06, | |
| "loss": 0.0184, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.574154940967168, | |
| "grad_norm": 0.23289534449577332, | |
| "learning_rate": 3.2244444444444444e-06, | |
| "loss": 0.0196, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.5781982856218664, | |
| "grad_norm": 0.23588407039642334, | |
| "learning_rate": 3.1688888888888893e-06, | |
| "loss": 0.0179, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.5822416302765647, | |
| "grad_norm": 0.2624644935131073, | |
| "learning_rate": 3.1133333333333337e-06, | |
| "loss": 0.0186, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.5862849749312632, | |
| "grad_norm": 0.21287862956523895, | |
| "learning_rate": 3.0577777777777778e-06, | |
| "loss": 0.0192, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.5903283195859615, | |
| "grad_norm": 0.24328123033046722, | |
| "learning_rate": 3.0022222222222227e-06, | |
| "loss": 0.0191, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.5943716642406599, | |
| "grad_norm": 0.1968221366405487, | |
| "learning_rate": 2.946666666666667e-06, | |
| "loss": 0.0189, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.5984150088953583, | |
| "grad_norm": 0.22048558294773102, | |
| "learning_rate": 2.891111111111111e-06, | |
| "loss": 0.0193, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.6024583535500566, | |
| "grad_norm": 0.23322191834449768, | |
| "learning_rate": 2.835555555555556e-06, | |
| "loss": 0.0203, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.606501698204755, | |
| "grad_norm": 0.2503873407840729, | |
| "learning_rate": 2.7800000000000005e-06, | |
| "loss": 0.0187, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.6105450428594533, | |
| "grad_norm": 0.23583608865737915, | |
| "learning_rate": 2.7244444444444445e-06, | |
| "loss": 0.0182, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.6145883875141517, | |
| "grad_norm": 0.21786954998970032, | |
| "learning_rate": 2.6688888888888894e-06, | |
| "loss": 0.0187, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.61863173216885, | |
| "grad_norm": 0.21708305180072784, | |
| "learning_rate": 2.6133333333333334e-06, | |
| "loss": 0.0207, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.6226750768235484, | |
| "grad_norm": 0.21468117833137512, | |
| "learning_rate": 2.557777777777778e-06, | |
| "loss": 0.0206, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.6267184214782469, | |
| "grad_norm": 0.2351790964603424, | |
| "learning_rate": 2.5022222222222224e-06, | |
| "loss": 0.0184, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.6307617661329452, | |
| "grad_norm": 0.20362141728401184, | |
| "learning_rate": 2.446666666666667e-06, | |
| "loss": 0.0181, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.6348051107876436, | |
| "grad_norm": 0.22791029512882233, | |
| "learning_rate": 2.3911111111111113e-06, | |
| "loss": 0.0176, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.6388484554423419, | |
| "grad_norm": 0.23184864223003387, | |
| "learning_rate": 2.3355555555555557e-06, | |
| "loss": 0.0201, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.6428918000970403, | |
| "grad_norm": 0.213025763630867, | |
| "learning_rate": 2.28e-06, | |
| "loss": 0.0178, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.6469351447517386, | |
| "grad_norm": 0.2062954306602478, | |
| "learning_rate": 2.2244444444444447e-06, | |
| "loss": 0.0198, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.6469351447517386, | |
| "eval_loss": 0.02430112473666668, | |
| "eval_runtime": 14235.4068, | |
| "eval_samples_per_second": 8.654, | |
| "eval_steps_per_second": 0.541, | |
| "eval_wer": 81.67647178155556, | |
| "step": 4000 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.693893124096e+19, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |