| { |
| "best_global_step": 3750, |
| "best_metric": 37.71349186892212, |
| "best_model_checkpoint": "./wav2vec2-base-sna-cleaned/checkpoint-3750", |
| "epoch": 4.796163069544365, |
| "eval_steps": 250, |
| "global_step": 4000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.02997601918465228, |
| "grad_norm": 87.95906829833984, |
| "learning_rate": 1.8e-06, |
| "loss": 65.4899169921875, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.05995203836930456, |
| "grad_norm": 163.35479736328125, |
| "learning_rate": 3.675e-06, |
| "loss": 59.3804345703125, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.08992805755395683, |
| "grad_norm": 117.41580963134766, |
| "learning_rate": 5.55e-06, |
| "loss": 38.9528857421875, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.11990407673860912, |
| "grad_norm": 42.569610595703125, |
| "learning_rate": 7.425e-06, |
| "loss": 17.7254150390625, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.1498800959232614, |
| "grad_norm": 15.584668159484863, |
| "learning_rate": 9.3e-06, |
| "loss": 13.227435302734374, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.17985611510791366, |
| "grad_norm": 19.06544303894043, |
| "learning_rate": 1.1175e-05, |
| "loss": 12.7419970703125, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.20983213429256595, |
| "grad_norm": 9.686856269836426, |
| "learning_rate": 1.305e-05, |
| "loss": 12.3552099609375, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.23980815347721823, |
| "grad_norm": 19.430574417114258, |
| "learning_rate": 1.4925e-05, |
| "loss": 12.184576416015625, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.2697841726618705, |
| "grad_norm": 8.859325408935547, |
| "learning_rate": 1.6800000000000002e-05, |
| "loss": 11.995687255859375, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.2997601918465228, |
| "grad_norm": 7.194825172424316, |
| "learning_rate": 1.8675000000000003e-05, |
| "loss": 11.898416748046875, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.2997601918465228, |
| "eval_loss": 2.9574084281921387, |
| "eval_runtime": 17.3027, |
| "eval_samples_per_second": 93.627, |
| "eval_steps_per_second": 11.732, |
| "eval_wer": 99.9972760208112, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.32973621103117506, |
| "grad_norm": 6.046307563781738, |
| "learning_rate": 2.055e-05, |
| "loss": 11.7874072265625, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.3597122302158273, |
| "grad_norm": 2.4176154136657715, |
| "learning_rate": 2.2425000000000003e-05, |
| "loss": 11.74424560546875, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.38968824940047964, |
| "grad_norm": 2.773775815963745, |
| "learning_rate": 2.43e-05, |
| "loss": 11.693775634765625, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.4196642685851319, |
| "grad_norm": 5.9898552894592285, |
| "learning_rate": 2.6175000000000003e-05, |
| "loss": 11.669410400390625, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.44964028776978415, |
| "grad_norm": 2.4337611198425293, |
| "learning_rate": 2.805e-05, |
| "loss": 11.601651611328125, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.47961630695443647, |
| "grad_norm": 6.943758964538574, |
| "learning_rate": 2.9925000000000002e-05, |
| "loss": 11.6068701171875, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.5095923261390888, |
| "grad_norm": 9.041548728942871, |
| "learning_rate": 2.98e-05, |
| "loss": 11.59481201171875, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.539568345323741, |
| "grad_norm": 2.0354104042053223, |
| "learning_rate": 2.9591666666666667e-05, |
| "loss": 11.573612060546875, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.5695443645083933, |
| "grad_norm": 1.5390305519104004, |
| "learning_rate": 2.9383333333333334e-05, |
| "loss": 11.53906005859375, |
| "step": 475 |
| }, |
| { |
| "epoch": 0.5995203836930456, |
| "grad_norm": 5.494091510772705, |
| "learning_rate": 2.9175e-05, |
| "loss": 11.54445556640625, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.5995203836930456, |
| "eval_loss": 2.8992974758148193, |
| "eval_runtime": 17.2261, |
| "eval_samples_per_second": 94.044, |
| "eval_steps_per_second": 11.784, |
| "eval_wer": 99.9972760208112, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.6294964028776978, |
| "grad_norm": 2.440443515777588, |
| "learning_rate": 2.8966666666666665e-05, |
| "loss": 11.5015576171875, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.6594724220623501, |
| "grad_norm": 4.654289722442627, |
| "learning_rate": 2.8758333333333332e-05, |
| "loss": 11.103582763671875, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.6894484412470024, |
| "grad_norm": 6.255335330963135, |
| "learning_rate": 2.855e-05, |
| "loss": 10.390032958984374, |
| "step": 575 |
| }, |
| { |
| "epoch": 0.7194244604316546, |
| "grad_norm": 10.0020112991333, |
| "learning_rate": 2.8341666666666667e-05, |
| "loss": 9.397643432617187, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.749400479616307, |
| "grad_norm": 7.839452266693115, |
| "learning_rate": 2.8133333333333334e-05, |
| "loss": 7.82263671875, |
| "step": 625 |
| }, |
| { |
| "epoch": 0.7793764988009593, |
| "grad_norm": 7.001938343048096, |
| "learning_rate": 2.7924999999999998e-05, |
| "loss": 6.276546020507812, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.8093525179856115, |
| "grad_norm": 6.308192729949951, |
| "learning_rate": 2.771666666666667e-05, |
| "loss": 5.47916259765625, |
| "step": 675 |
| }, |
| { |
| "epoch": 0.8393285371702638, |
| "grad_norm": 10.70156478881836, |
| "learning_rate": 2.7508333333333336e-05, |
| "loss": 4.5777520751953125, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.8693045563549161, |
| "grad_norm": 7.3394551277160645, |
| "learning_rate": 2.7300000000000003e-05, |
| "loss": 4.2768948364257815, |
| "step": 725 |
| }, |
| { |
| "epoch": 0.8992805755395683, |
| "grad_norm": 9.546968460083008, |
| "learning_rate": 2.7091666666666667e-05, |
| "loss": 3.8439666748046877, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.8992805755395683, |
| "eval_loss": 0.814090371131897, |
| "eval_runtime": 17.3358, |
| "eval_samples_per_second": 93.448, |
| "eval_steps_per_second": 11.71, |
| "eval_wer": 98.76603742747405, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.9292565947242206, |
| "grad_norm": 5.7955803871154785, |
| "learning_rate": 2.6883333333333334e-05, |
| "loss": 3.4320477294921874, |
| "step": 775 |
| }, |
| { |
| "epoch": 0.9592326139088729, |
| "grad_norm": 7.267416477203369, |
| "learning_rate": 2.6675e-05, |
| "loss": 3.3107342529296875, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.9892086330935251, |
| "grad_norm": 5.454007148742676, |
| "learning_rate": 2.646666666666667e-05, |
| "loss": 3.279764404296875, |
| "step": 825 |
| }, |
| { |
| "epoch": 1.0191846522781776, |
| "grad_norm": 6.052438735961914, |
| "learning_rate": 2.6258333333333336e-05, |
| "loss": 2.929551086425781, |
| "step": 850 |
| }, |
| { |
| "epoch": 1.0491606714628297, |
| "grad_norm": 5.4171319007873535, |
| "learning_rate": 2.605e-05, |
| "loss": 2.633551025390625, |
| "step": 875 |
| }, |
| { |
| "epoch": 1.079136690647482, |
| "grad_norm": 6.3133320808410645, |
| "learning_rate": 2.5841666666666667e-05, |
| "loss": 2.9070010375976563, |
| "step": 900 |
| }, |
| { |
| "epoch": 1.1091127098321343, |
| "grad_norm": 5.874911785125732, |
| "learning_rate": 2.5633333333333334e-05, |
| "loss": 2.554573516845703, |
| "step": 925 |
| }, |
| { |
| "epoch": 1.1390887290167866, |
| "grad_norm": 4.682479381561279, |
| "learning_rate": 2.5425e-05, |
| "loss": 2.424696960449219, |
| "step": 950 |
| }, |
| { |
| "epoch": 1.169064748201439, |
| "grad_norm": 5.821262836456299, |
| "learning_rate": 2.5216666666666665e-05, |
| "loss": 2.3822782897949217, |
| "step": 975 |
| }, |
| { |
| "epoch": 1.1990407673860912, |
| "grad_norm": 4.946887969970703, |
| "learning_rate": 2.5008333333333332e-05, |
| "loss": 2.3326264953613283, |
| "step": 1000 |
| }, |
| { |
| "epoch": 1.1990407673860912, |
| "eval_loss": 0.5055655241012573, |
| "eval_runtime": 17.356, |
| "eval_samples_per_second": 93.339, |
| "eval_steps_per_second": 11.696, |
| "eval_wer": 56.12759118520334, |
| "step": 1000 |
| }, |
| { |
| "epoch": 1.2290167865707433, |
| "grad_norm": 5.456410884857178, |
| "learning_rate": 2.48e-05, |
| "loss": 2.232008819580078, |
| "step": 1025 |
| }, |
| { |
| "epoch": 1.2589928057553956, |
| "grad_norm": 12.988375663757324, |
| "learning_rate": 2.4591666666666667e-05, |
| "loss": 2.308787841796875, |
| "step": 1050 |
| }, |
| { |
| "epoch": 1.288968824940048, |
| "grad_norm": 5.91463565826416, |
| "learning_rate": 2.4383333333333334e-05, |
| "loss": 2.133858489990234, |
| "step": 1075 |
| }, |
| { |
| "epoch": 1.3189448441247003, |
| "grad_norm": 5.080030918121338, |
| "learning_rate": 2.4174999999999998e-05, |
| "loss": 2.111269836425781, |
| "step": 1100 |
| }, |
| { |
| "epoch": 1.3489208633093526, |
| "grad_norm": 6.5078959465026855, |
| "learning_rate": 2.3966666666666665e-05, |
| "loss": 1.9835806274414063, |
| "step": 1125 |
| }, |
| { |
| "epoch": 1.3788968824940047, |
| "grad_norm": 6.049032688140869, |
| "learning_rate": 2.3758333333333333e-05, |
| "loss": 1.948318328857422, |
| "step": 1150 |
| }, |
| { |
| "epoch": 1.4088729016786572, |
| "grad_norm": 4.487147808074951, |
| "learning_rate": 2.3550000000000003e-05, |
| "loss": 1.9143856811523436, |
| "step": 1175 |
| }, |
| { |
| "epoch": 1.4388489208633093, |
| "grad_norm": 4.666418075561523, |
| "learning_rate": 2.3341666666666667e-05, |
| "loss": 1.9930302429199218, |
| "step": 1200 |
| }, |
| { |
| "epoch": 1.4688249400479616, |
| "grad_norm": 5.2837233543396, |
| "learning_rate": 2.3133333333333334e-05, |
| "loss": 1.9030076599121093, |
| "step": 1225 |
| }, |
| { |
| "epoch": 1.498800959232614, |
| "grad_norm": 4.892092227935791, |
| "learning_rate": 2.2925e-05, |
| "loss": 1.993743896484375, |
| "step": 1250 |
| }, |
| { |
| "epoch": 1.498800959232614, |
| "eval_loss": 0.4353557825088501, |
| "eval_runtime": 17.8784, |
| "eval_samples_per_second": 90.612, |
| "eval_steps_per_second": 11.355, |
| "eval_wer": 49.77254773773529, |
| "step": 1250 |
| }, |
| { |
| "epoch": 1.5287769784172662, |
| "grad_norm": 4.702995777130127, |
| "learning_rate": 2.271666666666667e-05, |
| "loss": 1.8019261169433594, |
| "step": 1275 |
| }, |
| { |
| "epoch": 1.5587529976019185, |
| "grad_norm": 5.374894618988037, |
| "learning_rate": 2.2508333333333336e-05, |
| "loss": 1.9548370361328125, |
| "step": 1300 |
| }, |
| { |
| "epoch": 1.5887290167865706, |
| "grad_norm": 5.36961555480957, |
| "learning_rate": 2.23e-05, |
| "loss": 1.6951516723632813, |
| "step": 1325 |
| }, |
| { |
| "epoch": 1.6187050359712232, |
| "grad_norm": 5.167247772216797, |
| "learning_rate": 2.2091666666666667e-05, |
| "loss": 1.887311248779297, |
| "step": 1350 |
| }, |
| { |
| "epoch": 1.6486810551558753, |
| "grad_norm": 4.589432716369629, |
| "learning_rate": 2.1883333333333334e-05, |
| "loss": 1.8049166870117188, |
| "step": 1375 |
| }, |
| { |
| "epoch": 1.6786570743405276, |
| "grad_norm": 4.192942142486572, |
| "learning_rate": 2.1675e-05, |
| "loss": 1.8473945617675782, |
| "step": 1400 |
| }, |
| { |
| "epoch": 1.70863309352518, |
| "grad_norm": 4.900947570800781, |
| "learning_rate": 2.1466666666666666e-05, |
| "loss": 1.7062265014648437, |
| "step": 1425 |
| }, |
| { |
| "epoch": 1.738609112709832, |
| "grad_norm": 4.3333587646484375, |
| "learning_rate": 2.1258333333333333e-05, |
| "loss": 1.7642361450195312, |
| "step": 1450 |
| }, |
| { |
| "epoch": 1.7685851318944845, |
| "grad_norm": 5.301547050476074, |
| "learning_rate": 2.105e-05, |
| "loss": 1.6472731018066407, |
| "step": 1475 |
| }, |
| { |
| "epoch": 1.7985611510791366, |
| "grad_norm": 4.614596366882324, |
| "learning_rate": 2.0841666666666667e-05, |
| "loss": 1.8639109802246094, |
| "step": 1500 |
| }, |
| { |
| "epoch": 1.7985611510791366, |
| "eval_loss": 0.3815246522426605, |
| "eval_runtime": 17.5598, |
| "eval_samples_per_second": 92.256, |
| "eval_steps_per_second": 11.561, |
| "eval_wer": 45.07640761624581, |
| "step": 1500 |
| }, |
| { |
| "epoch": 1.828537170263789, |
| "grad_norm": 4.178606033325195, |
| "learning_rate": 2.0633333333333335e-05, |
| "loss": 1.5441552734375, |
| "step": 1525 |
| }, |
| { |
| "epoch": 1.8585131894484412, |
| "grad_norm": 4.956863880157471, |
| "learning_rate": 2.0425e-05, |
| "loss": 1.9137911987304688, |
| "step": 1550 |
| }, |
| { |
| "epoch": 1.8884892086330936, |
| "grad_norm": 4.828672885894775, |
| "learning_rate": 2.0216666666666666e-05, |
| "loss": 1.5100914001464845, |
| "step": 1575 |
| }, |
| { |
| "epoch": 1.9184652278177459, |
| "grad_norm": 3.7223966121673584, |
| "learning_rate": 2.0008333333333333e-05, |
| "loss": 1.688656768798828, |
| "step": 1600 |
| }, |
| { |
| "epoch": 1.948441247002398, |
| "grad_norm": 5.0394062995910645, |
| "learning_rate": 1.98e-05, |
| "loss": 1.5531495666503907, |
| "step": 1625 |
| }, |
| { |
| "epoch": 1.9784172661870505, |
| "grad_norm": 4.3810954093933105, |
| "learning_rate": 1.9591666666666664e-05, |
| "loss": 1.7138851928710936, |
| "step": 1650 |
| }, |
| { |
| "epoch": 2.0083932853717026, |
| "grad_norm": 3.4982099533081055, |
| "learning_rate": 1.938333333333333e-05, |
| "loss": 1.5514004516601563, |
| "step": 1675 |
| }, |
| { |
| "epoch": 2.038369304556355, |
| "grad_norm": 4.8457136154174805, |
| "learning_rate": 1.9175000000000002e-05, |
| "loss": 1.5973490905761718, |
| "step": 1700 |
| }, |
| { |
| "epoch": 2.068345323741007, |
| "grad_norm": 4.1086602210998535, |
| "learning_rate": 1.896666666666667e-05, |
| "loss": 1.5693655395507813, |
| "step": 1725 |
| }, |
| { |
| "epoch": 2.0983213429256593, |
| "grad_norm": 4.984743118286133, |
| "learning_rate": 1.8758333333333336e-05, |
| "loss": 1.5768695068359375, |
| "step": 1750 |
| }, |
| { |
| "epoch": 2.0983213429256593, |
| "eval_loss": 0.3492273688316345, |
| "eval_runtime": 17.4669, |
| "eval_samples_per_second": 92.747, |
| "eval_steps_per_second": 11.622, |
| "eval_wer": 43.68445425076952, |
| "step": 1750 |
| }, |
| { |
| "epoch": 2.128297362110312, |
| "grad_norm": 4.139465808868408, |
| "learning_rate": 1.855e-05, |
| "loss": 1.5272630310058595, |
| "step": 1775 |
| }, |
| { |
| "epoch": 2.158273381294964, |
| "grad_norm": 4.634520053863525, |
| "learning_rate": 1.8341666666666668e-05, |
| "loss": 1.4187347412109375, |
| "step": 1800 |
| }, |
| { |
| "epoch": 2.1882494004796165, |
| "grad_norm": 3.5434200763702393, |
| "learning_rate": 1.8133333333333335e-05, |
| "loss": 1.5776219177246094, |
| "step": 1825 |
| }, |
| { |
| "epoch": 2.2182254196642686, |
| "grad_norm": 4.246279716491699, |
| "learning_rate": 1.7925000000000002e-05, |
| "loss": 1.4354205322265625, |
| "step": 1850 |
| }, |
| { |
| "epoch": 2.2482014388489207, |
| "grad_norm": 3.7664794921875, |
| "learning_rate": 1.7716666666666666e-05, |
| "loss": 1.4057354736328125, |
| "step": 1875 |
| }, |
| { |
| "epoch": 2.278177458033573, |
| "grad_norm": 5.111608028411865, |
| "learning_rate": 1.7508333333333333e-05, |
| "loss": 1.5491357421875, |
| "step": 1900 |
| }, |
| { |
| "epoch": 2.3081534772182253, |
| "grad_norm": 4.471536636352539, |
| "learning_rate": 1.73e-05, |
| "loss": 1.4842294311523438, |
| "step": 1925 |
| }, |
| { |
| "epoch": 2.338129496402878, |
| "grad_norm": 3.7081987857818604, |
| "learning_rate": 1.7091666666666668e-05, |
| "loss": 1.450211181640625, |
| "step": 1950 |
| }, |
| { |
| "epoch": 2.36810551558753, |
| "grad_norm": 3.8253395557403564, |
| "learning_rate": 1.6883333333333335e-05, |
| "loss": 1.376385498046875, |
| "step": 1975 |
| }, |
| { |
| "epoch": 2.3980815347721824, |
| "grad_norm": 3.773341417312622, |
| "learning_rate": 1.6675e-05, |
| "loss": 1.4288015747070313, |
| "step": 2000 |
| }, |
| { |
| "epoch": 2.3980815347721824, |
| "eval_loss": 0.33844250440597534, |
| "eval_runtime": 17.2583, |
| "eval_samples_per_second": 93.868, |
| "eval_steps_per_second": 11.762, |
| "eval_wer": 41.22470104328403, |
| "step": 2000 |
| }, |
| { |
| "epoch": 2.4280575539568345, |
| "grad_norm": 3.5804367065429688, |
| "learning_rate": 1.6466666666666666e-05, |
| "loss": 1.4616065979003907, |
| "step": 2025 |
| }, |
| { |
| "epoch": 2.4580335731414866, |
| "grad_norm": 4.723759651184082, |
| "learning_rate": 1.6258333333333333e-05, |
| "loss": 1.3814134216308593, |
| "step": 2050 |
| }, |
| { |
| "epoch": 2.488009592326139, |
| "grad_norm": 3.2633466720581055, |
| "learning_rate": 1.605e-05, |
| "loss": 1.393878173828125, |
| "step": 2075 |
| }, |
| { |
| "epoch": 2.5179856115107913, |
| "grad_norm": 4.216237545013428, |
| "learning_rate": 1.5841666666666664e-05, |
| "loss": 1.40193603515625, |
| "step": 2100 |
| }, |
| { |
| "epoch": 2.547961630695444, |
| "grad_norm": 3.037777900695801, |
| "learning_rate": 1.563333333333333e-05, |
| "loss": 1.3580673217773438, |
| "step": 2125 |
| }, |
| { |
| "epoch": 2.577937649880096, |
| "grad_norm": 5.0248847007751465, |
| "learning_rate": 1.5425e-05, |
| "loss": 1.572769012451172, |
| "step": 2150 |
| }, |
| { |
| "epoch": 2.6079136690647484, |
| "grad_norm": 3.7339730262756348, |
| "learning_rate": 1.5216666666666668e-05, |
| "loss": 1.5580809020996094, |
| "step": 2175 |
| }, |
| { |
| "epoch": 2.6378896882494005, |
| "grad_norm": 3.7687699794769287, |
| "learning_rate": 1.5008333333333335e-05, |
| "loss": 1.427168731689453, |
| "step": 2200 |
| }, |
| { |
| "epoch": 2.6678657074340526, |
| "grad_norm": 3.9816343784332275, |
| "learning_rate": 1.48e-05, |
| "loss": 1.3857545471191406, |
| "step": 2225 |
| }, |
| { |
| "epoch": 2.697841726618705, |
| "grad_norm": 3.972137212753296, |
| "learning_rate": 1.4591666666666666e-05, |
| "loss": 1.331875, |
| "step": 2250 |
| }, |
| { |
| "epoch": 2.697841726618705, |
| "eval_loss": 0.3375837802886963, |
| "eval_runtime": 17.4226, |
| "eval_samples_per_second": 92.983, |
| "eval_steps_per_second": 11.652, |
| "eval_wer": 40.461986870420304, |
| "step": 2250 |
| }, |
| { |
| "epoch": 2.7278177458033572, |
| "grad_norm": 3.696455717086792, |
| "learning_rate": 1.4383333333333333e-05, |
| "loss": 1.2858570861816405, |
| "step": 2275 |
| }, |
| { |
| "epoch": 2.7577937649880093, |
| "grad_norm": 3.8930845260620117, |
| "learning_rate": 1.4174999999999999e-05, |
| "loss": 1.3353477478027345, |
| "step": 2300 |
| }, |
| { |
| "epoch": 2.787769784172662, |
| "grad_norm": 3.240628719329834, |
| "learning_rate": 1.3966666666666666e-05, |
| "loss": 1.5282106018066406, |
| "step": 2325 |
| }, |
| { |
| "epoch": 2.8177458033573144, |
| "grad_norm": 4.53059196472168, |
| "learning_rate": 1.3758333333333335e-05, |
| "loss": 1.325364990234375, |
| "step": 2350 |
| }, |
| { |
| "epoch": 2.8477218225419665, |
| "grad_norm": 3.9437687397003174, |
| "learning_rate": 1.355e-05, |
| "loss": 1.3545150756835938, |
| "step": 2375 |
| }, |
| { |
| "epoch": 2.8776978417266186, |
| "grad_norm": 4.234551906585693, |
| "learning_rate": 1.3341666666666668e-05, |
| "loss": 1.3189064025878907, |
| "step": 2400 |
| }, |
| { |
| "epoch": 2.907673860911271, |
| "grad_norm": 3.8854103088378906, |
| "learning_rate": 1.3133333333333334e-05, |
| "loss": 1.3604727172851563, |
| "step": 2425 |
| }, |
| { |
| "epoch": 2.937649880095923, |
| "grad_norm": 4.403459072113037, |
| "learning_rate": 1.2925e-05, |
| "loss": 1.3202362060546875, |
| "step": 2450 |
| }, |
| { |
| "epoch": 2.9676258992805753, |
| "grad_norm": 3.7993664741516113, |
| "learning_rate": 1.2716666666666666e-05, |
| "loss": 1.3466416931152343, |
| "step": 2475 |
| }, |
| { |
| "epoch": 2.997601918465228, |
| "grad_norm": 4.246622562408447, |
| "learning_rate": 1.2508333333333334e-05, |
| "loss": 1.255322494506836, |
| "step": 2500 |
| }, |
| { |
| "epoch": 2.997601918465228, |
| "eval_loss": 0.32123640179634094, |
| "eval_runtime": 17.9027, |
| "eval_samples_per_second": 90.489, |
| "eval_steps_per_second": 11.339, |
| "eval_wer": 39.50587017515186, |
| "step": 2500 |
| }, |
| { |
| "epoch": 3.02757793764988, |
| "grad_norm": 4.5194411277771, |
| "learning_rate": 1.2299999999999999e-05, |
| "loss": 1.3707557678222657, |
| "step": 2525 |
| }, |
| { |
| "epoch": 3.0575539568345325, |
| "grad_norm": 4.761359214782715, |
| "learning_rate": 1.2091666666666666e-05, |
| "loss": 1.2971902465820313, |
| "step": 2550 |
| }, |
| { |
| "epoch": 3.0875299760191846, |
| "grad_norm": 3.902237892150879, |
| "learning_rate": 1.1883333333333334e-05, |
| "loss": 1.3517869567871095, |
| "step": 2575 |
| }, |
| { |
| "epoch": 3.117505995203837, |
| "grad_norm": 4.836638450622559, |
| "learning_rate": 1.1675000000000001e-05, |
| "loss": 1.204417724609375, |
| "step": 2600 |
| }, |
| { |
| "epoch": 3.147482014388489, |
| "grad_norm": 3.637927770614624, |
| "learning_rate": 1.1466666666666668e-05, |
| "loss": 1.3902195739746093, |
| "step": 2625 |
| }, |
| { |
| "epoch": 3.1774580335731413, |
| "grad_norm": 5.192939281463623, |
| "learning_rate": 1.1258333333333334e-05, |
| "loss": 1.210543670654297, |
| "step": 2650 |
| }, |
| { |
| "epoch": 3.207434052757794, |
| "grad_norm": 3.76882004737854, |
| "learning_rate": 1.1050000000000001e-05, |
| "loss": 1.292550048828125, |
| "step": 2675 |
| }, |
| { |
| "epoch": 3.237410071942446, |
| "grad_norm": 4.45828104019165, |
| "learning_rate": 1.0841666666666666e-05, |
| "loss": 1.1661454010009766, |
| "step": 2700 |
| }, |
| { |
| "epoch": 3.2673860911270984, |
| "grad_norm": 3.8488924503326416, |
| "learning_rate": 1.0633333333333334e-05, |
| "loss": 1.3198445129394532, |
| "step": 2725 |
| }, |
| { |
| "epoch": 3.2973621103117505, |
| "grad_norm": 4.573055744171143, |
| "learning_rate": 1.0425e-05, |
| "loss": 1.215043182373047, |
| "step": 2750 |
| }, |
| { |
| "epoch": 3.2973621103117505, |
| "eval_loss": 0.3164944350719452, |
| "eval_runtime": 17.9593, |
| "eval_samples_per_second": 90.204, |
| "eval_steps_per_second": 11.303, |
| "eval_wer": 39.07003350494402, |
| "step": 2750 |
| }, |
| { |
| "epoch": 3.327338129496403, |
| "grad_norm": 3.0178797245025635, |
| "learning_rate": 1.0216666666666667e-05, |
| "loss": 1.3084765625, |
| "step": 2775 |
| }, |
| { |
| "epoch": 3.357314148681055, |
| "grad_norm": 4.195224285125732, |
| "learning_rate": 1.0008333333333334e-05, |
| "loss": 1.2380980682373046, |
| "step": 2800 |
| }, |
| { |
| "epoch": 3.3872901678657072, |
| "grad_norm": 3.9489684104919434, |
| "learning_rate": 9.8e-06, |
| "loss": 1.3261349487304688, |
| "step": 2825 |
| }, |
| { |
| "epoch": 3.41726618705036, |
| "grad_norm": 4.707721710205078, |
| "learning_rate": 9.591666666666668e-06, |
| "loss": 1.166817626953125, |
| "step": 2850 |
| }, |
| { |
| "epoch": 3.447242206235012, |
| "grad_norm": 3.6897363662719727, |
| "learning_rate": 9.383333333333334e-06, |
| "loss": 1.3451451110839843, |
| "step": 2875 |
| }, |
| { |
| "epoch": 3.4772182254196644, |
| "grad_norm": 3.894786834716797, |
| "learning_rate": 9.175000000000001e-06, |
| "loss": 1.215664825439453, |
| "step": 2900 |
| }, |
| { |
| "epoch": 3.5071942446043165, |
| "grad_norm": 4.44349479675293, |
| "learning_rate": 8.966666666666667e-06, |
| "loss": 1.2665501403808594, |
| "step": 2925 |
| }, |
| { |
| "epoch": 3.537170263788969, |
| "grad_norm": 4.028308868408203, |
| "learning_rate": 8.758333333333334e-06, |
| "loss": 1.1728402709960937, |
| "step": 2950 |
| }, |
| { |
| "epoch": 3.567146282973621, |
| "grad_norm": 3.627992630004883, |
| "learning_rate": 8.55e-06, |
| "loss": 1.2784690856933594, |
| "step": 2975 |
| }, |
| { |
| "epoch": 3.597122302158273, |
| "grad_norm": 3.8370578289031982, |
| "learning_rate": 8.341666666666667e-06, |
| "loss": 1.1342899322509765, |
| "step": 3000 |
| }, |
| { |
| "epoch": 3.597122302158273, |
| "eval_loss": 0.310255765914917, |
| "eval_runtime": 18.0427, |
| "eval_samples_per_second": 89.787, |
| "eval_steps_per_second": 11.251, |
| "eval_wer": 38.22287597722753, |
| "step": 3000 |
| }, |
| { |
| "epoch": 3.6270983213429258, |
| "grad_norm": 2.9377591609954834, |
| "learning_rate": 8.133333333333334e-06, |
| "loss": 1.3874595642089844, |
| "step": 3025 |
| }, |
| { |
| "epoch": 3.657074340527578, |
| "grad_norm": 5.296197414398193, |
| "learning_rate": 7.925e-06, |
| "loss": 1.1686283111572267, |
| "step": 3050 |
| }, |
| { |
| "epoch": 3.68705035971223, |
| "grad_norm": 3.4494028091430664, |
| "learning_rate": 7.716666666666667e-06, |
| "loss": 1.202396469116211, |
| "step": 3075 |
| }, |
| { |
| "epoch": 3.7170263788968825, |
| "grad_norm": 4.737746238708496, |
| "learning_rate": 7.508333333333333e-06, |
| "loss": 1.1863296508789063, |
| "step": 3100 |
| }, |
| { |
| "epoch": 3.747002398081535, |
| "grad_norm": 3.7916765213012695, |
| "learning_rate": 7.3e-06, |
| "loss": 1.3395465087890626, |
| "step": 3125 |
| }, |
| { |
| "epoch": 3.776978417266187, |
| "grad_norm": 4.551313400268555, |
| "learning_rate": 7.091666666666667e-06, |
| "loss": 1.1056033325195314, |
| "step": 3150 |
| }, |
| { |
| "epoch": 3.806954436450839, |
| "grad_norm": 4.075565338134766, |
| "learning_rate": 6.883333333333334e-06, |
| "loss": 1.3138325500488282, |
| "step": 3175 |
| }, |
| { |
| "epoch": 3.8369304556354917, |
| "grad_norm": 3.972641706466675, |
| "learning_rate": 6.6750000000000005e-06, |
| "loss": 1.1527600860595704, |
| "step": 3200 |
| }, |
| { |
| "epoch": 3.866906474820144, |
| "grad_norm": 3.812631845474243, |
| "learning_rate": 6.466666666666667e-06, |
| "loss": 1.2789608001708985, |
| "step": 3225 |
| }, |
| { |
| "epoch": 3.896882494004796, |
| "grad_norm": 5.695910930633545, |
| "learning_rate": 6.258333333333333e-06, |
| "loss": 1.1754793548583984, |
| "step": 3250 |
| }, |
| { |
| "epoch": 3.896882494004796, |
| "eval_loss": 0.3044012486934662, |
| "eval_runtime": 18.0122, |
| "eval_samples_per_second": 89.939, |
| "eval_steps_per_second": 11.27, |
| "eval_wer": 38.53613358393942, |
| "step": 3250 |
| }, |
| { |
| "epoch": 3.9268585131894485, |
| "grad_norm": 3.0165011882781982, |
| "learning_rate": 6.05e-06, |
| "loss": 1.3710069274902343, |
| "step": 3275 |
| }, |
| { |
| "epoch": 3.956834532374101, |
| "grad_norm": 4.425358295440674, |
| "learning_rate": 5.841666666666667e-06, |
| "loss": 1.1081704711914062, |
| "step": 3300 |
| }, |
| { |
| "epoch": 3.986810551558753, |
| "grad_norm": 4.782137393951416, |
| "learning_rate": 5.633333333333333e-06, |
| "loss": 1.438135986328125, |
| "step": 3325 |
| }, |
| { |
| "epoch": 4.016786570743405, |
| "grad_norm": 3.2695047855377197, |
| "learning_rate": 5.4250000000000006e-06, |
| "loss": 1.1687757873535156, |
| "step": 3350 |
| }, |
| { |
| "epoch": 4.046762589928058, |
| "grad_norm": 3.6320788860321045, |
| "learning_rate": 5.216666666666667e-06, |
| "loss": 1.139478988647461, |
| "step": 3375 |
| }, |
| { |
| "epoch": 4.07673860911271, |
| "grad_norm": 3.5363335609436035, |
| "learning_rate": 5.008333333333333e-06, |
| "loss": 1.2308349609375, |
| "step": 3400 |
| }, |
| { |
| "epoch": 4.106714628297362, |
| "grad_norm": 3.7217462062835693, |
| "learning_rate": 4.800000000000001e-06, |
| "loss": 1.1875200653076172, |
| "step": 3425 |
| }, |
| { |
| "epoch": 4.136690647482014, |
| "grad_norm": 3.1459712982177734, |
| "learning_rate": 4.591666666666667e-06, |
| "loss": 1.1901798248291016, |
| "step": 3450 |
| }, |
| { |
| "epoch": 4.166666666666667, |
| "grad_norm": 3.7371957302093506, |
| "learning_rate": 4.3833333333333334e-06, |
| "loss": 1.0538123321533204, |
| "step": 3475 |
| }, |
| { |
| "epoch": 4.196642685851319, |
| "grad_norm": 3.7070775032043457, |
| "learning_rate": 4.175e-06, |
| "loss": 1.193154525756836, |
| "step": 3500 |
| }, |
| { |
| "epoch": 4.196642685851319, |
| "eval_loss": 0.30022621154785156, |
| "eval_runtime": 18.115, |
| "eval_samples_per_second": 89.428, |
| "eval_steps_per_second": 11.206, |
| "eval_wer": 37.855138786739666, |
| "step": 3500 |
| }, |
| { |
| "epoch": 4.226618705035971, |
| "grad_norm": 3.048323154449463, |
| "learning_rate": 3.966666666666666e-06, |
| "loss": 1.1228433990478515, |
| "step": 3525 |
| }, |
| { |
| "epoch": 4.256594724220624, |
| "grad_norm": 3.1846795082092285, |
| "learning_rate": 3.758333333333333e-06, |
| "loss": 1.2151748657226562, |
| "step": 3550 |
| }, |
| { |
| "epoch": 4.286570743405276, |
| "grad_norm": 4.277435302734375, |
| "learning_rate": 3.55e-06, |
| "loss": 1.146432876586914, |
| "step": 3575 |
| }, |
| { |
| "epoch": 4.316546762589928, |
| "grad_norm": 3.3862087726593018, |
| "learning_rate": 3.3416666666666667e-06, |
| "loss": 1.2336856079101564, |
| "step": 3600 |
| }, |
| { |
| "epoch": 4.34652278177458, |
| "grad_norm": 4.78780460357666, |
| "learning_rate": 3.1333333333333335e-06, |
| "loss": 1.1951210021972656, |
| "step": 3625 |
| }, |
| { |
| "epoch": 4.376498800959233, |
| "grad_norm": 3.9586503505706787, |
| "learning_rate": 2.9250000000000004e-06, |
| "loss": 1.227743148803711, |
| "step": 3650 |
| }, |
| { |
| "epoch": 4.406474820143885, |
| "grad_norm": 4.07125186920166, |
| "learning_rate": 2.7166666666666668e-06, |
| "loss": 1.1534527587890624, |
| "step": 3675 |
| }, |
| { |
| "epoch": 4.436450839328537, |
| "grad_norm": 3.8952131271362305, |
| "learning_rate": 2.508333333333333e-06, |
| "loss": 1.1760161590576172, |
| "step": 3700 |
| }, |
| { |
| "epoch": 4.46642685851319, |
| "grad_norm": 4.782440185546875, |
| "learning_rate": 2.3e-06, |
| "loss": 1.0955001068115235, |
| "step": 3725 |
| }, |
| { |
| "epoch": 4.496402877697841, |
| "grad_norm": 3.4196395874023438, |
| "learning_rate": 2.091666666666667e-06, |
| "loss": 1.3582955932617187, |
| "step": 3750 |
| }, |
| { |
| "epoch": 4.496402877697841, |
| "eval_loss": 0.29945749044418335, |
| "eval_runtime": 18.1526, |
| "eval_samples_per_second": 89.244, |
| "eval_steps_per_second": 11.183, |
| "eval_wer": 37.71349186892212, |
| "step": 3750 |
| }, |
| { |
| "epoch": 4.526378896882494, |
| "grad_norm": 4.029821395874023, |
| "learning_rate": 1.8833333333333334e-06, |
| "loss": 1.050459747314453, |
| "step": 3775 |
| }, |
| { |
| "epoch": 4.556354916067146, |
| "grad_norm": 3.5950191020965576, |
| "learning_rate": 1.675e-06, |
| "loss": 1.2073076629638673, |
| "step": 3800 |
| }, |
| { |
| "epoch": 4.586330935251799, |
| "grad_norm": 3.9220163822174072, |
| "learning_rate": 1.4666666666666669e-06, |
| "loss": 1.0614564514160156, |
| "step": 3825 |
| }, |
| { |
| "epoch": 4.616306954436451, |
| "grad_norm": 28.386432647705078, |
| "learning_rate": 1.2583333333333333e-06, |
| "loss": 1.43855224609375, |
| "step": 3850 |
| }, |
| { |
| "epoch": 4.646282973621103, |
| "grad_norm": 3.9794702529907227, |
| "learning_rate": 1.0500000000000001e-06, |
| "loss": 1.1082134246826172, |
| "step": 3875 |
| }, |
| { |
| "epoch": 4.676258992805756, |
| "grad_norm": 2.8433194160461426, |
| "learning_rate": 8.416666666666667e-07, |
| "loss": 1.1347400665283203, |
| "step": 3900 |
| }, |
| { |
| "epoch": 4.706235011990408, |
| "grad_norm": 3.9949467182159424, |
| "learning_rate": 6.333333333333333e-07, |
| "loss": 1.2218107604980468, |
| "step": 3925 |
| }, |
| { |
| "epoch": 4.73621103117506, |
| "grad_norm": 3.4458863735198975, |
| "learning_rate": 4.25e-07, |
| "loss": 1.222669906616211, |
| "step": 3950 |
| }, |
| { |
| "epoch": 4.766187050359712, |
| "grad_norm": 3.780194044113159, |
| "learning_rate": 2.1666666666666667e-07, |
| "loss": 1.122889175415039, |
| "step": 3975 |
| }, |
| { |
| "epoch": 4.796163069544365, |
| "grad_norm": 2.996938705444336, |
| "learning_rate": 8.333333333333334e-09, |
| "loss": 1.305466766357422, |
| "step": 4000 |
| }, |
| { |
| "epoch": 4.796163069544365, |
| "eval_loss": 0.30047205090522766, |
| "eval_runtime": 17.9707, |
| "eval_samples_per_second": 90.147, |
| "eval_steps_per_second": 11.296, |
| "eval_wer": 37.73255972324371, |
| "step": 4000 |
| }, |
| { |
| "epoch": 4.796163069544365, |
| "step": 4000, |
| "total_flos": 1.186592111130968e+19, |
| "train_loss": 4.035597569465637, |
| "train_runtime": 1379.9989, |
| "train_samples_per_second": 46.377, |
| "train_steps_per_second": 2.899 |
| } |
| ], |
| "logging_steps": 25, |
| "max_steps": 4000, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 5, |
| "save_steps": 250, |
| "stateful_callbacks": { |
| "EarlyStoppingCallback": { |
| "args": { |
| "early_stopping_patience": 5, |
| "early_stopping_threshold": 0.1 |
| }, |
| "attributes": { |
| "early_stopping_patience_counter": 1 |
| } |
| }, |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1.186592111130968e+19, |
| "train_batch_size": 4, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
