Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
msp_visual
Generated from Trainer
custom_code
Instructions to use MahmoodAnaam/MSP-Visual with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use MahmoodAnaam/MSP-Visual with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="MahmoodAnaam/MSP-Visual", trust_remote_code=True)# Load model directly from transformers import AutoModelForCTC model = AutoModelForCTC.from_pretrained("MahmoodAnaam/MSP-Visual", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
| { | |
| "best_global_step": 19000, | |
| "best_metric": 0.6492743254292723, | |
| "best_model_checkpoint": "/workspace/checkpoints-av/msp_visual/checkpoint-19000", | |
| "epoch": 1.0, | |
| "eval_steps": 1000, | |
| "global_step": 20000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00125, | |
| "grad_norm": 1.4223047494888306, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 4.481825256347657, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0025, | |
| "grad_norm": 1.9665943384170532, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 4.5632717895507815, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.00375, | |
| "grad_norm": 2.63791561126709, | |
| "learning_rate": 7.4e-06, | |
| "loss": 4.601844482421875, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.005, | |
| "grad_norm": 4.429030418395996, | |
| "learning_rate": 9.900000000000002e-06, | |
| "loss": 4.152791137695313, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.00625, | |
| "grad_norm": 3.0334951877593994, | |
| "learning_rate": 1.24e-05, | |
| "loss": 3.6490899658203126, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.0075, | |
| "grad_norm": 0.9445637464523315, | |
| "learning_rate": 1.49e-05, | |
| "loss": 3.24427490234375, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.00875, | |
| "grad_norm": 0.9154287576675415, | |
| "learning_rate": 1.74e-05, | |
| "loss": 3.070023193359375, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 0.5437336564064026, | |
| "learning_rate": 1.9900000000000003e-05, | |
| "loss": 3.0271185302734374, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.01125, | |
| "grad_norm": 1.1024788618087769, | |
| "learning_rate": 2.2400000000000002e-05, | |
| "loss": 3.0147573852539065, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.0125, | |
| "grad_norm": 0.4438971281051636, | |
| "learning_rate": 2.4900000000000002e-05, | |
| "loss": 2.967065734863281, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.01375, | |
| "grad_norm": 0.31962060928344727, | |
| "learning_rate": 2.7400000000000002e-05, | |
| "loss": 3.003431396484375, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.015, | |
| "grad_norm": 1.9898868799209595, | |
| "learning_rate": 2.9900000000000002e-05, | |
| "loss": 2.934626159667969, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.01625, | |
| "grad_norm": 1.0482736825942993, | |
| "learning_rate": 3.24e-05, | |
| "loss": 2.9529678344726564, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.0175, | |
| "grad_norm": 0.3708650469779968, | |
| "learning_rate": 3.49e-05, | |
| "loss": 2.926605224609375, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.01875, | |
| "grad_norm": 0.44377943873405457, | |
| "learning_rate": 3.74e-05, | |
| "loss": 2.930478515625, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 0.772573709487915, | |
| "learning_rate": 3.99e-05, | |
| "loss": 2.91692138671875, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.02125, | |
| "grad_norm": 1.4369664192199707, | |
| "learning_rate": 4.24e-05, | |
| "loss": 2.897421875, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.0225, | |
| "grad_norm": 0.746988832950592, | |
| "learning_rate": 4.49e-05, | |
| "loss": 2.9280184936523437, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.02375, | |
| "grad_norm": 0.34846651554107666, | |
| "learning_rate": 4.74e-05, | |
| "loss": 2.891307373046875, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.025, | |
| "grad_norm": 1.0560742616653442, | |
| "learning_rate": 4.99e-05, | |
| "loss": 2.8670343017578124, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02625, | |
| "grad_norm": 0.8927328586578369, | |
| "learning_rate": 5.2400000000000007e-05, | |
| "loss": 2.8965029907226563, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.0275, | |
| "grad_norm": 0.2764662802219391, | |
| "learning_rate": 5.4900000000000006e-05, | |
| "loss": 2.8731961059570312, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.02875, | |
| "grad_norm": 0.5751842856407166, | |
| "learning_rate": 5.74e-05, | |
| "loss": 2.890608825683594, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 0.41129082441329956, | |
| "learning_rate": 5.99e-05, | |
| "loss": 2.850989990234375, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.03125, | |
| "grad_norm": 0.5647749304771423, | |
| "learning_rate": 6.24e-05, | |
| "loss": 2.8839126586914063, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.0325, | |
| "grad_norm": 1.0460444688796997, | |
| "learning_rate": 6.49e-05, | |
| "loss": 2.8625051879882815, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.03375, | |
| "grad_norm": 0.367440402507782, | |
| "learning_rate": 6.740000000000001e-05, | |
| "loss": 2.8707711791992185, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.035, | |
| "grad_norm": 0.836260199546814, | |
| "learning_rate": 6.99e-05, | |
| "loss": 2.860828552246094, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.03625, | |
| "grad_norm": 0.26464590430259705, | |
| "learning_rate": 7.24e-05, | |
| "loss": 2.846402893066406, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.0375, | |
| "grad_norm": 0.3564036190509796, | |
| "learning_rate": 7.49e-05, | |
| "loss": 2.8481539916992187, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.03875, | |
| "grad_norm": 0.36544445157051086, | |
| "learning_rate": 7.740000000000001e-05, | |
| "loss": 2.8175527954101565, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.39201128482818604, | |
| "learning_rate": 7.99e-05, | |
| "loss": 2.8255255126953127, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.04125, | |
| "grad_norm": 0.35684701800346375, | |
| "learning_rate": 8.24e-05, | |
| "loss": 2.788352355957031, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.0425, | |
| "grad_norm": 0.6995617151260376, | |
| "learning_rate": 8.49e-05, | |
| "loss": 2.7405670166015623, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.04375, | |
| "grad_norm": 1.3969351053237915, | |
| "learning_rate": 8.740000000000001e-05, | |
| "loss": 2.7369400024414063, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.045, | |
| "grad_norm": 0.5219077467918396, | |
| "learning_rate": 8.99e-05, | |
| "loss": 2.7241180419921873, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.04625, | |
| "grad_norm": 0.3834463059902191, | |
| "learning_rate": 9.240000000000001e-05, | |
| "loss": 2.6655810546875, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.0475, | |
| "grad_norm": 0.7631888389587402, | |
| "learning_rate": 9.49e-05, | |
| "loss": 2.6604400634765626, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.04875, | |
| "grad_norm": 0.6729345917701721, | |
| "learning_rate": 9.74e-05, | |
| "loss": 2.5972296142578126, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 0.6313336491584778, | |
| "learning_rate": 9.99e-05, | |
| "loss": 2.5770745849609376, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_cer": 0.61139556712141, | |
| "eval_loss": 2.3988261222839355, | |
| "eval_runtime": 217.8229, | |
| "eval_samples_per_second": 9.182, | |
| "eval_steps_per_second": 0.289, | |
| "eval_wer": 0.9791240801308259, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.05125, | |
| "grad_norm": 0.39277583360671997, | |
| "learning_rate": 9.999960630992558e-05, | |
| "loss": 2.48320556640625, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.0525, | |
| "grad_norm": 0.7017425894737244, | |
| "learning_rate": 9.999835894802353e-05, | |
| "loss": 2.412620391845703, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.05375, | |
| "grad_norm": 0.567896842956543, | |
| "learning_rate": 9.999625725211721e-05, | |
| "loss": 2.3854339599609373, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.055, | |
| "grad_norm": 0.705840528011322, | |
| "learning_rate": 9.999330125811884e-05, | |
| "loss": 2.3765003967285154, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.05625, | |
| "grad_norm": 0.6153581142425537, | |
| "learning_rate": 9.998949101653817e-05, | |
| "loss": 2.4039566040039064, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.0575, | |
| "grad_norm": 0.5512921810150146, | |
| "learning_rate": 9.998482659248174e-05, | |
| "loss": 2.3075006103515623, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.05875, | |
| "grad_norm": 1.0332051515579224, | |
| "learning_rate": 9.997930806565166e-05, | |
| "loss": 2.3186570739746095, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.5854579210281372, | |
| "learning_rate": 9.997293553034433e-05, | |
| "loss": 2.3392803955078123, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.06125, | |
| "grad_norm": 0.826861560344696, | |
| "learning_rate": 9.996570909544882e-05, | |
| "loss": 2.267376251220703, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.0625, | |
| "grad_norm": 0.6796795129776001, | |
| "learning_rate": 9.995762888444495e-05, | |
| "loss": 2.238070220947266, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.06375, | |
| "grad_norm": 0.6117750406265259, | |
| "learning_rate": 9.994869503540122e-05, | |
| "loss": 2.2643505859375, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.065, | |
| "grad_norm": 0.6967955231666565, | |
| "learning_rate": 9.993890770097247e-05, | |
| "loss": 2.159381866455078, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.06625, | |
| "grad_norm": 0.6792699694633484, | |
| "learning_rate": 9.992826704839727e-05, | |
| "loss": 2.2520883178710935, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.0675, | |
| "grad_norm": 0.7921984195709229, | |
| "learning_rate": 9.991677325949497e-05, | |
| "loss": 2.2055706787109375, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.06875, | |
| "grad_norm": 0.7900427579879761, | |
| "learning_rate": 9.990442653066272e-05, | |
| "loss": 2.1528033447265624, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 0.8837091326713562, | |
| "learning_rate": 9.989122707287208e-05, | |
| "loss": 2.163330535888672, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.07125, | |
| "grad_norm": 0.7834476232528687, | |
| "learning_rate": 9.987717511166533e-05, | |
| "loss": 2.105241394042969, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.0725, | |
| "grad_norm": 0.7125497460365295, | |
| "learning_rate": 9.986227088715173e-05, | |
| "loss": 2.136973876953125, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.07375, | |
| "grad_norm": 0.7752950191497803, | |
| "learning_rate": 9.984651465400335e-05, | |
| "loss": 2.1693585205078123, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.075, | |
| "grad_norm": 0.7111746072769165, | |
| "learning_rate": 9.982990668145075e-05, | |
| "loss": 2.1592276000976565, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.07625, | |
| "grad_norm": 0.9812447428703308, | |
| "learning_rate": 9.981244725327836e-05, | |
| "loss": 2.1382072448730467, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.0775, | |
| "grad_norm": 0.8474680781364441, | |
| "learning_rate": 9.979413666781963e-05, | |
| "loss": 2.1538070678710937, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.07875, | |
| "grad_norm": 0.7857120037078857, | |
| "learning_rate": 9.977497523795194e-05, | |
| "loss": 2.1174755859375, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.8059459328651428, | |
| "learning_rate": 9.975496329109126e-05, | |
| "loss": 2.100527801513672, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.08125, | |
| "grad_norm": 0.6679235696792603, | |
| "learning_rate": 9.973410116918653e-05, | |
| "loss": 2.191019287109375, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.0825, | |
| "grad_norm": 0.5935495495796204, | |
| "learning_rate": 9.971238922871391e-05, | |
| "loss": 2.0978602600097656, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.08375, | |
| "grad_norm": 0.8274971842765808, | |
| "learning_rate": 9.968982784067049e-05, | |
| "loss": 2.0475253295898437, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.085, | |
| "grad_norm": 0.8631715178489685, | |
| "learning_rate": 9.966641739056818e-05, | |
| "loss": 2.0770193481445314, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.08625, | |
| "grad_norm": 0.8427679538726807, | |
| "learning_rate": 9.964215827842698e-05, | |
| "loss": 2.0830838012695314, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.0875, | |
| "grad_norm": 0.9500517845153809, | |
| "learning_rate": 9.961705091876816e-05, | |
| "loss": 2.090618133544922, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.08875, | |
| "grad_norm": 0.9094032645225525, | |
| "learning_rate": 9.959109574060726e-05, | |
| "loss": 2.0686489868164064, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.8317903876304626, | |
| "learning_rate": 9.956429318744662e-05, | |
| "loss": 2.0597463989257814, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.09125, | |
| "grad_norm": 0.8647030591964722, | |
| "learning_rate": 9.953664371726797e-05, | |
| "loss": 2.0847763061523437, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.0925, | |
| "grad_norm": 0.8953089714050293, | |
| "learning_rate": 9.950814780252442e-05, | |
| "loss": 2.013682556152344, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.09375, | |
| "grad_norm": 0.8105475902557373, | |
| "learning_rate": 9.947880593013255e-05, | |
| "loss": 1.9783633422851563, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.095, | |
| "grad_norm": 0.842832624912262, | |
| "learning_rate": 9.944861860146401e-05, | |
| "loss": 1.9547543334960937, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.09625, | |
| "grad_norm": 0.8718632459640503, | |
| "learning_rate": 9.941758633233696e-05, | |
| "loss": 2.015556640625, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.0975, | |
| "grad_norm": 0.8931266665458679, | |
| "learning_rate": 9.938570965300724e-05, | |
| "loss": 2.1189161682128907, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.09875, | |
| "grad_norm": 0.7824097275733948, | |
| "learning_rate": 9.935298910815936e-05, | |
| "loss": 2.0638720703125, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.9027556777000427, | |
| "learning_rate": 9.931942525689715e-05, | |
| "loss": 2.0986813354492186, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_cer": 0.47856812955964223, | |
| "eval_loss": 1.777026653289795, | |
| "eval_runtime": 221.039, | |
| "eval_samples_per_second": 9.048, | |
| "eval_steps_per_second": 0.285, | |
| "eval_wer": 0.8288276778413737, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.10125, | |
| "grad_norm": 0.8824305534362793, | |
| "learning_rate": 9.928501867273423e-05, | |
| "loss": 2.0886192321777344, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.1025, | |
| "grad_norm": 1.1464126110076904, | |
| "learning_rate": 9.924976994358417e-05, | |
| "loss": 2.1167315673828124, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.10375, | |
| "grad_norm": 0.6385474801063538, | |
| "learning_rate": 9.921367967175052e-05, | |
| "loss": 1.996148681640625, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.105, | |
| "grad_norm": 1.0066859722137451, | |
| "learning_rate": 9.917674847391645e-05, | |
| "loss": 2.04283935546875, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.10625, | |
| "grad_norm": 0.9596987962722778, | |
| "learning_rate": 9.913897698113422e-05, | |
| "loss": 2.050751495361328, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.1075, | |
| "grad_norm": 1.316969871520996, | |
| "learning_rate": 9.910036583881443e-05, | |
| "loss": 2.080152893066406, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.10875, | |
| "grad_norm": 0.8739728331565857, | |
| "learning_rate": 9.906091570671497e-05, | |
| "loss": 2.0330570983886718, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 1.230444073677063, | |
| "learning_rate": 9.902062725892976e-05, | |
| "loss": 1.9648872375488282, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.11125, | |
| "grad_norm": 0.6308285593986511, | |
| "learning_rate": 9.89795011838772e-05, | |
| "loss": 1.9692529296875, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.1125, | |
| "grad_norm": 0.928484320640564, | |
| "learning_rate": 9.893753818428845e-05, | |
| "loss": 1.9276266479492188, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.11375, | |
| "grad_norm": 0.8584057688713074, | |
| "learning_rate": 9.889473897719539e-05, | |
| "loss": 1.9632791137695313, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.115, | |
| "grad_norm": 0.8964471817016602, | |
| "learning_rate": 9.88511042939184e-05, | |
| "loss": 1.905766143798828, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.11625, | |
| "grad_norm": 0.9695539474487305, | |
| "learning_rate": 9.880663488005379e-05, | |
| "loss": 1.8976968383789063, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.1175, | |
| "grad_norm": 0.880225419998169, | |
| "learning_rate": 9.876133149546118e-05, | |
| "loss": 2.0029702758789063, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.11875, | |
| "grad_norm": 0.8994690179824829, | |
| "learning_rate": 9.871519491425038e-05, | |
| "loss": 2.0456982421875, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 1.4965325593948364, | |
| "learning_rate": 9.866822592476833e-05, | |
| "loss": 2.0431723022460937, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.12125, | |
| "grad_norm": 0.7884564995765686, | |
| "learning_rate": 9.86204253295854e-05, | |
| "loss": 1.9890243530273437, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.1225, | |
| "grad_norm": 1.0217492580413818, | |
| "learning_rate": 9.857179394548191e-05, | |
| "loss": 1.9927993774414063, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.12375, | |
| "grad_norm": 0.8048027753829956, | |
| "learning_rate": 9.852233260343403e-05, | |
| "loss": 1.987839813232422, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.125, | |
| "grad_norm": 0.9624154567718506, | |
| "learning_rate": 9.847204214859964e-05, | |
| "loss": 2.0316046142578124, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.12625, | |
| "grad_norm": 0.9308169484138489, | |
| "learning_rate": 9.842092344030382e-05, | |
| "loss": 2.0430287170410155, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.1275, | |
| "grad_norm": 1.0451279878616333, | |
| "learning_rate": 9.83689773520243e-05, | |
| "loss": 1.9393511962890626, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.12875, | |
| "grad_norm": 0.9306155443191528, | |
| "learning_rate": 9.831620477137638e-05, | |
| "loss": 1.9553242492675782, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.9762840270996094, | |
| "learning_rate": 9.826260660009785e-05, | |
| "loss": 1.9781422424316406, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.13125, | |
| "grad_norm": 1.1196411848068237, | |
| "learning_rate": 9.820818375403357e-05, | |
| "loss": 1.9470297241210937, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.1325, | |
| "grad_norm": 1.0263965129852295, | |
| "learning_rate": 9.815293716311987e-05, | |
| "loss": 1.93536865234375, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.13375, | |
| "grad_norm": 1.0180565118789673, | |
| "learning_rate": 9.809686777136852e-05, | |
| "loss": 1.9217897033691407, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.135, | |
| "grad_norm": 1.2505632638931274, | |
| "learning_rate": 9.803997653685072e-05, | |
| "loss": 1.9673397827148438, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.13625, | |
| "grad_norm": 0.6409063339233398, | |
| "learning_rate": 9.798226443168072e-05, | |
| "loss": 1.9945646667480468, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.1375, | |
| "grad_norm": 0.9332085251808167, | |
| "learning_rate": 9.792373244199913e-05, | |
| "loss": 1.9790260314941406, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.13875, | |
| "grad_norm": 0.6687518954277039, | |
| "learning_rate": 9.786438156795621e-05, | |
| "loss": 1.9665324401855468, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.8691813349723816, | |
| "learning_rate": 9.780421282369461e-05, | |
| "loss": 1.9200575256347656, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.14125, | |
| "grad_norm": 0.9970143437385559, | |
| "learning_rate": 9.774322723733216e-05, | |
| "loss": 1.9706582641601562, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.1425, | |
| "grad_norm": 0.782715380191803, | |
| "learning_rate": 9.768142585094426e-05, | |
| "loss": 1.9335671997070312, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.14375, | |
| "grad_norm": 1.1588191986083984, | |
| "learning_rate": 9.76188097205461e-05, | |
| "loss": 1.9765562438964843, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.145, | |
| "grad_norm": 0.9819175601005554, | |
| "learning_rate": 9.755537991607459e-05, | |
| "loss": 1.9710671997070313, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.14625, | |
| "grad_norm": 1.0242927074432373, | |
| "learning_rate": 9.749113752137007e-05, | |
| "loss": 2.0104330444335936, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.1475, | |
| "grad_norm": 0.8246551156044006, | |
| "learning_rate": 9.742608363415781e-05, | |
| "loss": 1.987698974609375, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.14875, | |
| "grad_norm": 0.9302476048469543, | |
| "learning_rate": 9.736021936602926e-05, | |
| "loss": 2.062389068603516, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 0.8409685492515564, | |
| "learning_rate": 9.729354584242302e-05, | |
| "loss": 1.9764314270019532, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_cer": 0.4485371897953407, | |
| "eval_loss": 1.6458523273468018, | |
| "eval_runtime": 214.2648, | |
| "eval_samples_per_second": 9.334, | |
| "eval_steps_per_second": 0.294, | |
| "eval_wer": 0.7827064595257563, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.15125, | |
| "grad_norm": 0.9748871326446533, | |
| "learning_rate": 9.72260642026057e-05, | |
| "loss": 2.041459197998047, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.1525, | |
| "grad_norm": 1.1625964641571045, | |
| "learning_rate": 9.715777559965228e-05, | |
| "loss": 2.011282501220703, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.15375, | |
| "grad_norm": 0.8373090624809265, | |
| "learning_rate": 9.70886812004266e-05, | |
| "loss": 2.059709930419922, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.155, | |
| "grad_norm": 0.8791937828063965, | |
| "learning_rate": 9.701878218556129e-05, | |
| "loss": 2.119442901611328, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.15625, | |
| "grad_norm": 0.8728455901145935, | |
| "learning_rate": 9.694807974943767e-05, | |
| "loss": 2.0219273376464844, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.1575, | |
| "grad_norm": 0.8951705694198608, | |
| "learning_rate": 9.687657510016527e-05, | |
| "loss": 2.020895538330078, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.15875, | |
| "grad_norm": 0.889247477054596, | |
| "learning_rate": 9.680426945956129e-05, | |
| "loss": 2.0021768188476563, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 1.3758327960968018, | |
| "learning_rate": 9.673116406312962e-05, | |
| "loss": 2.032898254394531, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.16125, | |
| "grad_norm": 1.0335829257965088, | |
| "learning_rate": 9.665726016003977e-05, | |
| "loss": 1.9508203125, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.1625, | |
| "grad_norm": 0.774730920791626, | |
| "learning_rate": 9.658255901310557e-05, | |
| "loss": 1.9228631591796874, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.16375, | |
| "grad_norm": 1.1470255851745605, | |
| "learning_rate": 9.650706189876346e-05, | |
| "loss": 1.9922433471679688, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.165, | |
| "grad_norm": 0.9730787873268127, | |
| "learning_rate": 9.643077010705087e-05, | |
| "loss": 2.0077645874023435, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.16625, | |
| "grad_norm": 1.0259824991226196, | |
| "learning_rate": 9.6353684941584e-05, | |
| "loss": 1.9907278442382812, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.1675, | |
| "grad_norm": 0.7878318428993225, | |
| "learning_rate": 9.627580771953563e-05, | |
| "loss": 1.9990422058105468, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.16875, | |
| "grad_norm": 0.9537973403930664, | |
| "learning_rate": 9.619713977161265e-05, | |
| "loss": 2.0272256469726564, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 1.0746439695358276, | |
| "learning_rate": 9.611768244203321e-05, | |
| "loss": 2.0527517700195315, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.17125, | |
| "grad_norm": 1.0745329856872559, | |
| "learning_rate": 9.603743708850386e-05, | |
| "loss": 1.9901165771484375, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.1725, | |
| "grad_norm": 0.8954412341117859, | |
| "learning_rate": 9.595640508219625e-05, | |
| "loss": 2.0291989135742186, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.17375, | |
| "grad_norm": 1.1868406534194946, | |
| "learning_rate": 9.587458780772385e-05, | |
| "loss": 2.003343505859375, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.175, | |
| "grad_norm": 0.8366594314575195, | |
| "learning_rate": 9.579198666311809e-05, | |
| "loss": 2.0220095825195314, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.17625, | |
| "grad_norm": 1.0375490188598633, | |
| "learning_rate": 9.570860305980466e-05, | |
| "loss": 1.8457562255859374, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.1775, | |
| "grad_norm": 1.0875084400177002, | |
| "learning_rate": 9.562443842257925e-05, | |
| "loss": 1.809892120361328, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.17875, | |
| "grad_norm": 1.2091565132141113, | |
| "learning_rate": 9.553949418958332e-05, | |
| "loss": 1.9217204284667968, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 1.1248983144760132, | |
| "learning_rate": 9.545377181227942e-05, | |
| "loss": 1.9127857971191407, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.18125, | |
| "grad_norm": 0.8328970670700073, | |
| "learning_rate": 9.536727275542648e-05, | |
| "loss": 1.97947509765625, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.1825, | |
| "grad_norm": 0.9299238920211792, | |
| "learning_rate": 9.527999849705471e-05, | |
| "loss": 2.0013667297363282, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.18375, | |
| "grad_norm": 1.0658174753189087, | |
| "learning_rate": 9.519195052844039e-05, | |
| "loss": 2.038532409667969, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.185, | |
| "grad_norm": 0.8259148001670837, | |
| "learning_rate": 9.510313035408035e-05, | |
| "loss": 1.9441246032714843, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.18625, | |
| "grad_norm": 1.071619987487793, | |
| "learning_rate": 9.501353949166637e-05, | |
| "loss": 2.0016786193847658, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.1875, | |
| "grad_norm": 1.2241653203964233, | |
| "learning_rate": 9.492317947205904e-05, | |
| "loss": 2.043337097167969, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.18875, | |
| "grad_norm": 0.8591672778129578, | |
| "learning_rate": 9.483205183926181e-05, | |
| "loss": 1.9712590026855468, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 1.109619140625, | |
| "learning_rate": 9.474015815039446e-05, | |
| "loss": 1.93697265625, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.19125, | |
| "grad_norm": 1.2231351137161255, | |
| "learning_rate": 9.464749997566664e-05, | |
| "loss": 1.902823486328125, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.1925, | |
| "grad_norm": 1.219138503074646, | |
| "learning_rate": 9.455407889835087e-05, | |
| "loss": 1.9404554748535157, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.19375, | |
| "grad_norm": 1.0045181512832642, | |
| "learning_rate": 9.44598965147556e-05, | |
| "loss": 1.9708486938476562, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.195, | |
| "grad_norm": 0.8377895951271057, | |
| "learning_rate": 9.436495443419795e-05, | |
| "loss": 1.9305537414550782, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.19625, | |
| "grad_norm": 0.7125170230865479, | |
| "learning_rate": 9.426925427897608e-05, | |
| "loss": 1.9963351440429689, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.1975, | |
| "grad_norm": 1.2973546981811523, | |
| "learning_rate": 9.41727976843416e-05, | |
| "loss": 1.9806707763671876, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.19875, | |
| "grad_norm": 1.7796871662139893, | |
| "learning_rate": 9.40755862984716e-05, | |
| "loss": 1.931557159423828, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.7939216494560242, | |
| "learning_rate": 9.397762178244043e-05, | |
| "loss": 1.9493400573730468, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_cer": 0.4387586350746676, | |
| "eval_loss": 1.6138834953308105, | |
| "eval_runtime": 195.2358, | |
| "eval_samples_per_second": 10.244, | |
| "eval_steps_per_second": 0.323, | |
| "eval_wer": 0.7613961569910057, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.20125, | |
| "grad_norm": 0.8293856978416443, | |
| "learning_rate": 9.387890581019139e-05, | |
| "loss": 1.9017347717285156, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.2025, | |
| "grad_norm": 1.160848617553711, | |
| "learning_rate": 9.377944006850807e-05, | |
| "loss": 1.9506214904785155, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.20375, | |
| "grad_norm": 1.8133563995361328, | |
| "learning_rate": 9.367922625698558e-05, | |
| "loss": 1.8418312072753906, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.205, | |
| "grad_norm": 1.0395394563674927, | |
| "learning_rate": 9.357826608800142e-05, | |
| "loss": 1.9256382751464844, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.20625, | |
| "grad_norm": 1.2605054378509521, | |
| "learning_rate": 9.347656128668636e-05, | |
| "loss": 1.9289976501464843, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.2075, | |
| "grad_norm": 1.1306040287017822, | |
| "learning_rate": 9.33741135908948e-05, | |
| "loss": 1.8459049987792968, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.20875, | |
| "grad_norm": 0.9057150483131409, | |
| "learning_rate": 9.327092475117525e-05, | |
| "loss": 1.870419921875, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 1.3892112970352173, | |
| "learning_rate": 9.316699653074023e-05, | |
| "loss": 1.9207023620605468, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.21125, | |
| "grad_norm": 1.0148946046829224, | |
| "learning_rate": 9.306233070543631e-05, | |
| "loss": 1.902631072998047, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.2125, | |
| "grad_norm": 1.286488652229309, | |
| "learning_rate": 9.295692906371363e-05, | |
| "loss": 1.9383331298828126, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.21375, | |
| "grad_norm": 1.4743224382400513, | |
| "learning_rate": 9.285079340659545e-05, | |
| "loss": 1.895093231201172, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.215, | |
| "grad_norm": 1.1783928871154785, | |
| "learning_rate": 9.274392554764733e-05, | |
| "loss": 1.9223080444335938, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.21625, | |
| "grad_norm": 1.0104583501815796, | |
| "learning_rate": 9.263632731294609e-05, | |
| "loss": 1.926556396484375, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.2175, | |
| "grad_norm": 0.8637489676475525, | |
| "learning_rate": 9.252800054104868e-05, | |
| "loss": 1.9078114318847657, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.21875, | |
| "grad_norm": 1.217969536781311, | |
| "learning_rate": 9.24189470829608e-05, | |
| "loss": 1.9689820861816407, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 1.0189920663833618, | |
| "learning_rate": 9.230916880210512e-05, | |
| "loss": 1.9385252380371094, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.22125, | |
| "grad_norm": 0.9831681847572327, | |
| "learning_rate": 9.219866757428959e-05, | |
| "loss": 1.9424336242675782, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.2225, | |
| "grad_norm": 0.910565197467804, | |
| "learning_rate": 9.208744528767528e-05, | |
| "loss": 1.9493093872070313, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.22375, | |
| "grad_norm": 1.1580244302749634, | |
| "learning_rate": 9.197550384274423e-05, | |
| "loss": 1.9484165954589843, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.225, | |
| "grad_norm": 0.874718427658081, | |
| "learning_rate": 9.186284515226686e-05, | |
| "loss": 1.9203952026367188, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.22625, | |
| "grad_norm": 1.0136849880218506, | |
| "learning_rate": 9.174947114126935e-05, | |
| "loss": 1.9253334045410155, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.2275, | |
| "grad_norm": 1.2775322198867798, | |
| "learning_rate": 9.163538374700076e-05, | |
| "loss": 1.9003428649902343, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.22875, | |
| "grad_norm": 0.958245038986206, | |
| "learning_rate": 9.152058491889986e-05, | |
| "loss": 1.852706298828125, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 0.9219448566436768, | |
| "learning_rate": 9.140507661856187e-05, | |
| "loss": 1.866653289794922, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.23125, | |
| "grad_norm": 0.8240512013435364, | |
| "learning_rate": 9.128886081970498e-05, | |
| "loss": 1.8764305114746094, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.2325, | |
| "grad_norm": 1.4956380128860474, | |
| "learning_rate": 9.117193950813652e-05, | |
| "loss": 1.8642387390136719, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.23375, | |
| "grad_norm": 0.7645339965820312, | |
| "learning_rate": 9.10543146817191e-05, | |
| "loss": 1.9245855712890625, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.235, | |
| "grad_norm": 1.2815724611282349, | |
| "learning_rate": 9.093598835033649e-05, | |
| "loss": 1.872388153076172, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.23625, | |
| "grad_norm": 0.7426547408103943, | |
| "learning_rate": 9.081696253585921e-05, | |
| "loss": 1.968475341796875, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.2375, | |
| "grad_norm": 0.7444338798522949, | |
| "learning_rate": 9.069723927211001e-05, | |
| "loss": 1.9433749389648438, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.23875, | |
| "grad_norm": 0.8147239089012146, | |
| "learning_rate": 9.057682060482916e-05, | |
| "loss": 1.9384831237792968, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.8725531697273254, | |
| "learning_rate": 9.045570859163943e-05, | |
| "loss": 1.9606639099121095, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.24125, | |
| "grad_norm": 0.832190215587616, | |
| "learning_rate": 9.033390530201097e-05, | |
| "loss": 1.9929269409179688, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.2425, | |
| "grad_norm": 1.0074503421783447, | |
| "learning_rate": 9.021141281722591e-05, | |
| "loss": 1.933099365234375, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.24375, | |
| "grad_norm": 0.777546226978302, | |
| "learning_rate": 9.008823323034288e-05, | |
| "loss": 1.9763279724121094, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.245, | |
| "grad_norm": 0.895451009273529, | |
| "learning_rate": 8.996436864616116e-05, | |
| "loss": 1.9091413879394532, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.24625, | |
| "grad_norm": 0.8923606872558594, | |
| "learning_rate": 8.98398211811847e-05, | |
| "loss": 2.0167835998535155, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.2475, | |
| "grad_norm": 0.9579914808273315, | |
| "learning_rate": 8.971459296358606e-05, | |
| "loss": 1.9921633911132812, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.24875, | |
| "grad_norm": 6.371490001678467, | |
| "learning_rate": 8.958868613316996e-05, | |
| "loss": 1.9475032043457032, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 1.0517321825027466, | |
| "learning_rate": 8.946210284133676e-05, | |
| "loss": 1.9069137573242188, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "eval_cer": 0.41909226803867716, | |
| "eval_loss": 1.5498154163360596, | |
| "eval_runtime": 217.6843, | |
| "eval_samples_per_second": 9.188, | |
| "eval_steps_per_second": 0.289, | |
| "eval_wer": 0.7350776778413737, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.25125, | |
| "grad_norm": 1.2655832767486572, | |
| "learning_rate": 8.933484525104562e-05, | |
| "loss": 1.9912879943847657, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 0.2525, | |
| "grad_norm": 1.2533960342407227, | |
| "learning_rate": 8.92069155367777e-05, | |
| "loss": 1.9497242736816407, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.25375, | |
| "grad_norm": 0.8946574330329895, | |
| "learning_rate": 8.907831588449879e-05, | |
| "loss": 1.882041473388672, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 0.255, | |
| "grad_norm": 0.846342921257019, | |
| "learning_rate": 8.894904849162218e-05, | |
| "loss": 1.9769656372070312, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.25625, | |
| "grad_norm": 0.966134250164032, | |
| "learning_rate": 8.881911556697093e-05, | |
| "loss": 2.048040008544922, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 0.2575, | |
| "grad_norm": 1.1368507146835327, | |
| "learning_rate": 8.868851933074021e-05, | |
| "loss": 1.9452519226074219, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.25875, | |
| "grad_norm": 0.8084460496902466, | |
| "learning_rate": 8.855726201445938e-05, | |
| "loss": 1.9749110412597657, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 0.9588484168052673, | |
| "learning_rate": 8.842534586095383e-05, | |
| "loss": 1.9269889831542968, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.26125, | |
| "grad_norm": 0.827880322933197, | |
| "learning_rate": 8.829277312430665e-05, | |
| "loss": 1.8800985717773437, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 0.2625, | |
| "grad_norm": 1.0797176361083984, | |
| "learning_rate": 8.815954606982015e-05, | |
| "loss": 2.028725433349609, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.26375, | |
| "grad_norm": 0.8844195604324341, | |
| "learning_rate": 8.802566697397708e-05, | |
| "loss": 2.014178924560547, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 0.265, | |
| "grad_norm": 0.8954948782920837, | |
| "learning_rate": 8.78911381244018e-05, | |
| "loss": 1.8876654052734374, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.26625, | |
| "grad_norm": 0.8572631478309631, | |
| "learning_rate": 8.775596181982118e-05, | |
| "loss": 1.9149148559570313, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 0.2675, | |
| "grad_norm": 0.7878565788269043, | |
| "learning_rate": 8.76201403700253e-05, | |
| "loss": 1.899776153564453, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.26875, | |
| "grad_norm": 0.9162899255752563, | |
| "learning_rate": 8.748367609582801e-05, | |
| "loss": 1.8620721435546874, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 0.7461763620376587, | |
| "learning_rate": 8.73465713290272e-05, | |
| "loss": 1.9585078430175782, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.27125, | |
| "grad_norm": 0.6477616429328918, | |
| "learning_rate": 8.720882841236508e-05, | |
| "loss": 1.9927880859375, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 0.2725, | |
| "grad_norm": 0.8808125257492065, | |
| "learning_rate": 8.707044969948806e-05, | |
| "loss": 1.8197854614257813, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.27375, | |
| "grad_norm": 0.8587700724601746, | |
| "learning_rate": 8.693143755490652e-05, | |
| "loss": 1.8663734436035155, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 0.275, | |
| "grad_norm": 0.7857425808906555, | |
| "learning_rate": 8.679179435395446e-05, | |
| "loss": 1.9393270874023438, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.27625, | |
| "grad_norm": 0.9678592681884766, | |
| "learning_rate": 8.66515224827489e-05, | |
| "loss": 1.8171139526367188, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 0.2775, | |
| "grad_norm": 0.6875806450843811, | |
| "learning_rate": 8.651062433814912e-05, | |
| "loss": 1.9486782836914063, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.27875, | |
| "grad_norm": 0.8405512571334839, | |
| "learning_rate": 8.636910232771562e-05, | |
| "loss": 1.9151707458496094, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 0.8750268220901489, | |
| "learning_rate": 8.622695886966911e-05, | |
| "loss": 1.9108184814453124, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.28125, | |
| "grad_norm": 1.0632688999176025, | |
| "learning_rate": 8.60841963928491e-05, | |
| "loss": 1.9389430236816407, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 0.2825, | |
| "grad_norm": 0.7698346376419067, | |
| "learning_rate": 8.594081733667243e-05, | |
| "loss": 1.8848039245605468, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.28375, | |
| "grad_norm": 1.2021455764770508, | |
| "learning_rate": 8.579682415109156e-05, | |
| "loss": 1.875746612548828, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 0.285, | |
| "grad_norm": 0.8804958462715149, | |
| "learning_rate": 8.565221929655275e-05, | |
| "loss": 1.8698390197753907, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.28625, | |
| "grad_norm": 0.9940733313560486, | |
| "learning_rate": 8.550700524395397e-05, | |
| "loss": 1.9087271118164062, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 0.2875, | |
| "grad_norm": 1.1654746532440186, | |
| "learning_rate": 8.536118447460275e-05, | |
| "loss": 1.9824313354492187, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.28875, | |
| "grad_norm": 1.1857008934020996, | |
| "learning_rate": 8.521475948017366e-05, | |
| "loss": 1.9514483642578124, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 1.0095324516296387, | |
| "learning_rate": 8.506773276266588e-05, | |
| "loss": 1.918441619873047, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.29125, | |
| "grad_norm": 0.7452509999275208, | |
| "learning_rate": 8.492010683436038e-05, | |
| "loss": 1.9532533264160157, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 0.2925, | |
| "grad_norm": 1.0081002712249756, | |
| "learning_rate": 8.477188421777692e-05, | |
| "loss": 1.97687255859375, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 0.29375, | |
| "grad_norm": 0.9298150539398193, | |
| "learning_rate": 8.462306744563108e-05, | |
| "loss": 1.8111131286621094, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 0.295, | |
| "grad_norm": 0.9799827337265015, | |
| "learning_rate": 8.447365906079088e-05, | |
| "loss": 1.8500839233398438, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.29625, | |
| "grad_norm": 0.8726683855056763, | |
| "learning_rate": 8.432366161623337e-05, | |
| "loss": 1.8856776428222657, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 0.2975, | |
| "grad_norm": 0.796048641204834, | |
| "learning_rate": 8.417307767500107e-05, | |
| "loss": 1.8410433959960937, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 0.29875, | |
| "grad_norm": 0.9114080667495728, | |
| "learning_rate": 8.402190981015803e-05, | |
| "loss": 1.8724212646484375, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 0.9777677059173584, | |
| "learning_rate": 8.387016060474597e-05, | |
| "loss": 1.9144281005859376, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "eval_cer": 0.4141856087882836, | |
| "eval_loss": 1.5211948156356812, | |
| "eval_runtime": 220.9481, | |
| "eval_samples_per_second": 9.052, | |
| "eval_steps_per_second": 0.285, | |
| "eval_wer": 0.7211774325429272, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.30125, | |
| "grad_norm": 0.9911900758743286, | |
| "learning_rate": 8.371783265174013e-05, | |
| "loss": 1.8932913208007813, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 0.3025, | |
| "grad_norm": 1.15787935256958, | |
| "learning_rate": 8.356492855400493e-05, | |
| "loss": 1.9449674987792969, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 0.30375, | |
| "grad_norm": 0.906575083732605, | |
| "learning_rate": 8.341145092424954e-05, | |
| "loss": 1.9541030883789063, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 0.305, | |
| "grad_norm": 0.7480375170707703, | |
| "learning_rate": 8.325740238498317e-05, | |
| "loss": 1.904345245361328, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.30625, | |
| "grad_norm": 0.8597415685653687, | |
| "learning_rate": 8.310278556847036e-05, | |
| "loss": 1.9497067260742187, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 0.3075, | |
| "grad_norm": 0.8572204113006592, | |
| "learning_rate": 8.294760311668586e-05, | |
| "loss": 1.926986541748047, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 0.30875, | |
| "grad_norm": 0.7788655757904053, | |
| "learning_rate": 8.279185768126962e-05, | |
| "loss": 1.898881072998047, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 0.9704040884971619, | |
| "learning_rate": 8.263555192348143e-05, | |
| "loss": 1.9204402160644531, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.31125, | |
| "grad_norm": 0.9288792610168457, | |
| "learning_rate": 8.247868851415542e-05, | |
| "loss": 1.956895751953125, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 0.3125, | |
| "grad_norm": 0.8298304080963135, | |
| "learning_rate": 8.232127013365445e-05, | |
| "loss": 1.7825811767578126, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.31375, | |
| "grad_norm": 0.8955572843551636, | |
| "learning_rate": 8.216329947182431e-05, | |
| "loss": 1.9356716918945311, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 0.315, | |
| "grad_norm": 0.9540830850601196, | |
| "learning_rate": 8.200477922794776e-05, | |
| "loss": 1.8685700988769531, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.31625, | |
| "grad_norm": 0.8996971845626831, | |
| "learning_rate": 8.18457121106984e-05, | |
| "loss": 1.8638067626953125, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 0.3175, | |
| "grad_norm": 0.8179593682289124, | |
| "learning_rate": 8.168610083809438e-05, | |
| "loss": 1.8482196044921875, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 0.31875, | |
| "grad_norm": 1.064643144607544, | |
| "learning_rate": 8.152594813745196e-05, | |
| "loss": 1.832763671875, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 0.9226500988006592, | |
| "learning_rate": 8.13652567453389e-05, | |
| "loss": 1.9133285522460937, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.32125, | |
| "grad_norm": 0.9603607654571533, | |
| "learning_rate": 8.120402940752778e-05, | |
| "loss": 1.8548468017578126, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 0.3225, | |
| "grad_norm": 1.0315053462982178, | |
| "learning_rate": 8.104226887894892e-05, | |
| "loss": 1.8654867553710937, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 0.32375, | |
| "grad_norm": 0.9519103169441223, | |
| "learning_rate": 8.087997792364344e-05, | |
| "loss": 1.82350830078125, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 0.325, | |
| "grad_norm": 0.87391597032547, | |
| "learning_rate": 8.071715931471602e-05, | |
| "loss": 1.927668914794922, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.32625, | |
| "grad_norm": 0.8425279855728149, | |
| "learning_rate": 8.055381583428743e-05, | |
| "loss": 1.9289253234863282, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 0.3275, | |
| "grad_norm": 0.9698518514633179, | |
| "learning_rate": 8.03899502734471e-05, | |
| "loss": 1.9301071166992188, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 0.32875, | |
| "grad_norm": 0.820851743221283, | |
| "learning_rate": 8.022556543220529e-05, | |
| "loss": 1.9006959533691405, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 0.8217940330505371, | |
| "learning_rate": 8.006066411944542e-05, | |
| "loss": 1.9452406311035155, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.33125, | |
| "grad_norm": 0.747125506401062, | |
| "learning_rate": 7.989524915287595e-05, | |
| "loss": 1.9340284729003907, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 0.3325, | |
| "grad_norm": 0.8333204984664917, | |
| "learning_rate": 7.972932335898226e-05, | |
| "loss": 1.8868572998046875, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 0.33375, | |
| "grad_norm": 1.0247576236724854, | |
| "learning_rate": 7.956288957297834e-05, | |
| "loss": 1.9146258544921875, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 0.335, | |
| "grad_norm": 1.011094331741333, | |
| "learning_rate": 7.939595063875842e-05, | |
| "loss": 1.8421237182617187, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.33625, | |
| "grad_norm": 1.0771154165267944, | |
| "learning_rate": 7.922850940884827e-05, | |
| "loss": 1.872168731689453, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 0.3375, | |
| "grad_norm": 0.9887160062789917, | |
| "learning_rate": 7.906056874435652e-05, | |
| "loss": 1.8505189514160156, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 0.33875, | |
| "grad_norm": 0.9073835611343384, | |
| "learning_rate": 7.889213151492578e-05, | |
| "loss": 1.8202755737304688, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.8542730808258057, | |
| "learning_rate": 7.872320059868355e-05, | |
| "loss": 1.89970703125, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.34125, | |
| "grad_norm": 1.0605671405792236, | |
| "learning_rate": 7.855377888219307e-05, | |
| "loss": 1.87497802734375, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 0.3425, | |
| "grad_norm": 2.405791997909546, | |
| "learning_rate": 7.838386926040407e-05, | |
| "loss": 1.9150254821777344, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 0.34375, | |
| "grad_norm": 0.7388380765914917, | |
| "learning_rate": 7.821347463660315e-05, | |
| "loss": 1.9328706359863281, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 0.345, | |
| "grad_norm": 0.8893718123435974, | |
| "learning_rate": 7.804259792236435e-05, | |
| "loss": 1.88892822265625, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.34625, | |
| "grad_norm": 0.847213089466095, | |
| "learning_rate": 7.78712420374992e-05, | |
| "loss": 1.9452713012695313, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 0.3475, | |
| "grad_norm": 0.7064250707626343, | |
| "learning_rate": 7.7699409910007e-05, | |
| "loss": 1.7696400451660157, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 0.34875, | |
| "grad_norm": 1.099586009979248, | |
| "learning_rate": 7.75271044760247e-05, | |
| "loss": 1.8268797302246094, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 0.7714620232582092, | |
| "learning_rate": 7.735432867977679e-05, | |
| "loss": 1.82887939453125, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "eval_cer": 0.40629423070237736, | |
| "eval_loss": 1.485690951347351, | |
| "eval_runtime": 232.1567, | |
| "eval_samples_per_second": 8.615, | |
| "eval_steps_per_second": 0.271, | |
| "eval_wer": 0.7139462387571546, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.35125, | |
| "grad_norm": 0.780205249786377, | |
| "learning_rate": 7.718108547352488e-05, | |
| "loss": 1.8689053344726563, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 0.3525, | |
| "grad_norm": 0.779252827167511, | |
| "learning_rate": 7.70073778175174e-05, | |
| "loss": 1.9316950988769532, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 0.35375, | |
| "grad_norm": 0.8561552166938782, | |
| "learning_rate": 7.683320867993892e-05, | |
| "loss": 1.9451611328125, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 0.355, | |
| "grad_norm": 1.0372787714004517, | |
| "learning_rate": 7.665858103685944e-05, | |
| "loss": 1.8706069946289063, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.35625, | |
| "grad_norm": 1.1541863679885864, | |
| "learning_rate": 7.648349787218355e-05, | |
| "loss": 1.8886134338378906, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 0.3575, | |
| "grad_norm": 0.7528390288352966, | |
| "learning_rate": 7.63079621775995e-05, | |
| "loss": 1.84392333984375, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 0.35875, | |
| "grad_norm": 0.8374006152153015, | |
| "learning_rate": 7.613197695252796e-05, | |
| "loss": 1.8741331481933594, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.8712308406829834, | |
| "learning_rate": 7.595554520407088e-05, | |
| "loss": 1.8904888916015625, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.36125, | |
| "grad_norm": 0.7469571828842163, | |
| "learning_rate": 7.577866994696007e-05, | |
| "loss": 1.8809963989257812, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 0.3625, | |
| "grad_norm": 0.8286027908325195, | |
| "learning_rate": 7.560135420350562e-05, | |
| "loss": 1.8762887573242188, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 0.36375, | |
| "grad_norm": 0.8619410395622253, | |
| "learning_rate": 7.542360100354436e-05, | |
| "loss": 1.8658367919921874, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 0.365, | |
| "grad_norm": 0.8482415676116943, | |
| "learning_rate": 7.524541338438807e-05, | |
| "loss": 1.890072784423828, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.36625, | |
| "grad_norm": 0.9963574409484863, | |
| "learning_rate": 7.506679439077148e-05, | |
| "loss": 1.813338623046875, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 0.3675, | |
| "grad_norm": 0.9238027334213257, | |
| "learning_rate": 7.488774707480042e-05, | |
| "loss": 1.8132342529296874, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 0.36875, | |
| "grad_norm": 1.035698652267456, | |
| "learning_rate": 7.470827449589947e-05, | |
| "loss": 1.8227348327636719, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 0.7535447478294373, | |
| "learning_rate": 7.452837972075983e-05, | |
| "loss": 1.7920472717285156, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.37125, | |
| "grad_norm": 1.0825896263122559, | |
| "learning_rate": 7.434806582328686e-05, | |
| "loss": 1.8934266662597656, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 0.3725, | |
| "grad_norm": 0.7823067307472229, | |
| "learning_rate": 7.416733588454758e-05, | |
| "loss": 1.85238525390625, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 0.37375, | |
| "grad_norm": 3.7756171226501465, | |
| "learning_rate": 7.398619299271798e-05, | |
| "loss": 1.9932785034179688, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 0.375, | |
| "grad_norm": 0.8524430394172668, | |
| "learning_rate": 7.380464024303028e-05, | |
| "loss": 1.87403564453125, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.37625, | |
| "grad_norm": 0.9760676026344299, | |
| "learning_rate": 7.362268073772007e-05, | |
| "loss": 1.8505853271484376, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 0.3775, | |
| "grad_norm": 0.9811142683029175, | |
| "learning_rate": 7.344031758597325e-05, | |
| "loss": 1.860489044189453, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 0.37875, | |
| "grad_norm": 0.7913584113121033, | |
| "learning_rate": 7.325755390387292e-05, | |
| "loss": 1.8794989013671874, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 0.8596822619438171, | |
| "learning_rate": 7.307439281434615e-05, | |
| "loss": 1.8047637939453125, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.38125, | |
| "grad_norm": 1.4376176595687866, | |
| "learning_rate": 7.28908374471106e-05, | |
| "loss": 1.844581298828125, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 0.3825, | |
| "grad_norm": 0.8676425218582153, | |
| "learning_rate": 7.270689093862105e-05, | |
| "loss": 1.7712818908691406, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 0.38375, | |
| "grad_norm": 0.9175459146499634, | |
| "learning_rate": 7.252255643201579e-05, | |
| "loss": 1.8562504577636718, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 0.385, | |
| "grad_norm": 1.0078133344650269, | |
| "learning_rate": 7.233783707706295e-05, | |
| "loss": 1.855503692626953, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.38625, | |
| "grad_norm": 0.9916481375694275, | |
| "learning_rate": 7.215273603010668e-05, | |
| "loss": 1.8638204956054687, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 0.3875, | |
| "grad_norm": 0.9561924338340759, | |
| "learning_rate": 7.196725645401309e-05, | |
| "loss": 1.848057861328125, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 0.38875, | |
| "grad_norm": 1.0393955707550049, | |
| "learning_rate": 7.178140151811647e-05, | |
| "loss": 1.8322097778320312, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 0.811954140663147, | |
| "learning_rate": 7.159517439816481e-05, | |
| "loss": 1.9043869018554687, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.39125, | |
| "grad_norm": 0.9596728086471558, | |
| "learning_rate": 7.14085782762658e-05, | |
| "loss": 1.834969024658203, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 0.3925, | |
| "grad_norm": 0.8850217461585999, | |
| "learning_rate": 7.122161634083234e-05, | |
| "loss": 1.9156986999511718, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 0.39375, | |
| "grad_norm": 0.7752941250801086, | |
| "learning_rate": 7.103429178652803e-05, | |
| "loss": 1.889502716064453, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 0.395, | |
| "grad_norm": 1.023840069770813, | |
| "learning_rate": 7.084660781421268e-05, | |
| "loss": 1.8967515563964843, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.39625, | |
| "grad_norm": 0.8335869312286377, | |
| "learning_rate": 7.065856763088754e-05, | |
| "loss": 1.8889625549316407, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 0.3975, | |
| "grad_norm": 0.8171396255493164, | |
| "learning_rate": 7.04701744496405e-05, | |
| "loss": 1.798343963623047, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 0.39875, | |
| "grad_norm": 0.9180098176002502, | |
| "learning_rate": 7.028143148959126e-05, | |
| "loss": 1.8260577392578126, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.9625085592269897, | |
| "learning_rate": 7.009234197583623e-05, | |
| "loss": 1.8844915771484374, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_cer": 0.3979409909663834, | |
| "eval_loss": 1.4832241535186768, | |
| "eval_runtime": 233.2525, | |
| "eval_samples_per_second": 8.574, | |
| "eval_steps_per_second": 0.27, | |
| "eval_wer": 0.6957532706459526, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.40125, | |
| "grad_norm": 0.9736983180046082, | |
| "learning_rate": 6.99029091393935e-05, | |
| "loss": 1.8964356994628906, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 0.4025, | |
| "grad_norm": 0.8102937936782837, | |
| "learning_rate": 6.971313621714756e-05, | |
| "loss": 1.7914996337890625, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 0.40375, | |
| "grad_norm": 0.9826732873916626, | |
| "learning_rate": 6.952302645179408e-05, | |
| "loss": 1.8220474243164062, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 0.405, | |
| "grad_norm": 1.3437153100967407, | |
| "learning_rate": 6.933258309178438e-05, | |
| "loss": 1.8824446105957031, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.40625, | |
| "grad_norm": 0.9603751301765442, | |
| "learning_rate": 6.914180939127005e-05, | |
| "loss": 1.881730499267578, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 0.4075, | |
| "grad_norm": 0.8415713310241699, | |
| "learning_rate": 6.895070861004729e-05, | |
| "loss": 1.9096884155273437, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 0.40875, | |
| "grad_norm": 0.9102653861045837, | |
| "learning_rate": 6.875928401350116e-05, | |
| "loss": 1.847227783203125, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 0.6367753148078918, | |
| "learning_rate": 6.856753887254986e-05, | |
| "loss": 1.9515626525878906, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.41125, | |
| "grad_norm": 0.831386387348175, | |
| "learning_rate": 6.837547646358882e-05, | |
| "loss": 1.8950588989257813, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 0.4125, | |
| "grad_norm": 0.9433812499046326, | |
| "learning_rate": 6.818310006843468e-05, | |
| "loss": 1.8682933044433594, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 0.41375, | |
| "grad_norm": 0.9233819246292114, | |
| "learning_rate": 6.799041297426925e-05, | |
| "loss": 1.860150146484375, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 0.415, | |
| "grad_norm": 0.8766050338745117, | |
| "learning_rate": 6.779741847358332e-05, | |
| "loss": 1.765559539794922, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.41625, | |
| "grad_norm": 0.8227185010910034, | |
| "learning_rate": 6.760411986412043e-05, | |
| "loss": 1.860703582763672, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 0.4175, | |
| "grad_norm": 0.9237884283065796, | |
| "learning_rate": 6.741052044882048e-05, | |
| "loss": 1.8087977600097656, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 0.41875, | |
| "grad_norm": 1.1772114038467407, | |
| "learning_rate": 6.721662353576332e-05, | |
| "loss": 1.756484832763672, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 0.8661752939224243, | |
| "learning_rate": 6.702243243811221e-05, | |
| "loss": 1.8151901245117188, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.42125, | |
| "grad_norm": 1.0115716457366943, | |
| "learning_rate": 6.682795047405724e-05, | |
| "loss": 1.8263777160644532, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 0.4225, | |
| "grad_norm": 1.04531729221344, | |
| "learning_rate": 6.663318096675854e-05, | |
| "loss": 1.8317567443847655, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 0.42375, | |
| "grad_norm": 0.8114931583404541, | |
| "learning_rate": 6.643812724428964e-05, | |
| "loss": 1.8492442321777345, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 0.425, | |
| "grad_norm": 0.9216915965080261, | |
| "learning_rate": 6.624279263958047e-05, | |
| "loss": 1.8715736389160156, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.42625, | |
| "grad_norm": 0.9087522625923157, | |
| "learning_rate": 6.604718049036048e-05, | |
| "loss": 1.8946910095214844, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 0.4275, | |
| "grad_norm": 0.9846459627151489, | |
| "learning_rate": 6.585129413910159e-05, | |
| "loss": 1.7952665710449218, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 0.42875, | |
| "grad_norm": 0.9188402891159058, | |
| "learning_rate": 6.565513693296104e-05, | |
| "loss": 1.8260617065429687, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 0.973331868648529, | |
| "learning_rate": 6.545871222372436e-05, | |
| "loss": 1.8059585571289063, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.43125, | |
| "grad_norm": 1.011702537536621, | |
| "learning_rate": 6.526202336774776e-05, | |
| "loss": 1.8327102661132812, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 0.4325, | |
| "grad_norm": 0.8708674311637878, | |
| "learning_rate": 6.506507372590119e-05, | |
| "loss": 1.8367582702636718, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 0.43375, | |
| "grad_norm": 0.8796373605728149, | |
| "learning_rate": 6.486786666351057e-05, | |
| "loss": 1.8570964050292968, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 0.435, | |
| "grad_norm": 1.0652796030044556, | |
| "learning_rate": 6.467040555030052e-05, | |
| "loss": 1.7981742858886718, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.43625, | |
| "grad_norm": 0.9230595827102661, | |
| "learning_rate": 6.447269376033664e-05, | |
| "loss": 1.8698123168945313, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 0.4375, | |
| "grad_norm": 1.1134626865386963, | |
| "learning_rate": 6.427473467196793e-05, | |
| "loss": 1.8410800170898438, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 0.43875, | |
| "grad_norm": 0.7997915744781494, | |
| "learning_rate": 6.4076531667769e-05, | |
| "loss": 1.7883843994140625, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.9203572273254395, | |
| "learning_rate": 6.387808813448234e-05, | |
| "loss": 1.7845350646972655, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.44125, | |
| "grad_norm": 0.9123949408531189, | |
| "learning_rate": 6.367940746296041e-05, | |
| "loss": 1.780988006591797, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 0.4425, | |
| "grad_norm": 0.9542874693870544, | |
| "learning_rate": 6.348049304810771e-05, | |
| "loss": 1.9114262390136718, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 0.44375, | |
| "grad_norm": 0.9000183343887329, | |
| "learning_rate": 6.328134828882276e-05, | |
| "loss": 1.923231964111328, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 0.445, | |
| "grad_norm": 1.1514582633972168, | |
| "learning_rate": 6.308197658794003e-05, | |
| "loss": 1.90771484375, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.44625, | |
| "grad_norm": 0.9541072249412537, | |
| "learning_rate": 6.288238135217183e-05, | |
| "loss": 1.8115818786621094, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 0.4475, | |
| "grad_norm": 0.9740118384361267, | |
| "learning_rate": 6.268256599205003e-05, | |
| "loss": 1.7607234191894532, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 0.44875, | |
| "grad_norm": 1.038560152053833, | |
| "learning_rate": 6.248253392186781e-05, | |
| "loss": 1.7897041320800782, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 1.0086963176727295, | |
| "learning_rate": 6.228228855962133e-05, | |
| "loss": 1.7763148498535157, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "eval_cer": 0.3946185668383335, | |
| "eval_loss": 1.4465574026107788, | |
| "eval_runtime": 219.3161, | |
| "eval_samples_per_second": 9.119, | |
| "eval_steps_per_second": 0.287, | |
| "eval_wer": 0.6937857726901063, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.45125, | |
| "grad_norm": 0.9602487683296204, | |
| "learning_rate": 6.208183332695135e-05, | |
| "loss": 1.8459107971191406, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 0.4525, | |
| "grad_norm": 1.2571187019348145, | |
| "learning_rate": 6.188117164908474e-05, | |
| "loss": 1.845450439453125, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 0.45375, | |
| "grad_norm": 1.0202518701553345, | |
| "learning_rate": 6.168030695477591e-05, | |
| "loss": 1.797836151123047, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 0.455, | |
| "grad_norm": 0.8882875442504883, | |
| "learning_rate": 6.147924267624829e-05, | |
| "loss": 1.7729702758789063, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.45625, | |
| "grad_norm": 0.7974418997764587, | |
| "learning_rate": 6.127798224913564e-05, | |
| "loss": 1.8427424621582031, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 0.4575, | |
| "grad_norm": 0.9373735785484314, | |
| "learning_rate": 6.107652911242336e-05, | |
| "loss": 1.8868913269042968, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 0.45875, | |
| "grad_norm": 1.0345109701156616, | |
| "learning_rate": 6.0874886708389745e-05, | |
| "loss": 1.8657403564453126, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 0.7744643092155457, | |
| "learning_rate": 6.067305848254709e-05, | |
| "loss": 1.8451573181152343, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.46125, | |
| "grad_norm": 0.9977173209190369, | |
| "learning_rate": 6.047104788358297e-05, | |
| "loss": 1.82953857421875, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 0.4625, | |
| "grad_norm": 1.1429433822631836, | |
| "learning_rate": 6.0268858363301105e-05, | |
| "loss": 1.842634735107422, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 0.46375, | |
| "grad_norm": 0.7920505404472351, | |
| "learning_rate": 6.006649337656256e-05, | |
| "loss": 1.9003070068359376, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 0.465, | |
| "grad_norm": 1.2922805547714233, | |
| "learning_rate": 5.9863956381226607e-05, | |
| "loss": 1.9844851684570313, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.46625, | |
| "grad_norm": 0.8898814916610718, | |
| "learning_rate": 5.9661250838091675e-05, | |
| "loss": 1.8634324645996094, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 0.4675, | |
| "grad_norm": 0.9574541449546814, | |
| "learning_rate": 5.945838021083623e-05, | |
| "loss": 1.8424851989746094, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 0.46875, | |
| "grad_norm": 0.9146164059638977, | |
| "learning_rate": 5.925534796595953e-05, | |
| "loss": 1.8926506042480469, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 0.8964924812316895, | |
| "learning_rate": 5.905215757272248e-05, | |
| "loss": 1.9138760375976562, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.47125, | |
| "grad_norm": 1.3129435777664185, | |
| "learning_rate": 5.8848812503088244e-05, | |
| "loss": 1.8968862915039062, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 0.4725, | |
| "grad_norm": 0.9037919044494629, | |
| "learning_rate": 5.864531623166305e-05, | |
| "loss": 1.8491824340820313, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 0.47375, | |
| "grad_norm": 0.7935272455215454, | |
| "learning_rate": 5.844167223563669e-05, | |
| "loss": 1.925780029296875, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 0.475, | |
| "grad_norm": 0.9731354117393494, | |
| "learning_rate": 5.82378839947232e-05, | |
| "loss": 1.9472561645507813, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.47625, | |
| "grad_norm": 1.2544164657592773, | |
| "learning_rate": 5.803395499110131e-05, | |
| "loss": 1.8374139404296874, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 0.4775, | |
| "grad_norm": 0.7206196188926697, | |
| "learning_rate": 5.782988870935509e-05, | |
| "loss": 1.8779299926757813, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 0.47875, | |
| "grad_norm": 0.9583224654197693, | |
| "learning_rate": 5.76256886364142e-05, | |
| "loss": 1.9695869445800782, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 1.2145416736602783, | |
| "learning_rate": 5.742135826149453e-05, | |
| "loss": 1.9242964172363282, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.48125, | |
| "grad_norm": 1.039480447769165, | |
| "learning_rate": 5.721690107603838e-05, | |
| "loss": 1.7802023315429687, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 0.4825, | |
| "grad_norm": 0.8302348852157593, | |
| "learning_rate": 5.7012320573654945e-05, | |
| "loss": 1.8482383728027343, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 0.48375, | |
| "grad_norm": 1.02461576461792, | |
| "learning_rate": 5.6807620250060554e-05, | |
| "loss": 1.8193183898925782, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 0.485, | |
| "grad_norm": 1.1633189916610718, | |
| "learning_rate": 5.660280360301896e-05, | |
| "loss": 1.8072488403320313, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.48625, | |
| "grad_norm": 1.2958931922912598, | |
| "learning_rate": 5.6397874132281535e-05, | |
| "loss": 1.801346893310547, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 0.4875, | |
| "grad_norm": 0.744342029094696, | |
| "learning_rate": 5.619283533952754e-05, | |
| "loss": 1.7783653259277343, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 0.48875, | |
| "grad_norm": 0.9049773216247559, | |
| "learning_rate": 5.5987690728304195e-05, | |
| "loss": 1.8744984436035157, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 0.9979234933853149, | |
| "learning_rate": 5.578244380396691e-05, | |
| "loss": 1.8616281127929688, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.49125, | |
| "grad_norm": 1.021041989326477, | |
| "learning_rate": 5.5577098073619314e-05, | |
| "loss": 1.8032304382324218, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 0.4925, | |
| "grad_norm": 0.9475110173225403, | |
| "learning_rate": 5.5371657046053384e-05, | |
| "loss": 1.7942196655273437, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 0.49375, | |
| "grad_norm": 1.221889615058899, | |
| "learning_rate": 5.516612423168943e-05, | |
| "loss": 1.8513232421875, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 0.495, | |
| "grad_norm": 0.9770843982696533, | |
| "learning_rate": 5.49605031425162e-05, | |
| "loss": 1.813275146484375, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.49625, | |
| "grad_norm": 1.0538535118103027, | |
| "learning_rate": 5.4754797292030734e-05, | |
| "loss": 1.8585232543945311, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 0.4975, | |
| "grad_norm": 1.1689670085906982, | |
| "learning_rate": 5.4549010195178505e-05, | |
| "loss": 2.0056752014160155, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 0.49875, | |
| "grad_norm": 0.9732740521430969, | |
| "learning_rate": 5.4343145368293166e-05, | |
| "loss": 1.9135684204101562, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 1.0371040105819702, | |
| "learning_rate": 5.413720632903664e-05, | |
| "loss": 1.9369529724121093, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "eval_cer": 0.3916139829856128, | |
| "eval_loss": 1.4235085248947144, | |
| "eval_runtime": 220.6314, | |
| "eval_samples_per_second": 9.065, | |
| "eval_steps_per_second": 0.286, | |
| "eval_wer": 0.682466271463614, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.50125, | |
| "grad_norm": 0.8778891563415527, | |
| "learning_rate": 5.393119659633888e-05, | |
| "loss": 1.92362548828125, | |
| "step": 10025 | |
| }, | |
| { | |
| "epoch": 0.5025, | |
| "grad_norm": 0.8749571442604065, | |
| "learning_rate": 5.3725119690337846e-05, | |
| "loss": 1.8752864074707032, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 0.50375, | |
| "grad_norm": 1.1466436386108398, | |
| "learning_rate": 5.351897913231925e-05, | |
| "loss": 1.7905665588378907, | |
| "step": 10075 | |
| }, | |
| { | |
| "epoch": 0.505, | |
| "grad_norm": 1.1386643648147583, | |
| "learning_rate": 5.331277844465647e-05, | |
| "loss": 1.8082792663574219, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 0.50625, | |
| "grad_norm": 0.8335414528846741, | |
| "learning_rate": 5.3106521150750345e-05, | |
| "loss": 1.8966696166992187, | |
| "step": 10125 | |
| }, | |
| { | |
| "epoch": 0.5075, | |
| "grad_norm": 1.1373592615127563, | |
| "learning_rate": 5.290021077496893e-05, | |
| "loss": 1.8417135620117187, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 0.50875, | |
| "grad_norm": 1.0658506155014038, | |
| "learning_rate": 5.2693850842587336e-05, | |
| "loss": 1.7909144592285156, | |
| "step": 10175 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 0.8956384658813477, | |
| "learning_rate": 5.248744487972742e-05, | |
| "loss": 1.7616961669921876, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.51125, | |
| "grad_norm": 0.9346334934234619, | |
| "learning_rate": 5.22809964132976e-05, | |
| "loss": 1.760950469970703, | |
| "step": 10225 | |
| }, | |
| { | |
| "epoch": 0.5125, | |
| "grad_norm": 1.2858171463012695, | |
| "learning_rate": 5.207450897093257e-05, | |
| "loss": 1.9476943969726563, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 0.51375, | |
| "grad_norm": 0.9099264740943909, | |
| "learning_rate": 5.186798608093298e-05, | |
| "loss": 1.9370100402832031, | |
| "step": 10275 | |
| }, | |
| { | |
| "epoch": 0.515, | |
| "grad_norm": 1.0369707345962524, | |
| "learning_rate": 5.166143127220524e-05, | |
| "loss": 1.8100379943847655, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 0.51625, | |
| "grad_norm": 1.0726726055145264, | |
| "learning_rate": 5.145484807420109e-05, | |
| "loss": 1.7876473999023437, | |
| "step": 10325 | |
| }, | |
| { | |
| "epoch": 0.5175, | |
| "grad_norm": 1.460521936416626, | |
| "learning_rate": 5.124824001685741e-05, | |
| "loss": 1.7366654968261719, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 0.51875, | |
| "grad_norm": 1.117113471031189, | |
| "learning_rate": 5.104161063053585e-05, | |
| "loss": 1.7056991577148437, | |
| "step": 10375 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 0.9027568697929382, | |
| "learning_rate": 5.0834963445962524e-05, | |
| "loss": 1.7607815551757813, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 0.52125, | |
| "grad_norm": 0.9944213032722473, | |
| "learning_rate": 5.062830199416764e-05, | |
| "loss": 1.866544189453125, | |
| "step": 10425 | |
| }, | |
| { | |
| "epoch": 0.5225, | |
| "grad_norm": 0.9849388599395752, | |
| "learning_rate": 5.042162980642523e-05, | |
| "loss": 1.8680609130859376, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 0.52375, | |
| "grad_norm": 0.9672307372093201, | |
| "learning_rate": 5.021495041419273e-05, | |
| "loss": 1.8579081726074218, | |
| "step": 10475 | |
| }, | |
| { | |
| "epoch": 0.525, | |
| "grad_norm": 0.9958829283714294, | |
| "learning_rate": 5.000826734905073e-05, | |
| "loss": 1.8764848327636718, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.52625, | |
| "grad_norm": 1.0822824239730835, | |
| "learning_rate": 4.980158414264254e-05, | |
| "loss": 1.8176104736328125, | |
| "step": 10525 | |
| }, | |
| { | |
| "epoch": 0.5275, | |
| "grad_norm": 1.1756185293197632, | |
| "learning_rate": 4.959490432661391e-05, | |
| "loss": 1.8305401611328125, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 0.52875, | |
| "grad_norm": 1.0455031394958496, | |
| "learning_rate": 4.938823143255264e-05, | |
| "loss": 1.8224380493164063, | |
| "step": 10575 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 1.2272131443023682, | |
| "learning_rate": 4.918156899192826e-05, | |
| "loss": 1.8834970092773438, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 0.53125, | |
| "grad_norm": 0.9592375159263611, | |
| "learning_rate": 4.897492053603168e-05, | |
| "loss": 1.8808551025390625, | |
| "step": 10625 | |
| }, | |
| { | |
| "epoch": 0.5325, | |
| "grad_norm": 0.9799576997756958, | |
| "learning_rate": 4.876828959591485e-05, | |
| "loss": 1.8907269287109374, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 0.53375, | |
| "grad_norm": 1.0906355381011963, | |
| "learning_rate": 4.856167970233042e-05, | |
| "loss": 1.9806098937988281, | |
| "step": 10675 | |
| }, | |
| { | |
| "epoch": 0.535, | |
| "grad_norm": 0.8884340524673462, | |
| "learning_rate": 4.835509438567142e-05, | |
| "loss": 1.9684288024902343, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 0.53625, | |
| "grad_norm": 0.9048951268196106, | |
| "learning_rate": 4.814853717591091e-05, | |
| "loss": 1.7778785705566407, | |
| "step": 10725 | |
| }, | |
| { | |
| "epoch": 0.5375, | |
| "grad_norm": 0.9540491104125977, | |
| "learning_rate": 4.794201160254171e-05, | |
| "loss": 1.874403839111328, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 0.53875, | |
| "grad_norm": 1.1666539907455444, | |
| "learning_rate": 4.773552119451603e-05, | |
| "loss": 1.8194976806640626, | |
| "step": 10775 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 0.8927249908447266, | |
| "learning_rate": 4.752906948018525e-05, | |
| "loss": 1.755653076171875, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 0.54125, | |
| "grad_norm": 1.0932155847549438, | |
| "learning_rate": 4.7322659987239514e-05, | |
| "loss": 1.7385015869140625, | |
| "step": 10825 | |
| }, | |
| { | |
| "epoch": 0.5425, | |
| "grad_norm": 0.8828619718551636, | |
| "learning_rate": 4.7116296242647554e-05, | |
| "loss": 1.780015411376953, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 0.54375, | |
| "grad_norm": 0.9021382331848145, | |
| "learning_rate": 4.6909981772596415e-05, | |
| "loss": 1.800555419921875, | |
| "step": 10875 | |
| }, | |
| { | |
| "epoch": 0.545, | |
| "grad_norm": 1.1465368270874023, | |
| "learning_rate": 4.670372010243111e-05, | |
| "loss": 1.7544993591308593, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 0.54625, | |
| "grad_norm": 0.951725959777832, | |
| "learning_rate": 4.6497514756594526e-05, | |
| "loss": 1.7968252563476563, | |
| "step": 10925 | |
| }, | |
| { | |
| "epoch": 0.5475, | |
| "grad_norm": 1.015174388885498, | |
| "learning_rate": 4.629136925856705e-05, | |
| "loss": 1.902012481689453, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 0.54875, | |
| "grad_norm": 1.1253248453140259, | |
| "learning_rate": 4.608528713080651e-05, | |
| "loss": 1.7841081237792968, | |
| "step": 10975 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 1.062878131866455, | |
| "learning_rate": 4.5879271894687814e-05, | |
| "loss": 1.767802734375, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "eval_cer": 0.38568924160330553, | |
| "eval_loss": 1.4164085388183594, | |
| "eval_runtime": 221.7788, | |
| "eval_samples_per_second": 9.018, | |
| "eval_steps_per_second": 0.284, | |
| "eval_wer": 0.6784290678659035, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.55125, | |
| "grad_norm": 1.0052728652954102, | |
| "learning_rate": 4.567332707044298e-05, | |
| "loss": 1.8400352478027344, | |
| "step": 11025 | |
| }, | |
| { | |
| "epoch": 0.5525, | |
| "grad_norm": 0.8470182418823242, | |
| "learning_rate": 4.546745617710081e-05, | |
| "loss": 1.7786236572265626, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 0.55375, | |
| "grad_norm": 0.8040934205055237, | |
| "learning_rate": 4.5261662732426865e-05, | |
| "loss": 1.7493003845214843, | |
| "step": 11075 | |
| }, | |
| { | |
| "epoch": 0.555, | |
| "grad_norm": 0.9572652578353882, | |
| "learning_rate": 4.5055950252863296e-05, | |
| "loss": 1.7914910888671876, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 0.55625, | |
| "grad_norm": 1.0314631462097168, | |
| "learning_rate": 4.485032225346882e-05, | |
| "loss": 1.7660665893554688, | |
| "step": 11125 | |
| }, | |
| { | |
| "epoch": 0.5575, | |
| "grad_norm": 1.09141206741333, | |
| "learning_rate": 4.46447822478586e-05, | |
| "loss": 1.7657220458984375, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 0.55875, | |
| "grad_norm": 1.0272183418273926, | |
| "learning_rate": 4.4439333748144185e-05, | |
| "loss": 1.7961570739746093, | |
| "step": 11175 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 1.0564382076263428, | |
| "learning_rate": 4.4233980264873636e-05, | |
| "loss": 1.7062783813476563, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.56125, | |
| "grad_norm": 0.9640118479728699, | |
| "learning_rate": 4.402872530697135e-05, | |
| "loss": 1.781280517578125, | |
| "step": 11225 | |
| }, | |
| { | |
| "epoch": 0.5625, | |
| "grad_norm": 1.1091479063034058, | |
| "learning_rate": 4.3823572381678286e-05, | |
| "loss": 1.8192706298828125, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 0.56375, | |
| "grad_norm": 0.9963802099227905, | |
| "learning_rate": 4.361852499449185e-05, | |
| "loss": 1.8918753051757813, | |
| "step": 11275 | |
| }, | |
| { | |
| "epoch": 0.565, | |
| "grad_norm": 1.0567357540130615, | |
| "learning_rate": 4.34135866491062e-05, | |
| "loss": 1.879267578125, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 0.56625, | |
| "grad_norm": 0.9304851293563843, | |
| "learning_rate": 4.3208760847352146e-05, | |
| "loss": 1.8433013916015626, | |
| "step": 11325 | |
| }, | |
| { | |
| "epoch": 0.5675, | |
| "grad_norm": 1.175447940826416, | |
| "learning_rate": 4.3004051089137576e-05, | |
| "loss": 1.9177244567871095, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 0.56875, | |
| "grad_norm": 0.9959896802902222, | |
| "learning_rate": 4.2799460872387394e-05, | |
| "loss": 1.8130079650878905, | |
| "step": 11375 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 1.0206068754196167, | |
| "learning_rate": 4.2594993692983955e-05, | |
| "loss": 1.8562202453613281, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 0.57125, | |
| "grad_norm": 1.1970798969268799, | |
| "learning_rate": 4.239065304470718e-05, | |
| "loss": 1.8565621948242188, | |
| "step": 11425 | |
| }, | |
| { | |
| "epoch": 0.5725, | |
| "grad_norm": 0.9450256824493408, | |
| "learning_rate": 4.2186442419174984e-05, | |
| "loss": 1.8496116638183593, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 0.57375, | |
| "grad_norm": 1.106532096862793, | |
| "learning_rate": 4.198236530578349e-05, | |
| "loss": 1.859767303466797, | |
| "step": 11475 | |
| }, | |
| { | |
| "epoch": 0.575, | |
| "grad_norm": 0.9269622564315796, | |
| "learning_rate": 4.177842519164752e-05, | |
| "loss": 1.9058389282226562, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.57625, | |
| "grad_norm": 1.0396665334701538, | |
| "learning_rate": 4.1574625561540935e-05, | |
| "loss": 1.8232623291015626, | |
| "step": 11525 | |
| }, | |
| { | |
| "epoch": 0.5775, | |
| "grad_norm": 0.9114131927490234, | |
| "learning_rate": 4.13709698978371e-05, | |
| "loss": 1.8967308044433593, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 0.57875, | |
| "grad_norm": 1.1759916543960571, | |
| "learning_rate": 4.1167461680449405e-05, | |
| "loss": 1.84106689453125, | |
| "step": 11575 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 1.0380804538726807, | |
| "learning_rate": 4.0964104386771785e-05, | |
| "loss": 1.843013153076172, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 0.58125, | |
| "grad_norm": 0.9165422916412354, | |
| "learning_rate": 4.076090149161932e-05, | |
| "loss": 1.8948208618164062, | |
| "step": 11625 | |
| }, | |
| { | |
| "epoch": 0.5825, | |
| "grad_norm": 1.1688545942306519, | |
| "learning_rate": 4.055785646716882e-05, | |
| "loss": 1.8211753845214844, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 0.58375, | |
| "grad_norm": 0.7985947132110596, | |
| "learning_rate": 4.035497278289953e-05, | |
| "loss": 1.7845936584472657, | |
| "step": 11675 | |
| }, | |
| { | |
| "epoch": 0.585, | |
| "grad_norm": 1.287771463394165, | |
| "learning_rate": 4.015225390553385e-05, | |
| "loss": 1.7396124267578126, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 0.58625, | |
| "grad_norm": 1.0075743198394775, | |
| "learning_rate": 3.994970329897806e-05, | |
| "loss": 1.7854393005371094, | |
| "step": 11725 | |
| }, | |
| { | |
| "epoch": 0.5875, | |
| "grad_norm": 1.0524576902389526, | |
| "learning_rate": 3.974732442426319e-05, | |
| "loss": 1.9185934448242188, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 0.58875, | |
| "grad_norm": 1.0832891464233398, | |
| "learning_rate": 3.954512073948581e-05, | |
| "loss": 1.9127937316894532, | |
| "step": 11775 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 0.9133186340332031, | |
| "learning_rate": 3.9343095699749e-05, | |
| "loss": 1.9142672729492187, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 0.59125, | |
| "grad_norm": 0.9331486225128174, | |
| "learning_rate": 3.914125275710329e-05, | |
| "loss": 1.8372850036621093, | |
| "step": 11825 | |
| }, | |
| { | |
| "epoch": 0.5925, | |
| "grad_norm": 1.0655385255813599, | |
| "learning_rate": 3.8939595360487656e-05, | |
| "loss": 1.8685006713867187, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 0.59375, | |
| "grad_norm": 1.0582547187805176, | |
| "learning_rate": 3.873812695567059e-05, | |
| "loss": 1.850763397216797, | |
| "step": 11875 | |
| }, | |
| { | |
| "epoch": 0.595, | |
| "grad_norm": 0.9080460071563721, | |
| "learning_rate": 3.853685098519132e-05, | |
| "loss": 1.813930206298828, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 0.59625, | |
| "grad_norm": 1.1732892990112305, | |
| "learning_rate": 3.833577088830079e-05, | |
| "loss": 1.7444282531738282, | |
| "step": 11925 | |
| }, | |
| { | |
| "epoch": 0.5975, | |
| "grad_norm": 1.1060677766799927, | |
| "learning_rate": 3.81348901009031e-05, | |
| "loss": 1.7383247375488282, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 0.59875, | |
| "grad_norm": 0.9363880157470703, | |
| "learning_rate": 3.7934212055496595e-05, | |
| "loss": 1.8123402404785156, | |
| "step": 11975 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.920231282711029, | |
| "learning_rate": 3.7733740181115455e-05, | |
| "loss": 1.8402847290039062, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_cer": 0.3867917500583535, | |
| "eval_loss": 1.3981289863586426, | |
| "eval_runtime": 231.6217, | |
| "eval_samples_per_second": 8.635, | |
| "eval_steps_per_second": 0.272, | |
| "eval_wer": 0.669562551103843, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.60125, | |
| "grad_norm": 1.0437941551208496, | |
| "learning_rate": 3.753347790327084e-05, | |
| "loss": 1.796068115234375, | |
| "step": 12025 | |
| }, | |
| { | |
| "epoch": 0.6025, | |
| "grad_norm": 0.9707103371620178, | |
| "learning_rate": 3.7333428643892567e-05, | |
| "loss": 1.8068885803222656, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 0.60375, | |
| "grad_norm": 1.3142566680908203, | |
| "learning_rate": 3.713359582127049e-05, | |
| "loss": 1.7970997619628906, | |
| "step": 12075 | |
| }, | |
| { | |
| "epoch": 0.605, | |
| "grad_norm": 0.9063726663589478, | |
| "learning_rate": 3.693398284999623e-05, | |
| "loss": 1.8356533813476563, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 0.60625, | |
| "grad_norm": 1.1845322847366333, | |
| "learning_rate": 3.6734593140904676e-05, | |
| "loss": 1.827454833984375, | |
| "step": 12125 | |
| }, | |
| { | |
| "epoch": 0.6075, | |
| "grad_norm": 1.0417838096618652, | |
| "learning_rate": 3.6535430101015866e-05, | |
| "loss": 1.8590811157226563, | |
| "step": 12150 | |
| }, | |
| { | |
| "epoch": 0.60875, | |
| "grad_norm": 1.2524938583374023, | |
| "learning_rate": 3.6336497133476616e-05, | |
| "loss": 1.7782191467285156, | |
| "step": 12175 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 0.9469017386436462, | |
| "learning_rate": 3.6137797637502444e-05, | |
| "loss": 1.8327203369140626, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 0.61125, | |
| "grad_norm": 0.8404561281204224, | |
| "learning_rate": 3.593933500831955e-05, | |
| "loss": 1.8657476806640625, | |
| "step": 12225 | |
| }, | |
| { | |
| "epoch": 0.6125, | |
| "grad_norm": 1.105899691581726, | |
| "learning_rate": 3.5741112637106655e-05, | |
| "loss": 1.8155101013183594, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 0.61375, | |
| "grad_norm": 1.0258651971817017, | |
| "learning_rate": 3.55431339109372e-05, | |
| "loss": 1.7249429321289063, | |
| "step": 12275 | |
| }, | |
| { | |
| "epoch": 0.615, | |
| "grad_norm": 1.3268970251083374, | |
| "learning_rate": 3.5345402212721335e-05, | |
| "loss": 1.8360076904296876, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 0.61625, | |
| "grad_norm": 0.9011600613594055, | |
| "learning_rate": 3.5147920921148267e-05, | |
| "loss": 1.8858851623535156, | |
| "step": 12325 | |
| }, | |
| { | |
| "epoch": 0.6175, | |
| "grad_norm": 1.2116265296936035, | |
| "learning_rate": 3.495069341062836e-05, | |
| "loss": 1.812566375732422, | |
| "step": 12350 | |
| }, | |
| { | |
| "epoch": 0.61875, | |
| "grad_norm": 1.003523826599121, | |
| "learning_rate": 3.475372305123562e-05, | |
| "loss": 1.8327554321289063, | |
| "step": 12375 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 1.1544911861419678, | |
| "learning_rate": 3.4557013208650016e-05, | |
| "loss": 1.8175526428222657, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 0.62125, | |
| "grad_norm": 1.0257184505462646, | |
| "learning_rate": 3.436056724410005e-05, | |
| "loss": 1.7979986572265625, | |
| "step": 12425 | |
| }, | |
| { | |
| "epoch": 0.6225, | |
| "grad_norm": 1.7117664813995361, | |
| "learning_rate": 3.416438851430519e-05, | |
| "loss": 1.781560516357422, | |
| "step": 12450 | |
| }, | |
| { | |
| "epoch": 0.62375, | |
| "grad_norm": 1.3236567974090576, | |
| "learning_rate": 3.3968480371418706e-05, | |
| "loss": 1.8611416625976562, | |
| "step": 12475 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 0.7625004649162292, | |
| "learning_rate": 3.377284616297021e-05, | |
| "loss": 1.7145248413085938, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.62625, | |
| "grad_norm": 0.7954502701759338, | |
| "learning_rate": 3.357748923180852e-05, | |
| "loss": 1.7421397399902343, | |
| "step": 12525 | |
| }, | |
| { | |
| "epoch": 0.6275, | |
| "grad_norm": 1.0269776582717896, | |
| "learning_rate": 3.3382412916044645e-05, | |
| "loss": 1.6932389831542969, | |
| "step": 12550 | |
| }, | |
| { | |
| "epoch": 0.62875, | |
| "grad_norm": 1.1194988489151, | |
| "learning_rate": 3.318762054899453e-05, | |
| "loss": 1.7593035888671875, | |
| "step": 12575 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 1.716532826423645, | |
| "learning_rate": 3.2993115459122305e-05, | |
| "loss": 1.7944053649902343, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 0.63125, | |
| "grad_norm": 1.2546732425689697, | |
| "learning_rate": 3.2798900969983285e-05, | |
| "loss": 1.67067626953125, | |
| "step": 12625 | |
| }, | |
| { | |
| "epoch": 0.6325, | |
| "grad_norm": 1.072123646736145, | |
| "learning_rate": 3.2604980400167254e-05, | |
| "loss": 1.6887086486816407, | |
| "step": 12650 | |
| }, | |
| { | |
| "epoch": 0.63375, | |
| "grad_norm": 0.9773191213607788, | |
| "learning_rate": 3.241135706324168e-05, | |
| "loss": 1.6930819702148439, | |
| "step": 12675 | |
| }, | |
| { | |
| "epoch": 0.635, | |
| "grad_norm": 1.000594139099121, | |
| "learning_rate": 3.221803426769518e-05, | |
| "loss": 1.7778182983398438, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 0.63625, | |
| "grad_norm": 1.350174069404602, | |
| "learning_rate": 3.2025015316880935e-05, | |
| "loss": 1.6934915161132813, | |
| "step": 12725 | |
| }, | |
| { | |
| "epoch": 0.6375, | |
| "grad_norm": 0.8144739270210266, | |
| "learning_rate": 3.183230350896026e-05, | |
| "loss": 1.7666160583496093, | |
| "step": 12750 | |
| }, | |
| { | |
| "epoch": 0.63875, | |
| "grad_norm": 0.8452905416488647, | |
| "learning_rate": 3.163990213684623e-05, | |
| "loss": 1.78269775390625, | |
| "step": 12775 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 1.1083192825317383, | |
| "learning_rate": 3.144781448814746e-05, | |
| "loss": 1.8267408752441405, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.64125, | |
| "grad_norm": 0.9822102785110474, | |
| "learning_rate": 3.125604384511186e-05, | |
| "loss": 1.8954287719726564, | |
| "step": 12825 | |
| }, | |
| { | |
| "epoch": 0.6425, | |
| "grad_norm": 1.1005773544311523, | |
| "learning_rate": 3.10645934845706e-05, | |
| "loss": 1.8386691284179688, | |
| "step": 12850 | |
| }, | |
| { | |
| "epoch": 0.64375, | |
| "grad_norm": 0.8736421465873718, | |
| "learning_rate": 3.0873466677882144e-05, | |
| "loss": 1.8711613464355468, | |
| "step": 12875 | |
| }, | |
| { | |
| "epoch": 0.645, | |
| "grad_norm": 0.9403691291809082, | |
| "learning_rate": 3.068266669087625e-05, | |
| "loss": 1.834996795654297, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 0.64625, | |
| "grad_norm": 1.164202094078064, | |
| "learning_rate": 3.0492196783798277e-05, | |
| "loss": 1.7883038330078125, | |
| "step": 12925 | |
| }, | |
| { | |
| "epoch": 0.6475, | |
| "grad_norm": 1.035305380821228, | |
| "learning_rate": 3.0302060211253408e-05, | |
| "loss": 1.7249766540527345, | |
| "step": 12950 | |
| }, | |
| { | |
| "epoch": 0.64875, | |
| "grad_norm": 1.1645963191986084, | |
| "learning_rate": 3.0112260222151067e-05, | |
| "loss": 1.6899990844726562, | |
| "step": 12975 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 0.9183026552200317, | |
| "learning_rate": 2.9922800059649382e-05, | |
| "loss": 1.672317352294922, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "eval_cer": 0.3769436677774522, | |
| "eval_loss": 1.3849124908447266, | |
| "eval_runtime": 224.4076, | |
| "eval_samples_per_second": 8.912, | |
| "eval_steps_per_second": 0.281, | |
| "eval_wer": 0.6630723630417007, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.65125, | |
| "grad_norm": 1.0245611667633057, | |
| "learning_rate": 2.97336829610998e-05, | |
| "loss": 1.68808837890625, | |
| "step": 13025 | |
| }, | |
| { | |
| "epoch": 0.6525, | |
| "grad_norm": 1.2814953327178955, | |
| "learning_rate": 2.9544912157991745e-05, | |
| "loss": 1.7415261840820313, | |
| "step": 13050 | |
| }, | |
| { | |
| "epoch": 0.65375, | |
| "grad_norm": 1.1460628509521484, | |
| "learning_rate": 2.9356490875897403e-05, | |
| "loss": 1.7819834899902345, | |
| "step": 13075 | |
| }, | |
| { | |
| "epoch": 0.655, | |
| "grad_norm": 1.0019270181655884, | |
| "learning_rate": 2.916842233441661e-05, | |
| "loss": 1.7759149169921875, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 0.65625, | |
| "grad_norm": 1.1836796998977661, | |
| "learning_rate": 2.898070974712186e-05, | |
| "loss": 1.6897103881835938, | |
| "step": 13125 | |
| }, | |
| { | |
| "epoch": 0.6575, | |
| "grad_norm": 0.8534275889396667, | |
| "learning_rate": 2.8793356321503306e-05, | |
| "loss": 1.7299311828613282, | |
| "step": 13150 | |
| }, | |
| { | |
| "epoch": 0.65875, | |
| "grad_norm": 0.9979979395866394, | |
| "learning_rate": 2.8606365258914125e-05, | |
| "loss": 1.6920565795898437, | |
| "step": 13175 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 1.1418875455856323, | |
| "learning_rate": 2.8419739754515616e-05, | |
| "loss": 1.6824465942382814, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 0.66125, | |
| "grad_norm": 1.0111151933670044, | |
| "learning_rate": 2.8233482997222728e-05, | |
| "loss": 1.7573744201660155, | |
| "step": 13225 | |
| }, | |
| { | |
| "epoch": 0.6625, | |
| "grad_norm": 0.7626604437828064, | |
| "learning_rate": 2.804759816964957e-05, | |
| "loss": 1.9496656799316405, | |
| "step": 13250 | |
| }, | |
| { | |
| "epoch": 0.66375, | |
| "grad_norm": 1.1094297170639038, | |
| "learning_rate": 2.7862088448054936e-05, | |
| "loss": 1.8660400390625, | |
| "step": 13275 | |
| }, | |
| { | |
| "epoch": 0.665, | |
| "grad_norm": 1.0341440439224243, | |
| "learning_rate": 2.7676957002288163e-05, | |
| "loss": 1.7471173095703125, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 0.66625, | |
| "grad_norm": 1.339675784111023, | |
| "learning_rate": 2.7492206995734837e-05, | |
| "loss": 1.7066221618652344, | |
| "step": 13325 | |
| }, | |
| { | |
| "epoch": 0.6675, | |
| "grad_norm": 0.9849182963371277, | |
| "learning_rate": 2.730784158526286e-05, | |
| "loss": 1.794190673828125, | |
| "step": 13350 | |
| }, | |
| { | |
| "epoch": 0.66875, | |
| "grad_norm": 1.0803141593933105, | |
| "learning_rate": 2.712386392116837e-05, | |
| "loss": 1.70763916015625, | |
| "step": 13375 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 0.9134665727615356, | |
| "learning_rate": 2.6940277147122085e-05, | |
| "loss": 1.8360369873046876, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 0.67125, | |
| "grad_norm": 0.9723681807518005, | |
| "learning_rate": 2.6757084400115394e-05, | |
| "loss": 1.789735107421875, | |
| "step": 13425 | |
| }, | |
| { | |
| "epoch": 0.6725, | |
| "grad_norm": 0.9309587478637695, | |
| "learning_rate": 2.6574288810406946e-05, | |
| "loss": 1.72171875, | |
| "step": 13450 | |
| }, | |
| { | |
| "epoch": 0.67375, | |
| "grad_norm": 1.118057131767273, | |
| "learning_rate": 2.6391893501468968e-05, | |
| "loss": 1.7434465026855468, | |
| "step": 13475 | |
| }, | |
| { | |
| "epoch": 0.675, | |
| "grad_norm": 0.9700250029563904, | |
| "learning_rate": 2.620990158993406e-05, | |
| "loss": 1.7831150817871093, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.67625, | |
| "grad_norm": 0.9992932081222534, | |
| "learning_rate": 2.6028316185541896e-05, | |
| "loss": 1.7688768005371094, | |
| "step": 13525 | |
| }, | |
| { | |
| "epoch": 0.6775, | |
| "grad_norm": 1.0656723976135254, | |
| "learning_rate": 2.5847140391085972e-05, | |
| "loss": 1.7229135131835938, | |
| "step": 13550 | |
| }, | |
| { | |
| "epoch": 0.67875, | |
| "grad_norm": 0.9617791771888733, | |
| "learning_rate": 2.5666377302360776e-05, | |
| "loss": 1.8882518005371094, | |
| "step": 13575 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 1.0879966020584106, | |
| "learning_rate": 2.548603000810872e-05, | |
| "loss": 1.8817071533203125, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 0.68125, | |
| "grad_norm": 1.2155507802963257, | |
| "learning_rate": 2.530610158996749e-05, | |
| "loss": 1.790412139892578, | |
| "step": 13625 | |
| }, | |
| { | |
| "epoch": 0.6825, | |
| "grad_norm": 1.12446129322052, | |
| "learning_rate": 2.5126595122417295e-05, | |
| "loss": 1.6981216430664063, | |
| "step": 13650 | |
| }, | |
| { | |
| "epoch": 0.68375, | |
| "grad_norm": 1.1474330425262451, | |
| "learning_rate": 2.4947513672728414e-05, | |
| "loss": 1.7459504699707031, | |
| "step": 13675 | |
| }, | |
| { | |
| "epoch": 0.685, | |
| "grad_norm": 1.083685040473938, | |
| "learning_rate": 2.4768860300908685e-05, | |
| "loss": 1.8179473876953125, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 0.68625, | |
| "grad_norm": 1.052711009979248, | |
| "learning_rate": 2.459063805965135e-05, | |
| "loss": 1.779862518310547, | |
| "step": 13725 | |
| }, | |
| { | |
| "epoch": 0.6875, | |
| "grad_norm": 1.0125257968902588, | |
| "learning_rate": 2.4412849994282742e-05, | |
| "loss": 1.7436151123046875, | |
| "step": 13750 | |
| }, | |
| { | |
| "epoch": 0.68875, | |
| "grad_norm": 1.0145083665847778, | |
| "learning_rate": 2.4235499142710404e-05, | |
| "loss": 1.7439036560058594, | |
| "step": 13775 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 0.8667387366294861, | |
| "learning_rate": 2.4058588535371017e-05, | |
| "loss": 1.7218632507324219, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 0.69125, | |
| "grad_norm": 1.1944369077682495, | |
| "learning_rate": 2.3882121195178797e-05, | |
| "loss": 1.7588336181640625, | |
| "step": 13825 | |
| }, | |
| { | |
| "epoch": 0.6925, | |
| "grad_norm": 1.2647547721862793, | |
| "learning_rate": 2.3706100137473667e-05, | |
| "loss": 1.8015342712402345, | |
| "step": 13850 | |
| }, | |
| { | |
| "epoch": 0.69375, | |
| "grad_norm": 0.9314883947372437, | |
| "learning_rate": 2.3530528369969833e-05, | |
| "loss": 1.8014820861816405, | |
| "step": 13875 | |
| }, | |
| { | |
| "epoch": 0.695, | |
| "grad_norm": 0.9026237726211548, | |
| "learning_rate": 2.3355408892704424e-05, | |
| "loss": 1.789105682373047, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 0.69625, | |
| "grad_norm": 0.9898682236671448, | |
| "learning_rate": 2.3180744697986086e-05, | |
| "loss": 1.9098811340332031, | |
| "step": 13925 | |
| }, | |
| { | |
| "epoch": 0.6975, | |
| "grad_norm": 0.9056411981582642, | |
| "learning_rate": 2.3006538770344032e-05, | |
| "loss": 1.8874795532226563, | |
| "step": 13950 | |
| }, | |
| { | |
| "epoch": 0.69875, | |
| "grad_norm": 1.1745656728744507, | |
| "learning_rate": 2.283279408647689e-05, | |
| "loss": 1.7552346801757812, | |
| "step": 13975 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 1.1037813425064087, | |
| "learning_rate": 2.265951361520195e-05, | |
| "loss": 1.70399658203125, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_cer": 0.37631791973539797, | |
| "eval_loss": 1.388388991355896, | |
| "eval_runtime": 236.4134, | |
| "eval_samples_per_second": 8.46, | |
| "eval_steps_per_second": 0.266, | |
| "eval_wer": 0.6578853229762878, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.70125, | |
| "grad_norm": 0.8479313850402832, | |
| "learning_rate": 2.2486700317404343e-05, | |
| "loss": 1.6616452026367188, | |
| "step": 14025 | |
| }, | |
| { | |
| "epoch": 0.7025, | |
| "grad_norm": 1.3055832386016846, | |
| "learning_rate": 2.2314357145986552e-05, | |
| "loss": 1.7221562194824218, | |
| "step": 14050 | |
| }, | |
| { | |
| "epoch": 0.70375, | |
| "grad_norm": 1.0889098644256592, | |
| "learning_rate": 2.2142487045817833e-05, | |
| "loss": 1.74091552734375, | |
| "step": 14075 | |
| }, | |
| { | |
| "epoch": 0.705, | |
| "grad_norm": 1.1302812099456787, | |
| "learning_rate": 2.1971092953684026e-05, | |
| "loss": 1.7212419128417968, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 0.70625, | |
| "grad_norm": 1.6330608129501343, | |
| "learning_rate": 2.180017779823723e-05, | |
| "loss": 1.8517782592773437, | |
| "step": 14125 | |
| }, | |
| { | |
| "epoch": 0.7075, | |
| "grad_norm": 1.7381510734558105, | |
| "learning_rate": 2.162974449994593e-05, | |
| "loss": 1.7808000183105468, | |
| "step": 14150 | |
| }, | |
| { | |
| "epoch": 0.70875, | |
| "grad_norm": 1.47407066822052, | |
| "learning_rate": 2.145979597104491e-05, | |
| "loss": 1.7250691223144532, | |
| "step": 14175 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 1.228254795074463, | |
| "learning_rate": 2.129033511548566e-05, | |
| "loss": 1.7454202270507813, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 0.71125, | |
| "grad_norm": 1.0281528234481812, | |
| "learning_rate": 2.112136482888663e-05, | |
| "loss": 1.7192562866210936, | |
| "step": 14225 | |
| }, | |
| { | |
| "epoch": 0.7125, | |
| "grad_norm": 1.3011813163757324, | |
| "learning_rate": 2.095288799848379e-05, | |
| "loss": 1.657694091796875, | |
| "step": 14250 | |
| }, | |
| { | |
| "epoch": 0.71375, | |
| "grad_norm": 1.226781964302063, | |
| "learning_rate": 2.078490750308136e-05, | |
| "loss": 1.7552859497070312, | |
| "step": 14275 | |
| }, | |
| { | |
| "epoch": 0.715, | |
| "grad_norm": 1.1064605712890625, | |
| "learning_rate": 2.0617426213002506e-05, | |
| "loss": 1.6327297973632813, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 0.71625, | |
| "grad_norm": 1.4926190376281738, | |
| "learning_rate": 2.0450446990040412e-05, | |
| "loss": 1.6457408142089844, | |
| "step": 14325 | |
| }, | |
| { | |
| "epoch": 0.7175, | |
| "grad_norm": 1.2489418983459473, | |
| "learning_rate": 2.0283972687409247e-05, | |
| "loss": 1.6671981811523438, | |
| "step": 14350 | |
| }, | |
| { | |
| "epoch": 0.71875, | |
| "grad_norm": 1.1891186237335205, | |
| "learning_rate": 2.0118006149695563e-05, | |
| "loss": 1.712366943359375, | |
| "step": 14375 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 1.0904715061187744, | |
| "learning_rate": 1.995255021280954e-05, | |
| "loss": 1.6890667724609374, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 0.72125, | |
| "grad_norm": 1.1137077808380127, | |
| "learning_rate": 1.978760770393666e-05, | |
| "loss": 1.8110963439941405, | |
| "step": 14425 | |
| }, | |
| { | |
| "epoch": 0.7225, | |
| "grad_norm": 1.2658946514129639, | |
| "learning_rate": 1.962318144148928e-05, | |
| "loss": 1.71916259765625, | |
| "step": 14450 | |
| }, | |
| { | |
| "epoch": 0.72375, | |
| "grad_norm": 1.4173874855041504, | |
| "learning_rate": 1.945927423505859e-05, | |
| "loss": 1.7222340393066407, | |
| "step": 14475 | |
| }, | |
| { | |
| "epoch": 0.725, | |
| "grad_norm": 0.9890522956848145, | |
| "learning_rate": 1.929588888536647e-05, | |
| "loss": 1.7548519897460937, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.72625, | |
| "grad_norm": 1.276092767715454, | |
| "learning_rate": 1.9133028184217795e-05, | |
| "loss": 1.8688414001464844, | |
| "step": 14525 | |
| }, | |
| { | |
| "epoch": 0.7275, | |
| "grad_norm": 1.1105726957321167, | |
| "learning_rate": 1.897069491445258e-05, | |
| "loss": 1.7756446838378905, | |
| "step": 14550 | |
| }, | |
| { | |
| "epoch": 0.72875, | |
| "grad_norm": 1.1034342050552368, | |
| "learning_rate": 1.880889184989849e-05, | |
| "loss": 1.6495399475097656, | |
| "step": 14575 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 1.4259778261184692, | |
| "learning_rate": 1.8647621755323513e-05, | |
| "loss": 1.7023257446289062, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 0.73125, | |
| "grad_norm": 1.3064128160476685, | |
| "learning_rate": 1.848688738638857e-05, | |
| "loss": 1.760780029296875, | |
| "step": 14625 | |
| }, | |
| { | |
| "epoch": 0.7325, | |
| "grad_norm": 1.0586611032485962, | |
| "learning_rate": 1.832669148960057e-05, | |
| "loss": 1.709996337890625, | |
| "step": 14650 | |
| }, | |
| { | |
| "epoch": 0.73375, | |
| "grad_norm": 1.2333015203475952, | |
| "learning_rate": 1.8167036802265358e-05, | |
| "loss": 1.707982940673828, | |
| "step": 14675 | |
| }, | |
| { | |
| "epoch": 0.735, | |
| "grad_norm": 1.2188390493392944, | |
| "learning_rate": 1.800792605244109e-05, | |
| "loss": 1.71076904296875, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 0.73625, | |
| "grad_norm": 1.0349574089050293, | |
| "learning_rate": 1.7849361958891426e-05, | |
| "loss": 1.7050798034667969, | |
| "step": 14725 | |
| }, | |
| { | |
| "epoch": 0.7375, | |
| "grad_norm": 0.9727810025215149, | |
| "learning_rate": 1.7691347231039275e-05, | |
| "loss": 1.668623809814453, | |
| "step": 14750 | |
| }, | |
| { | |
| "epoch": 0.73875, | |
| "grad_norm": 1.195213794708252, | |
| "learning_rate": 1.7533884568920322e-05, | |
| "loss": 1.7399934387207032, | |
| "step": 14775 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 1.4273074865341187, | |
| "learning_rate": 1.7376976663137047e-05, | |
| "loss": 1.7789492797851563, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 0.74125, | |
| "grad_norm": 1.2793517112731934, | |
| "learning_rate": 1.72206261948126e-05, | |
| "loss": 1.7787957763671876, | |
| "step": 14825 | |
| }, | |
| { | |
| "epoch": 0.7425, | |
| "grad_norm": 1.3987388610839844, | |
| "learning_rate": 1.706483583554513e-05, | |
| "loss": 1.8119566345214844, | |
| "step": 14850 | |
| }, | |
| { | |
| "epoch": 0.74375, | |
| "grad_norm": 1.2237513065338135, | |
| "learning_rate": 1.690960824736202e-05, | |
| "loss": 1.86758544921875, | |
| "step": 14875 | |
| }, | |
| { | |
| "epoch": 0.745, | |
| "grad_norm": 1.6185275316238403, | |
| "learning_rate": 1.6754946082674444e-05, | |
| "loss": 1.799109649658203, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 0.74625, | |
| "grad_norm": 1.1656967401504517, | |
| "learning_rate": 1.6600851984232113e-05, | |
| "loss": 1.7074261474609376, | |
| "step": 14925 | |
| }, | |
| { | |
| "epoch": 0.7475, | |
| "grad_norm": 1.0714638233184814, | |
| "learning_rate": 1.644732858507797e-05, | |
| "loss": 1.6924021911621094, | |
| "step": 14950 | |
| }, | |
| { | |
| "epoch": 0.74875, | |
| "grad_norm": 1.5699371099472046, | |
| "learning_rate": 1.6294378508503356e-05, | |
| "loss": 1.638306884765625, | |
| "step": 14975 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 1.1188901662826538, | |
| "learning_rate": 1.614200436800304e-05, | |
| "loss": 1.5820968627929688, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "eval_cer": 0.37558788035300134, | |
| "eval_loss": 1.3598655462265015, | |
| "eval_runtime": 216.0532, | |
| "eval_samples_per_second": 9.257, | |
| "eval_steps_per_second": 0.292, | |
| "eval_wer": 0.6587540883074408, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.75125, | |
| "grad_norm": 1.4203516244888306, | |
| "learning_rate": 1.5990208767230704e-05, | |
| "loss": 1.6576242065429687, | |
| "step": 15025 | |
| }, | |
| { | |
| "epoch": 0.7525, | |
| "grad_norm": 1.2917144298553467, | |
| "learning_rate": 1.583899429995431e-05, | |
| "loss": 1.7396479797363282, | |
| "step": 15050 | |
| }, | |
| { | |
| "epoch": 0.75375, | |
| "grad_norm": 1.2932450771331787, | |
| "learning_rate": 1.568836355001193e-05, | |
| "loss": 1.8346478271484374, | |
| "step": 15075 | |
| }, | |
| { | |
| "epoch": 0.755, | |
| "grad_norm": 1.0382546186447144, | |
| "learning_rate": 1.553831909126744e-05, | |
| "loss": 2.438699493408203, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 0.75625, | |
| "grad_norm": 1.1095161437988281, | |
| "learning_rate": 1.538886348756668e-05, | |
| "loss": 1.7449612426757812, | |
| "step": 15125 | |
| }, | |
| { | |
| "epoch": 0.7575, | |
| "grad_norm": 1.4348697662353516, | |
| "learning_rate": 1.5239999292693524e-05, | |
| "loss": 1.6115505981445313, | |
| "step": 15150 | |
| }, | |
| { | |
| "epoch": 0.75875, | |
| "grad_norm": 1.2121785879135132, | |
| "learning_rate": 1.5091729050326375e-05, | |
| "loss": 1.7153553771972656, | |
| "step": 15175 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 1.2374008893966675, | |
| "learning_rate": 1.4944055293994551e-05, | |
| "loss": 1.6517231750488282, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 0.76125, | |
| "grad_norm": 1.177483320236206, | |
| "learning_rate": 1.4796980547035105e-05, | |
| "loss": 1.7158457946777343, | |
| "step": 15225 | |
| }, | |
| { | |
| "epoch": 0.7625, | |
| "grad_norm": 1.102067470550537, | |
| "learning_rate": 1.4650507322549684e-05, | |
| "loss": 1.8409674072265625, | |
| "step": 15250 | |
| }, | |
| { | |
| "epoch": 0.76375, | |
| "grad_norm": 1.2028313875198364, | |
| "learning_rate": 1.4504638123361542e-05, | |
| "loss": 1.6845973205566407, | |
| "step": 15275 | |
| }, | |
| { | |
| "epoch": 0.765, | |
| "grad_norm": 1.4570106267929077, | |
| "learning_rate": 1.4359375441972844e-05, | |
| "loss": 1.7015574645996094, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 0.76625, | |
| "grad_norm": 1.247830867767334, | |
| "learning_rate": 1.4214721760522003e-05, | |
| "loss": 1.6135427856445312, | |
| "step": 15325 | |
| }, | |
| { | |
| "epoch": 0.7675, | |
| "grad_norm": 1.198233723640442, | |
| "learning_rate": 1.407067955074135e-05, | |
| "loss": 1.6854086303710938, | |
| "step": 15350 | |
| }, | |
| { | |
| "epoch": 0.76875, | |
| "grad_norm": 1.4148765802383423, | |
| "learning_rate": 1.3927251273914792e-05, | |
| "loss": 1.7009700012207032, | |
| "step": 15375 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 1.2017967700958252, | |
| "learning_rate": 1.3784439380835879e-05, | |
| "loss": 1.7337741088867187, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 0.77125, | |
| "grad_norm": 1.0875744819641113, | |
| "learning_rate": 1.36422463117658e-05, | |
| "loss": 1.5848745727539062, | |
| "step": 15425 | |
| }, | |
| { | |
| "epoch": 0.7725, | |
| "grad_norm": 1.1721950769424438, | |
| "learning_rate": 1.3500674496391814e-05, | |
| "loss": 1.6446456909179688, | |
| "step": 15450 | |
| }, | |
| { | |
| "epoch": 0.77375, | |
| "grad_norm": 1.02508544921875, | |
| "learning_rate": 1.33597263537856e-05, | |
| "loss": 1.804439239501953, | |
| "step": 15475 | |
| }, | |
| { | |
| "epoch": 0.775, | |
| "grad_norm": 1.2351601123809814, | |
| "learning_rate": 1.3219404292362065e-05, | |
| "loss": 1.699874267578125, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.77625, | |
| "grad_norm": 0.9497158527374268, | |
| "learning_rate": 1.3079710709838038e-05, | |
| "loss": 1.6298439025878906, | |
| "step": 15525 | |
| }, | |
| { | |
| "epoch": 0.7775, | |
| "grad_norm": 1.103649377822876, | |
| "learning_rate": 1.2940647993191457e-05, | |
| "loss": 1.691442413330078, | |
| "step": 15550 | |
| }, | |
| { | |
| "epoch": 0.77875, | |
| "grad_norm": 0.8841238021850586, | |
| "learning_rate": 1.280221851862044e-05, | |
| "loss": 1.7583200073242187, | |
| "step": 15575 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 1.4351425170898438, | |
| "learning_rate": 1.2664424651502755e-05, | |
| "loss": 1.72744140625, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 0.78125, | |
| "grad_norm": 1.3280922174453735, | |
| "learning_rate": 1.2527268746355431e-05, | |
| "loss": 1.8163755798339845, | |
| "step": 15625 | |
| }, | |
| { | |
| "epoch": 0.7825, | |
| "grad_norm": 1.3615809679031372, | |
| "learning_rate": 1.2390753146794437e-05, | |
| "loss": 1.6660487365722656, | |
| "step": 15650 | |
| }, | |
| { | |
| "epoch": 0.78375, | |
| "grad_norm": 1.5741857290267944, | |
| "learning_rate": 1.2254880185494732e-05, | |
| "loss": 1.5826646423339843, | |
| "step": 15675 | |
| }, | |
| { | |
| "epoch": 0.785, | |
| "grad_norm": 1.3928883075714111, | |
| "learning_rate": 1.211965218415032e-05, | |
| "loss": 1.638283233642578, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 0.78625, | |
| "grad_norm": 1.5434808731079102, | |
| "learning_rate": 1.198507145343466e-05, | |
| "loss": 1.5825973510742188, | |
| "step": 15725 | |
| }, | |
| { | |
| "epoch": 0.7875, | |
| "grad_norm": 1.2133275270462036, | |
| "learning_rate": 1.1851140292961088e-05, | |
| "loss": 1.7104928588867188, | |
| "step": 15750 | |
| }, | |
| { | |
| "epoch": 0.78875, | |
| "grad_norm": 1.4504660367965698, | |
| "learning_rate": 1.171786099124364e-05, | |
| "loss": 1.6134979248046875, | |
| "step": 15775 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 1.4318393468856812, | |
| "learning_rate": 1.158523582565782e-05, | |
| "loss": 1.5858187866210938, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 0.79125, | |
| "grad_norm": 1.1338497400283813, | |
| "learning_rate": 1.1453267062401812e-05, | |
| "loss": 1.7505503845214845, | |
| "step": 15825 | |
| }, | |
| { | |
| "epoch": 0.7925, | |
| "grad_norm": 1.3158544301986694, | |
| "learning_rate": 1.1321956956457646e-05, | |
| "loss": 1.7246611022949219, | |
| "step": 15850 | |
| }, | |
| { | |
| "epoch": 0.79375, | |
| "grad_norm": 1.23163902759552, | |
| "learning_rate": 1.1191307751552737e-05, | |
| "loss": 1.6535879516601562, | |
| "step": 15875 | |
| }, | |
| { | |
| "epoch": 0.795, | |
| "grad_norm": 1.2866789102554321, | |
| "learning_rate": 1.106132168012155e-05, | |
| "loss": 1.7222795104980468, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 0.79625, | |
| "grad_norm": 1.5627771615982056, | |
| "learning_rate": 1.0932000963267363e-05, | |
| "loss": 1.7361250305175782, | |
| "step": 15925 | |
| }, | |
| { | |
| "epoch": 0.7975, | |
| "grad_norm": 1.0179498195648193, | |
| "learning_rate": 1.0803347810724452e-05, | |
| "loss": 1.7269871520996094, | |
| "step": 15950 | |
| }, | |
| { | |
| "epoch": 0.79875, | |
| "grad_norm": 1.1671440601348877, | |
| "learning_rate": 1.06753644208202e-05, | |
| "loss": 1.6840296936035157, | |
| "step": 15975 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.9704040884971619, | |
| "learning_rate": 1.0548052980437645e-05, | |
| "loss": 1.572103271484375, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_cer": 0.3728266429610795, | |
| "eval_loss": 1.3480268716812134, | |
| "eval_runtime": 215.5049, | |
| "eval_samples_per_second": 9.281, | |
| "eval_steps_per_second": 0.292, | |
| "eval_wer": 0.6519061733442355, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.80125, | |
| "grad_norm": 1.2414360046386719, | |
| "learning_rate": 1.0421415664978019e-05, | |
| "loss": 1.7525442504882813, | |
| "step": 16025 | |
| }, | |
| { | |
| "epoch": 0.8025, | |
| "grad_norm": 1.1043179035186768, | |
| "learning_rate": 1.0295454638323666e-05, | |
| "loss": 1.6157144165039063, | |
| "step": 16050 | |
| }, | |
| { | |
| "epoch": 0.80375, | |
| "grad_norm": 1.4600468873977661, | |
| "learning_rate": 1.0170172052800959e-05, | |
| "loss": 1.6451036071777343, | |
| "step": 16075 | |
| }, | |
| { | |
| "epoch": 0.805, | |
| "grad_norm": 1.6918089389801025, | |
| "learning_rate": 1.004557004914365e-05, | |
| "loss": 1.7464239501953125, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 0.80625, | |
| "grad_norm": 1.0915247201919556, | |
| "learning_rate": 9.921650756456164e-06, | |
| "loss": 1.679139404296875, | |
| "step": 16125 | |
| }, | |
| { | |
| "epoch": 0.8075, | |
| "grad_norm": 0.8941418528556824, | |
| "learning_rate": 9.798416292177337e-06, | |
| "loss": 1.6927119445800782, | |
| "step": 16150 | |
| }, | |
| { | |
| "epoch": 0.80875, | |
| "grad_norm": 1.1518326997756958, | |
| "learning_rate": 9.675868762044122e-06, | |
| "loss": 1.614804229736328, | |
| "step": 16175 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 1.5816043615341187, | |
| "learning_rate": 9.554010260055713e-06, | |
| "loss": 1.6708433532714844, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 0.81125, | |
| "grad_norm": 2.2120110988616943, | |
| "learning_rate": 9.43284286843768e-06, | |
| "loss": 1.6525192260742188, | |
| "step": 16225 | |
| }, | |
| { | |
| "epoch": 0.8125, | |
| "grad_norm": 1.2603182792663574, | |
| "learning_rate": 9.312368657606412e-06, | |
| "loss": 1.6440431213378905, | |
| "step": 16250 | |
| }, | |
| { | |
| "epoch": 0.81375, | |
| "grad_norm": 1.708512783050537, | |
| "learning_rate": 9.192589686133812e-06, | |
| "loss": 1.6378236389160157, | |
| "step": 16275 | |
| }, | |
| { | |
| "epoch": 0.815, | |
| "grad_norm": 1.0281175374984741, | |
| "learning_rate": 9.073508000711983e-06, | |
| "loss": 1.6738160705566407, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 0.81625, | |
| "grad_norm": 1.44629967212677, | |
| "learning_rate": 8.955125636118406e-06, | |
| "loss": 1.6968197631835937, | |
| "step": 16325 | |
| }, | |
| { | |
| "epoch": 0.8175, | |
| "grad_norm": 1.1274943351745605, | |
| "learning_rate": 8.837444615181029e-06, | |
| "loss": 1.61760498046875, | |
| "step": 16350 | |
| }, | |
| { | |
| "epoch": 0.81875, | |
| "grad_norm": 2.11869215965271, | |
| "learning_rate": 8.720466948743838e-06, | |
| "loss": 1.6444927978515624, | |
| "step": 16375 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 1.5141431093215942, | |
| "learning_rate": 8.604194635632373e-06, | |
| "loss": 1.6296272277832031, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 0.82125, | |
| "grad_norm": 1.4636927843093872, | |
| "learning_rate": 8.48862966261969e-06, | |
| "loss": 1.7262242126464844, | |
| "step": 16425 | |
| }, | |
| { | |
| "epoch": 0.8225, | |
| "grad_norm": 1.3123044967651367, | |
| "learning_rate": 8.373774004392293e-06, | |
| "loss": 1.8966561889648437, | |
| "step": 16450 | |
| }, | |
| { | |
| "epoch": 0.82375, | |
| "grad_norm": 1.0209351778030396, | |
| "learning_rate": 8.259629623516501e-06, | |
| "loss": 1.8558387756347656, | |
| "step": 16475 | |
| }, | |
| { | |
| "epoch": 0.825, | |
| "grad_norm": 1.9075822830200195, | |
| "learning_rate": 8.146198470404843e-06, | |
| "loss": 1.81407958984375, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.82625, | |
| "grad_norm": 1.4140610694885254, | |
| "learning_rate": 8.033482483282762e-06, | |
| "loss": 1.7187783813476563, | |
| "step": 16525 | |
| }, | |
| { | |
| "epoch": 0.8275, | |
| "grad_norm": 1.4152733087539673, | |
| "learning_rate": 7.92148358815547e-06, | |
| "loss": 1.8056149291992187, | |
| "step": 16550 | |
| }, | |
| { | |
| "epoch": 0.82875, | |
| "grad_norm": 1.3810051679611206, | |
| "learning_rate": 7.810203698775093e-06, | |
| "loss": 1.7580088806152343, | |
| "step": 16575 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 1.0717037916183472, | |
| "learning_rate": 7.699644716607895e-06, | |
| "loss": 1.66962646484375, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 0.83125, | |
| "grad_norm": 1.2882596254348755, | |
| "learning_rate": 7.5898085308018285e-06, | |
| "loss": 1.722180633544922, | |
| "step": 16625 | |
| }, | |
| { | |
| "epoch": 0.8325, | |
| "grad_norm": 1.2467365264892578, | |
| "learning_rate": 7.480697018154286e-06, | |
| "loss": 1.628616943359375, | |
| "step": 16650 | |
| }, | |
| { | |
| "epoch": 0.83375, | |
| "grad_norm": 1.1477034091949463, | |
| "learning_rate": 7.372312043079943e-06, | |
| "loss": 1.5673805236816407, | |
| "step": 16675 | |
| }, | |
| { | |
| "epoch": 0.835, | |
| "grad_norm": 1.2944529056549072, | |
| "learning_rate": 7.264655457579e-06, | |
| "loss": 1.7042782592773438, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 0.83625, | |
| "grad_norm": 1.4483259916305542, | |
| "learning_rate": 7.157729101205452e-06, | |
| "loss": 1.723576202392578, | |
| "step": 16725 | |
| }, | |
| { | |
| "epoch": 0.8375, | |
| "grad_norm": 1.2930117845535278, | |
| "learning_rate": 7.051534801035725e-06, | |
| "loss": 1.6988841247558595, | |
| "step": 16750 | |
| }, | |
| { | |
| "epoch": 0.83875, | |
| "grad_norm": 2.3856797218322754, | |
| "learning_rate": 6.94607437163739e-06, | |
| "loss": 1.7573260498046874, | |
| "step": 16775 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 1.2205097675323486, | |
| "learning_rate": 6.8413496150382394e-06, | |
| "loss": 1.7281820678710937, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 0.84125, | |
| "grad_norm": 1.1864348649978638, | |
| "learning_rate": 6.737362320695384e-06, | |
| "loss": 1.6891987609863282, | |
| "step": 16825 | |
| }, | |
| { | |
| "epoch": 0.8425, | |
| "grad_norm": 1.104697346687317, | |
| "learning_rate": 6.634114265464803e-06, | |
| "loss": 1.7520260620117187, | |
| "step": 16850 | |
| }, | |
| { | |
| "epoch": 0.84375, | |
| "grad_norm": 1.4675954580307007, | |
| "learning_rate": 6.531607213570861e-06, | |
| "loss": 1.6536885070800782, | |
| "step": 16875 | |
| }, | |
| { | |
| "epoch": 0.845, | |
| "grad_norm": 1.331076979637146, | |
| "learning_rate": 6.429842916576279e-06, | |
| "loss": 1.7051416015625, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 0.84625, | |
| "grad_norm": 1.1980607509613037, | |
| "learning_rate": 6.3288231133521e-06, | |
| "loss": 1.7295429992675782, | |
| "step": 16925 | |
| }, | |
| { | |
| "epoch": 0.8475, | |
| "grad_norm": 1.0040066242218018, | |
| "learning_rate": 6.228549530048022e-06, | |
| "loss": 1.7798513793945312, | |
| "step": 16950 | |
| }, | |
| { | |
| "epoch": 0.84875, | |
| "grad_norm": 1.06290864944458, | |
| "learning_rate": 6.129023880062951e-06, | |
| "loss": 1.7142842102050782, | |
| "step": 16975 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 1.1379823684692383, | |
| "learning_rate": 6.0302478640156145e-06, | |
| "loss": 1.6634608459472657, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "eval_cer": 0.373248774576751, | |
| "eval_loss": 1.343159794807434, | |
| "eval_runtime": 223.74, | |
| "eval_samples_per_second": 8.939, | |
| "eval_steps_per_second": 0.282, | |
| "eval_wer": 0.6527493867538839, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.85125, | |
| "grad_norm": 1.147297978401184, | |
| "learning_rate": 5.932223169715623e-06, | |
| "loss": 1.6187948608398437, | |
| "step": 17025 | |
| }, | |
| { | |
| "epoch": 0.8525, | |
| "grad_norm": 1.2166903018951416, | |
| "learning_rate": 5.834951472134514e-06, | |
| "loss": 1.6052098083496094, | |
| "step": 17050 | |
| }, | |
| { | |
| "epoch": 0.85375, | |
| "grad_norm": 1.2225483655929565, | |
| "learning_rate": 5.738434433377243e-06, | |
| "loss": 1.6365573120117187, | |
| "step": 17075 | |
| }, | |
| { | |
| "epoch": 0.855, | |
| "grad_norm": 1.4318076372146606, | |
| "learning_rate": 5.642673702653683e-06, | |
| "loss": 1.70055419921875, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 0.85625, | |
| "grad_norm": 1.1895360946655273, | |
| "learning_rate": 5.5476709162505305e-06, | |
| "loss": 1.7699407958984374, | |
| "step": 17125 | |
| }, | |
| { | |
| "epoch": 0.8575, | |
| "grad_norm": 1.3633129596710205, | |
| "learning_rate": 5.453427697503255e-06, | |
| "loss": 1.7118313598632813, | |
| "step": 17150 | |
| }, | |
| { | |
| "epoch": 0.85875, | |
| "grad_norm": 1.0259132385253906, | |
| "learning_rate": 5.359945656768467e-06, | |
| "loss": 1.7853955078125, | |
| "step": 17175 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 1.2191566228866577, | |
| "learning_rate": 5.267226391396296e-06, | |
| "loss": 1.8883969116210937, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 0.86125, | |
| "grad_norm": 1.1236616373062134, | |
| "learning_rate": 5.175271485703187e-06, | |
| "loss": 1.7460699462890625, | |
| "step": 17225 | |
| }, | |
| { | |
| "epoch": 0.8625, | |
| "grad_norm": 1.0902246236801147, | |
| "learning_rate": 5.084082510944749e-06, | |
| "loss": 1.6689810180664062, | |
| "step": 17250 | |
| }, | |
| { | |
| "epoch": 0.86375, | |
| "grad_norm": 1.423059344291687, | |
| "learning_rate": 4.993661025288959e-06, | |
| "loss": 1.7480343627929686, | |
| "step": 17275 | |
| }, | |
| { | |
| "epoch": 0.865, | |
| "grad_norm": 1.4627045392990112, | |
| "learning_rate": 4.904008573789548e-06, | |
| "loss": 1.747147216796875, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 0.86625, | |
| "grad_norm": 1.5385794639587402, | |
| "learning_rate": 4.815126688359522e-06, | |
| "loss": 1.707109375, | |
| "step": 17325 | |
| }, | |
| { | |
| "epoch": 0.8675, | |
| "grad_norm": 1.1001863479614258, | |
| "learning_rate": 4.727016887745095e-06, | |
| "loss": 1.7542826843261718, | |
| "step": 17350 | |
| }, | |
| { | |
| "epoch": 0.86875, | |
| "grad_norm": 1.3000284433364868, | |
| "learning_rate": 4.639680677499642e-06, | |
| "loss": 1.72843505859375, | |
| "step": 17375 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 1.055073857307434, | |
| "learning_rate": 4.553119549958035e-06, | |
| "loss": 1.819022216796875, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 0.87125, | |
| "grad_norm": 1.3849037885665894, | |
| "learning_rate": 4.467334984211102e-06, | |
| "loss": 1.7520419311523439, | |
| "step": 17425 | |
| }, | |
| { | |
| "epoch": 0.8725, | |
| "grad_norm": 1.2385969161987305, | |
| "learning_rate": 4.3823284460804025e-06, | |
| "loss": 1.7134429931640625, | |
| "step": 17450 | |
| }, | |
| { | |
| "epoch": 0.87375, | |
| "grad_norm": 0.9723396301269531, | |
| "learning_rate": 4.298101388093118e-06, | |
| "loss": 1.6656005859375, | |
| "step": 17475 | |
| }, | |
| { | |
| "epoch": 0.875, | |
| "grad_norm": 1.5931142568588257, | |
| "learning_rate": 4.214655249457284e-06, | |
| "loss": 1.6508407592773438, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.87625, | |
| "grad_norm": 1.1754169464111328, | |
| "learning_rate": 4.131991456037171e-06, | |
| "loss": 1.7136148071289063, | |
| "step": 17525 | |
| }, | |
| { | |
| "epoch": 0.8775, | |
| "grad_norm": 1.2141945362091064, | |
| "learning_rate": 4.050111420328939e-06, | |
| "loss": 1.8041830444335938, | |
| "step": 17550 | |
| }, | |
| { | |
| "epoch": 0.87875, | |
| "grad_norm": 1.0505759716033936, | |
| "learning_rate": 3.969016541436476e-06, | |
| "loss": 1.763004913330078, | |
| "step": 17575 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 1.0453979969024658, | |
| "learning_rate": 3.888708205047509e-06, | |
| "loss": 1.7114891052246093, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 0.88125, | |
| "grad_norm": 1.0407336950302124, | |
| "learning_rate": 3.8091877834099455e-06, | |
| "loss": 1.64724365234375, | |
| "step": 17625 | |
| }, | |
| { | |
| "epoch": 0.8825, | |
| "grad_norm": 1.174785852432251, | |
| "learning_rate": 3.7304566353083658e-06, | |
| "loss": 1.6655709838867188, | |
| "step": 17650 | |
| }, | |
| { | |
| "epoch": 0.88375, | |
| "grad_norm": 1.1461564302444458, | |
| "learning_rate": 3.6525161060408875e-06, | |
| "loss": 1.6740159606933593, | |
| "step": 17675 | |
| }, | |
| { | |
| "epoch": 0.885, | |
| "grad_norm": 1.4013136625289917, | |
| "learning_rate": 3.575367527396084e-06, | |
| "loss": 1.6717343139648437, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 0.88625, | |
| "grad_norm": 1.3019026517868042, | |
| "learning_rate": 3.499012217630332e-06, | |
| "loss": 1.6947007751464844, | |
| "step": 17725 | |
| }, | |
| { | |
| "epoch": 0.8875, | |
| "grad_norm": 1.7853046655654907, | |
| "learning_rate": 3.4234514814451836e-06, | |
| "loss": 1.7914909362792968, | |
| "step": 17750 | |
| }, | |
| { | |
| "epoch": 0.88875, | |
| "grad_norm": 1.379899501800537, | |
| "learning_rate": 3.348686609965157e-06, | |
| "loss": 1.7445704650878906, | |
| "step": 17775 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 1.8439922332763672, | |
| "learning_rate": 3.2747188807155993e-06, | |
| "loss": 1.7227520751953125, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 0.89125, | |
| "grad_norm": 0.9510113596916199, | |
| "learning_rate": 3.201549557600947e-06, | |
| "loss": 1.7174844360351562, | |
| "step": 17825 | |
| }, | |
| { | |
| "epoch": 0.8925, | |
| "grad_norm": 1.3749337196350098, | |
| "learning_rate": 3.1291798908830273e-06, | |
| "loss": 1.7721444702148437, | |
| "step": 17850 | |
| }, | |
| { | |
| "epoch": 0.89375, | |
| "grad_norm": 1.2029380798339844, | |
| "learning_rate": 3.057611117159792e-06, | |
| "loss": 1.760584716796875, | |
| "step": 17875 | |
| }, | |
| { | |
| "epoch": 0.895, | |
| "grad_norm": 1.3879034519195557, | |
| "learning_rate": 2.9868444593440957e-06, | |
| "loss": 1.6648373413085937, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 0.89625, | |
| "grad_norm": 2.085763692855835, | |
| "learning_rate": 2.91688112664289e-06, | |
| "loss": 1.6733528137207032, | |
| "step": 17925 | |
| }, | |
| { | |
| "epoch": 0.8975, | |
| "grad_norm": 1.299283504486084, | |
| "learning_rate": 2.847722314536483e-06, | |
| "loss": 1.6162384033203125, | |
| "step": 17950 | |
| }, | |
| { | |
| "epoch": 0.89875, | |
| "grad_norm": 1.5633270740509033, | |
| "learning_rate": 2.7793692047581586e-06, | |
| "loss": 1.624728240966797, | |
| "step": 17975 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 1.3784313201904297, | |
| "learning_rate": 2.7118229652739747e-06, | |
| "loss": 1.655706024169922, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "eval_cer": 0.37475354963026236, | |
| "eval_loss": 1.350075602531433, | |
| "eval_runtime": 218.1843, | |
| "eval_samples_per_second": 9.167, | |
| "eval_steps_per_second": 0.289, | |
| "eval_wer": 0.6524938675388389, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.90125, | |
| "grad_norm": 1.0730280876159668, | |
| "learning_rate": 2.6450847502627884e-06, | |
| "loss": 1.6174893188476562, | |
| "step": 18025 | |
| }, | |
| { | |
| "epoch": 0.9025, | |
| "grad_norm": 1.5405020713806152, | |
| "learning_rate": 2.579155700096575e-06, | |
| "loss": 1.6413883972167969, | |
| "step": 18050 | |
| }, | |
| { | |
| "epoch": 0.90375, | |
| "grad_norm": 1.1906652450561523, | |
| "learning_rate": 2.5140369413208743e-06, | |
| "loss": 1.6562530517578125, | |
| "step": 18075 | |
| }, | |
| { | |
| "epoch": 0.905, | |
| "grad_norm": 1.449783444404602, | |
| "learning_rate": 2.4497295866356296e-06, | |
| "loss": 1.6450199890136719, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 0.90625, | |
| "grad_norm": 1.4530644416809082, | |
| "learning_rate": 2.3862347348760804e-06, | |
| "loss": 1.6267514038085937, | |
| "step": 18125 | |
| }, | |
| { | |
| "epoch": 0.9075, | |
| "grad_norm": 1.7504948377609253, | |
| "learning_rate": 2.3235534709940665e-06, | |
| "loss": 1.7457493591308593, | |
| "step": 18150 | |
| }, | |
| { | |
| "epoch": 0.90875, | |
| "grad_norm": 1.5501203536987305, | |
| "learning_rate": 2.261686866039442e-06, | |
| "loss": 1.672449188232422, | |
| "step": 18175 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 1.2946059703826904, | |
| "learning_rate": 2.200635977141796e-06, | |
| "loss": 1.7002857971191405, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 0.91125, | |
| "grad_norm": 1.1127476692199707, | |
| "learning_rate": 2.1404018474923628e-06, | |
| "loss": 1.7305221557617188, | |
| "step": 18225 | |
| }, | |
| { | |
| "epoch": 0.9125, | |
| "grad_norm": 1.534924030303955, | |
| "learning_rate": 2.0809855063262273e-06, | |
| "loss": 1.7657546997070312, | |
| "step": 18250 | |
| }, | |
| { | |
| "epoch": 0.91375, | |
| "grad_norm": 1.0926239490509033, | |
| "learning_rate": 2.0223879689047433e-06, | |
| "loss": 1.6709626770019532, | |
| "step": 18275 | |
| }, | |
| { | |
| "epoch": 0.915, | |
| "grad_norm": 1.290826439857483, | |
| "learning_rate": 1.9646102364981266e-06, | |
| "loss": 1.71489013671875, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 0.91625, | |
| "grad_norm": 1.1290301084518433, | |
| "learning_rate": 1.9076532963684224e-06, | |
| "loss": 1.5985736083984374, | |
| "step": 18325 | |
| }, | |
| { | |
| "epoch": 0.9175, | |
| "grad_norm": 1.4048781394958496, | |
| "learning_rate": 1.8515181217525824e-06, | |
| "loss": 1.617904815673828, | |
| "step": 18350 | |
| }, | |
| { | |
| "epoch": 0.91875, | |
| "grad_norm": 1.1143537759780884, | |
| "learning_rate": 1.7962056718458632e-06, | |
| "loss": 1.6652728271484376, | |
| "step": 18375 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 1.091009497642517, | |
| "learning_rate": 1.7417168917854165e-06, | |
| "loss": 1.6322047424316406, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 0.92125, | |
| "grad_norm": 1.9022687673568726, | |
| "learning_rate": 1.6880527126341638e-06, | |
| "loss": 1.7680978393554687, | |
| "step": 18425 | |
| }, | |
| { | |
| "epoch": 0.9225, | |
| "grad_norm": 1.1953799724578857, | |
| "learning_rate": 1.6352140513648417e-06, | |
| "loss": 1.8073098754882813, | |
| "step": 18450 | |
| }, | |
| { | |
| "epoch": 0.92375, | |
| "grad_norm": 1.0084136724472046, | |
| "learning_rate": 1.5832018108444092e-06, | |
| "loss": 1.8635337829589844, | |
| "step": 18475 | |
| }, | |
| { | |
| "epoch": 0.925, | |
| "grad_norm": 1.2110964059829712, | |
| "learning_rate": 1.532016879818532e-06, | |
| "loss": 1.749373779296875, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.92625, | |
| "grad_norm": 1.3123623132705688, | |
| "learning_rate": 1.481660132896473e-06, | |
| "loss": 1.6708609008789062, | |
| "step": 18525 | |
| }, | |
| { | |
| "epoch": 0.9275, | |
| "grad_norm": 1.1336241960525513, | |
| "learning_rate": 1.432132430536076e-06, | |
| "loss": 1.7288685607910157, | |
| "step": 18550 | |
| }, | |
| { | |
| "epoch": 0.92875, | |
| "grad_norm": 1.2451077699661255, | |
| "learning_rate": 1.3834346190291448e-06, | |
| "loss": 1.7669725036621093, | |
| "step": 18575 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 1.4456207752227783, | |
| "learning_rate": 1.3355675304869086e-06, | |
| "loss": 1.7303396606445312, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 0.93125, | |
| "grad_norm": 1.4477598667144775, | |
| "learning_rate": 1.2885319828258357e-06, | |
| "loss": 1.8188290405273437, | |
| "step": 18625 | |
| }, | |
| { | |
| "epoch": 0.9325, | |
| "grad_norm": 1.234950065612793, | |
| "learning_rate": 1.2423287797536654e-06, | |
| "loss": 1.74578857421875, | |
| "step": 18650 | |
| }, | |
| { | |
| "epoch": 0.93375, | |
| "grad_norm": 1.2804648876190186, | |
| "learning_rate": 1.1969587107556524e-06, | |
| "loss": 1.7694448852539062, | |
| "step": 18675 | |
| }, | |
| { | |
| "epoch": 0.935, | |
| "grad_norm": 1.4425557851791382, | |
| "learning_rate": 1.1524225510811116e-06, | |
| "loss": 1.8393618774414062, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 0.93625, | |
| "grad_norm": 1.1845166683197021, | |
| "learning_rate": 1.1087210617301114e-06, | |
| "loss": 1.6897073364257813, | |
| "step": 18725 | |
| }, | |
| { | |
| "epoch": 0.9375, | |
| "grad_norm": 1.8836590051651, | |
| "learning_rate": 1.0658549894405456e-06, | |
| "loss": 1.7422689819335937, | |
| "step": 18750 | |
| }, | |
| { | |
| "epoch": 0.93875, | |
| "grad_norm": 1.3005623817443848, | |
| "learning_rate": 1.0238250666753046e-06, | |
| "loss": 1.5547370910644531, | |
| "step": 18775 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 1.5120162963867188, | |
| "learning_rate": 9.826320116098132e-07, | |
| "loss": 1.7445518493652343, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 0.94125, | |
| "grad_norm": 1.0026639699935913, | |
| "learning_rate": 9.422765281197232e-07, | |
| "loss": 1.7480233764648438, | |
| "step": 18825 | |
| }, | |
| { | |
| "epoch": 0.9425, | |
| "grad_norm": 1.2218565940856934, | |
| "learning_rate": 9.027593057689076e-07, | |
| "loss": 1.7591806030273438, | |
| "step": 18850 | |
| }, | |
| { | |
| "epoch": 0.94375, | |
| "grad_norm": 1.0779987573623657, | |
| "learning_rate": 8.640810197976579e-07, | |
| "loss": 1.8884674072265626, | |
| "step": 18875 | |
| }, | |
| { | |
| "epoch": 0.945, | |
| "grad_norm": 1.1344575881958008, | |
| "learning_rate": 8.262423311111711e-07, | |
| "loss": 1.8649324035644532, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 0.94625, | |
| "grad_norm": 1.1154437065124512, | |
| "learning_rate": 7.892438862682261e-07, | |
| "loss": 1.862205047607422, | |
| "step": 18925 | |
| }, | |
| { | |
| "epoch": 0.9475, | |
| "grad_norm": 0.9772124290466309, | |
| "learning_rate": 7.530863174701752e-07, | |
| "loss": 1.810330352783203, | |
| "step": 18950 | |
| }, | |
| { | |
| "epoch": 0.94875, | |
| "grad_norm": 0.781427264213562, | |
| "learning_rate": 7.177702425500976e-07, | |
| "loss": 1.8051651000976563, | |
| "step": 18975 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 1.2141427993774414, | |
| "learning_rate": 6.832962649622798e-07, | |
| "loss": 1.7991664123535156, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "eval_cer": 0.37249390392284426, | |
| "eval_loss": 1.3333868980407715, | |
| "eval_runtime": 211.2761, | |
| "eval_samples_per_second": 9.466, | |
| "eval_steps_per_second": 0.298, | |
| "eval_wer": 0.6492743254292723, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.95125, | |
| "grad_norm": 1.0704618692398071, | |
| "learning_rate": 6.496649737718796e-07, | |
| "loss": 1.704537353515625, | |
| "step": 19025 | |
| }, | |
| { | |
| "epoch": 0.9525, | |
| "grad_norm": 1.1346384286880493, | |
| "learning_rate": 6.168769436448673e-07, | |
| "loss": 1.7066073608398438, | |
| "step": 19050 | |
| }, | |
| { | |
| "epoch": 0.95375, | |
| "grad_norm": 0.9572947025299072, | |
| "learning_rate": 5.849327348382228e-07, | |
| "loss": 1.7915672302246093, | |
| "step": 19075 | |
| }, | |
| { | |
| "epoch": 0.955, | |
| "grad_norm": 0.8648141026496887, | |
| "learning_rate": 5.538328931903259e-07, | |
| "loss": 1.7891120910644531, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 0.95625, | |
| "grad_norm": 1.3389594554901123, | |
| "learning_rate": 5.2357795011167e-07, | |
| "loss": 1.6836885070800782, | |
| "step": 19125 | |
| }, | |
| { | |
| "epoch": 0.9575, | |
| "grad_norm": 1.1919796466827393, | |
| "learning_rate": 4.941684225757526e-07, | |
| "loss": 1.7464437866210938, | |
| "step": 19150 | |
| }, | |
| { | |
| "epoch": 0.95875, | |
| "grad_norm": 1.043595314025879, | |
| "learning_rate": 4.6560481311025974e-07, | |
| "loss": 1.7690708923339844, | |
| "step": 19175 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 1.039276123046875, | |
| "learning_rate": 4.378876097884621e-07, | |
| "loss": 1.8303811645507813, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 0.96125, | |
| "grad_norm": 1.0408260822296143, | |
| "learning_rate": 4.1101728622089384e-07, | |
| "loss": 1.7924273681640626, | |
| "step": 19225 | |
| }, | |
| { | |
| "epoch": 0.9625, | |
| "grad_norm": 1.0000412464141846, | |
| "learning_rate": 3.849943015472479e-07, | |
| "loss": 1.8378707885742187, | |
| "step": 19250 | |
| }, | |
| { | |
| "epoch": 0.96375, | |
| "grad_norm": 0.9676679968833923, | |
| "learning_rate": 3.598191004285323e-07, | |
| "loss": 1.8996051025390626, | |
| "step": 19275 | |
| }, | |
| { | |
| "epoch": 0.965, | |
| "grad_norm": 1.1249455213546753, | |
| "learning_rate": 3.354921130394706e-07, | |
| "loss": 1.8261538696289064, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 0.96625, | |
| "grad_norm": 1.0636478662490845, | |
| "learning_rate": 3.120137550611579e-07, | |
| "loss": 1.8052239990234376, | |
| "step": 19325 | |
| }, | |
| { | |
| "epoch": 0.9675, | |
| "grad_norm": 0.9989639520645142, | |
| "learning_rate": 2.893844276739499e-07, | |
| "loss": 1.8380400085449218, | |
| "step": 19350 | |
| }, | |
| { | |
| "epoch": 0.96875, | |
| "grad_norm": 0.8790108561515808, | |
| "learning_rate": 2.676045175506237e-07, | |
| "loss": 1.750625762939453, | |
| "step": 19375 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 4.106517791748047, | |
| "learning_rate": 2.4667439684974423e-07, | |
| "loss": 1.7750701904296875, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 0.97125, | |
| "grad_norm": 1.1543933153152466, | |
| "learning_rate": 2.2659442320932534e-07, | |
| "loss": 1.7718186950683594, | |
| "step": 19425 | |
| }, | |
| { | |
| "epoch": 0.9725, | |
| "grad_norm": 1.0098258256912231, | |
| "learning_rate": 2.0736493974071736e-07, | |
| "loss": 1.8053097534179687, | |
| "step": 19450 | |
| }, | |
| { | |
| "epoch": 0.97375, | |
| "grad_norm": 1.089660882949829, | |
| "learning_rate": 1.8898627502273448e-07, | |
| "loss": 1.7882542419433594, | |
| "step": 19475 | |
| }, | |
| { | |
| "epoch": 0.975, | |
| "grad_norm": 1.056114673614502, | |
| "learning_rate": 1.7145874309604792e-07, | |
| "loss": 1.7408137512207031, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.97625, | |
| "grad_norm": 1.0057376623153687, | |
| "learning_rate": 1.5478264345782368e-07, | |
| "loss": 1.735123748779297, | |
| "step": 19525 | |
| }, | |
| { | |
| "epoch": 0.9775, | |
| "grad_norm": 1.1148862838745117, | |
| "learning_rate": 1.389582610565876e-07, | |
| "loss": 1.800911865234375, | |
| "step": 19550 | |
| }, | |
| { | |
| "epoch": 0.97875, | |
| "grad_norm": 1.095569133758545, | |
| "learning_rate": 1.2398586628737385e-07, | |
| "loss": 1.800084228515625, | |
| "step": 19575 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 0.8736165761947632, | |
| "learning_rate": 1.0986571498710074e-07, | |
| "loss": 1.8828060913085938, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 0.98125, | |
| "grad_norm": 1.0789073705673218, | |
| "learning_rate": 9.659804843019648e-08, | |
| "loss": 1.9125828552246094, | |
| "step": 19625 | |
| }, | |
| { | |
| "epoch": 0.9825, | |
| "grad_norm": 1.0751843452453613, | |
| "learning_rate": 8.418309332447471e-08, | |
| "loss": 1.8400242614746094, | |
| "step": 19650 | |
| }, | |
| { | |
| "epoch": 0.98375, | |
| "grad_norm": 0.9620322585105896, | |
| "learning_rate": 7.262106180726535e-08, | |
| "loss": 1.821525115966797, | |
| "step": 19675 | |
| }, | |
| { | |
| "epoch": 0.985, | |
| "grad_norm": 1.219050407409668, | |
| "learning_rate": 6.191215144178419e-08, | |
| "loss": 1.7966372680664062, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 0.98625, | |
| "grad_norm": 1.0730938911437988, | |
| "learning_rate": 5.2056545213763354e-08, | |
| "loss": 1.8384300231933595, | |
| "step": 19725 | |
| }, | |
| { | |
| "epoch": 0.9875, | |
| "grad_norm": 1.1275181770324707, | |
| "learning_rate": 4.305441152831491e-08, | |
| "loss": 1.7794735717773438, | |
| "step": 19750 | |
| }, | |
| { | |
| "epoch": 0.98875, | |
| "grad_norm": 1.2597466707229614, | |
| "learning_rate": 3.490590420707762e-08, | |
| "loss": 1.7876008605957032, | |
| "step": 19775 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 1.1383274793624878, | |
| "learning_rate": 2.76111624855524e-08, | |
| "loss": 1.8828434753417969, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 0.99125, | |
| "grad_norm": 1.0261435508728027, | |
| "learning_rate": 2.1170311010748623e-08, | |
| "loss": 1.79291259765625, | |
| "step": 19825 | |
| }, | |
| { | |
| "epoch": 0.9925, | |
| "grad_norm": 1.016748309135437, | |
| "learning_rate": 1.5583459839046964e-08, | |
| "loss": 1.843469696044922, | |
| "step": 19850 | |
| }, | |
| { | |
| "epoch": 0.99375, | |
| "grad_norm": 1.3169704675674438, | |
| "learning_rate": 1.0850704434317572e-08, | |
| "loss": 1.7497860717773437, | |
| "step": 19875 | |
| }, | |
| { | |
| "epoch": 0.995, | |
| "grad_norm": 0.9610996246337891, | |
| "learning_rate": 6.972125666299123e-09, | |
| "loss": 1.7596157836914061, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 0.99625, | |
| "grad_norm": 0.9990308880805969, | |
| "learning_rate": 3.947789809194414e-09, | |
| "loss": 1.7010501098632813, | |
| "step": 19925 | |
| }, | |
| { | |
| "epoch": 0.9975, | |
| "grad_norm": 0.8817222118377686, | |
| "learning_rate": 1.7777485405601203e-09, | |
| "loss": 1.7759243774414062, | |
| "step": 19950 | |
| }, | |
| { | |
| "epoch": 0.99875, | |
| "grad_norm": 1.117990255355835, | |
| "learning_rate": 4.620389404186298e-10, | |
| "loss": 1.7576724243164064, | |
| "step": 19975 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 1.1643476486206055, | |
| "learning_rate": 6.834906085551041e-13, | |
| "loss": 1.7090470886230469, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_cer": 0.37243430887122003, | |
| "eval_loss": 1.3328464031219482, | |
| "eval_runtime": 216.0709, | |
| "eval_samples_per_second": 9.256, | |
| "eval_steps_per_second": 0.292, | |
| "eval_wer": 0.6499386753883892, | |
| "step": 20000 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 20000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "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": 3.867001152205912e+21, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |