Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
msp_audio
Generated from Trainer
custom_code
Instructions to use MahmoodAnaam/MSP-ASR with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use MahmoodAnaam/MSP-ASR with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="MahmoodAnaam/MSP-ASR", trust_remote_code=True)# Load model directly from transformers import AutoModelForCTC model = AutoModelForCTC.from_pretrained("MahmoodAnaam/MSP-ASR", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
| { | |
| "best_global_step": 17000, | |
| "best_metric": 0.2039707879188227, | |
| "best_model_checkpoint": "/workspace/checkpoints/msp_asr/checkpoint-17000", | |
| "epoch": 1.0, | |
| "eval_steps": 1000, | |
| "global_step": 40000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.000625, | |
| "grad_norm": 5.267065525054932, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 4.323406982421875, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.00125, | |
| "grad_norm": 3.8583974838256836, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 4.264261779785156, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.001875, | |
| "grad_norm": 3.3579821586608887, | |
| "learning_rate": 7.4e-06, | |
| "loss": 4.044583129882812, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0025, | |
| "grad_norm": 2.367401361465454, | |
| "learning_rate": 9.900000000000002e-06, | |
| "loss": 3.791492919921875, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.003125, | |
| "grad_norm": 2.434986114501953, | |
| "learning_rate": 1.24e-05, | |
| "loss": 3.6091009521484376, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.00375, | |
| "grad_norm": 2.1168673038482666, | |
| "learning_rate": 1.49e-05, | |
| "loss": 3.6343560791015626, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.004375, | |
| "grad_norm": 2.473984479904175, | |
| "learning_rate": 1.74e-05, | |
| "loss": 3.5921762084960935, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.005, | |
| "grad_norm": 1.7123606204986572, | |
| "learning_rate": 1.9900000000000003e-05, | |
| "loss": 3.4881491088867187, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.005625, | |
| "grad_norm": 2.603245973587036, | |
| "learning_rate": 2.2400000000000002e-05, | |
| "loss": 3.484436340332031, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.00625, | |
| "grad_norm": 2.6283440589904785, | |
| "learning_rate": 2.4900000000000002e-05, | |
| "loss": 3.3572451782226564, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.006875, | |
| "grad_norm": 3.7932605743408203, | |
| "learning_rate": 2.7400000000000002e-05, | |
| "loss": 3.3410797119140625, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.0075, | |
| "grad_norm": 2.569643497467041, | |
| "learning_rate": 2.9900000000000002e-05, | |
| "loss": 3.3989813232421877, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.008125, | |
| "grad_norm": 2.44470477104187, | |
| "learning_rate": 3.24e-05, | |
| "loss": 3.4272494506835938, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.00875, | |
| "grad_norm": 4.493594169616699, | |
| "learning_rate": 3.49e-05, | |
| "loss": 3.2092889404296874, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.009375, | |
| "grad_norm": 2.4080135822296143, | |
| "learning_rate": 3.74e-05, | |
| "loss": 3.2225802612304686, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 3.196730375289917, | |
| "learning_rate": 3.99e-05, | |
| "loss": 3.251342468261719, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.010625, | |
| "grad_norm": 3.765761375427246, | |
| "learning_rate": 4.24e-05, | |
| "loss": 3.421331787109375, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.01125, | |
| "grad_norm": 3.0019078254699707, | |
| "learning_rate": 4.49e-05, | |
| "loss": 3.247855224609375, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.011875, | |
| "grad_norm": 3.268728256225586, | |
| "learning_rate": 4.74e-05, | |
| "loss": 2.974139404296875, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.0125, | |
| "grad_norm": 2.966991901397705, | |
| "learning_rate": 4.99e-05, | |
| "loss": 3.136421813964844, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.013125, | |
| "grad_norm": 2.091470241546631, | |
| "learning_rate": 5.2400000000000007e-05, | |
| "loss": 2.9975115966796877, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.01375, | |
| "grad_norm": 2.0409600734710693, | |
| "learning_rate": 5.4900000000000006e-05, | |
| "loss": 2.9733895874023437, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.014375, | |
| "grad_norm": 3.4123849868774414, | |
| "learning_rate": 5.74e-05, | |
| "loss": 2.943175964355469, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.015, | |
| "grad_norm": 2.6855201721191406, | |
| "learning_rate": 5.99e-05, | |
| "loss": 3.1415582275390626, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.015625, | |
| "grad_norm": 2.6875081062316895, | |
| "learning_rate": 6.24e-05, | |
| "loss": 3.185292053222656, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.01625, | |
| "grad_norm": 3.4700987339019775, | |
| "learning_rate": 6.49e-05, | |
| "loss": 3.0090676879882814, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.016875, | |
| "grad_norm": 2.850792646408081, | |
| "learning_rate": 6.740000000000001e-05, | |
| "loss": 3.0521066284179685, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.0175, | |
| "grad_norm": 3.1411142349243164, | |
| "learning_rate": 6.99e-05, | |
| "loss": 2.9812518310546876, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.018125, | |
| "grad_norm": 2.2294530868530273, | |
| "learning_rate": 7.24e-05, | |
| "loss": 2.95865966796875, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.01875, | |
| "grad_norm": 3.497708797454834, | |
| "learning_rate": 7.49e-05, | |
| "loss": 3.053566589355469, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.019375, | |
| "grad_norm": 2.695732593536377, | |
| "learning_rate": 7.740000000000001e-05, | |
| "loss": 2.9046136474609376, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 2.7966887950897217, | |
| "learning_rate": 7.99e-05, | |
| "loss": 2.958673400878906, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.020625, | |
| "grad_norm": 3.018155574798584, | |
| "learning_rate": 8.24e-05, | |
| "loss": 2.743119812011719, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.02125, | |
| "grad_norm": 3.1703360080718994, | |
| "learning_rate": 8.49e-05, | |
| "loss": 2.84368896484375, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.021875, | |
| "grad_norm": 4.187642574310303, | |
| "learning_rate": 8.740000000000001e-05, | |
| "loss": 2.9081005859375, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.0225, | |
| "grad_norm": 2.7985174655914307, | |
| "learning_rate": 8.99e-05, | |
| "loss": 3.0562255859375, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.023125, | |
| "grad_norm": 2.9112160205841064, | |
| "learning_rate": 9.240000000000001e-05, | |
| "loss": 2.9217318725585937, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.02375, | |
| "grad_norm": 3.2347748279571533, | |
| "learning_rate": 9.49e-05, | |
| "loss": 2.8495794677734376, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.024375, | |
| "grad_norm": 3.2869646549224854, | |
| "learning_rate": 9.74e-05, | |
| "loss": 3.0750582885742186, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.025, | |
| "grad_norm": 3.9301090240478516, | |
| "learning_rate": 9.99e-05, | |
| "loss": 2.9990963745117187, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.025, | |
| "eval_loss": 0.7600362300872803, | |
| "eval_runtime": 20.7898, | |
| "eval_samples_per_second": 70.034, | |
| "eval_steps_per_second": 4.377, | |
| "eval_wer": 0.42731843370398914, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.025625, | |
| "grad_norm": 3.2349042892456055, | |
| "learning_rate": 9.999990655998743e-05, | |
| "loss": 2.9402279663085937, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.02625, | |
| "grad_norm": 2.61457896232605, | |
| "learning_rate": 9.999961050477646e-05, | |
| "loss": 2.9998614501953127, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.026875, | |
| "grad_norm": 1.9102178812026978, | |
| "learning_rate": 9.999911167334542e-05, | |
| "loss": 3.1586492919921874, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.0275, | |
| "grad_norm": 2.3976991176605225, | |
| "learning_rate": 9.999841006771737e-05, | |
| "loss": 2.975727844238281, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.028125, | |
| "grad_norm": 2.706573963165283, | |
| "learning_rate": 9.999750569073771e-05, | |
| "loss": 2.7479977416992187, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.02875, | |
| "grad_norm": 3.162564516067505, | |
| "learning_rate": 9.999639854607419e-05, | |
| "loss": 2.6489334106445312, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.029375, | |
| "grad_norm": 3.254704475402832, | |
| "learning_rate": 9.999508863821689e-05, | |
| "loss": 2.6830172729492188, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 1.984910011291504, | |
| "learning_rate": 9.999357597247825e-05, | |
| "loss": 2.794486083984375, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.030625, | |
| "grad_norm": 2.546842336654663, | |
| "learning_rate": 9.999186055499292e-05, | |
| "loss": 2.9072146606445313, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.03125, | |
| "grad_norm": 3.029797315597534, | |
| "learning_rate": 9.99899423927179e-05, | |
| "loss": 2.964591064453125, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.031875, | |
| "grad_norm": 2.4892096519470215, | |
| "learning_rate": 9.99878214934324e-05, | |
| "loss": 2.7791043090820313, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.0325, | |
| "grad_norm": 2.67838978767395, | |
| "learning_rate": 9.998549786573785e-05, | |
| "loss": 2.712342529296875, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.033125, | |
| "grad_norm": 2.953914165496826, | |
| "learning_rate": 9.998297151905785e-05, | |
| "loss": 2.7532568359375, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.03375, | |
| "grad_norm": 4.061647891998291, | |
| "learning_rate": 9.998024246363813e-05, | |
| "loss": 2.7488946533203125, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.034375, | |
| "grad_norm": 2.453488826751709, | |
| "learning_rate": 9.997731071054653e-05, | |
| "loss": 2.6597412109375, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.035, | |
| "grad_norm": 2.5735793113708496, | |
| "learning_rate": 9.997417627167296e-05, | |
| "loss": 2.611119689941406, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.035625, | |
| "grad_norm": 2.86007022857666, | |
| "learning_rate": 9.997083915972929e-05, | |
| "loss": 2.744207458496094, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.03625, | |
| "grad_norm": 3.4510135650634766, | |
| "learning_rate": 9.996729938824935e-05, | |
| "loss": 2.64264892578125, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.036875, | |
| "grad_norm": 2.464531183242798, | |
| "learning_rate": 9.996355697158893e-05, | |
| "loss": 2.7909088134765625, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.0375, | |
| "grad_norm": 2.5669782161712646, | |
| "learning_rate": 9.995961192492558e-05, | |
| "loss": 2.6939956665039064, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.038125, | |
| "grad_norm": 5.631814479827881, | |
| "learning_rate": 9.995546426425866e-05, | |
| "loss": 2.7690692138671875, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.03875, | |
| "grad_norm": 2.090285062789917, | |
| "learning_rate": 9.995111400640922e-05, | |
| "loss": 2.6084750366210936, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.039375, | |
| "grad_norm": 2.9554991722106934, | |
| "learning_rate": 9.994656116902002e-05, | |
| "loss": 2.691404113769531, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 2.8547191619873047, | |
| "learning_rate": 9.99418057705553e-05, | |
| "loss": 2.608034362792969, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.040625, | |
| "grad_norm": 3.4367880821228027, | |
| "learning_rate": 9.993684783030088e-05, | |
| "loss": 2.744490966796875, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.04125, | |
| "grad_norm": 2.45200514793396, | |
| "learning_rate": 9.993168736836394e-05, | |
| "loss": 2.6521151733398436, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.041875, | |
| "grad_norm": 3.7969329357147217, | |
| "learning_rate": 9.992632440567304e-05, | |
| "loss": 2.6562158203125, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.0425, | |
| "grad_norm": 4.674492835998535, | |
| "learning_rate": 9.992075896397797e-05, | |
| "loss": 2.4041812133789064, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.043125, | |
| "grad_norm": 2.933290719985962, | |
| "learning_rate": 9.99149910658497e-05, | |
| "loss": 2.653353576660156, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.04375, | |
| "grad_norm": 5.6183342933654785, | |
| "learning_rate": 9.990902073468023e-05, | |
| "loss": 2.6427932739257813, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.044375, | |
| "grad_norm": 2.1874606609344482, | |
| "learning_rate": 9.990284799468262e-05, | |
| "loss": 2.522238311767578, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.045, | |
| "grad_norm": 3.279247760772705, | |
| "learning_rate": 9.98964728708907e-05, | |
| "loss": 2.683359375, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.045625, | |
| "grad_norm": 2.838366985321045, | |
| "learning_rate": 9.988989538915921e-05, | |
| "loss": 2.5724777221679687, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.04625, | |
| "grad_norm": 3.5663628578186035, | |
| "learning_rate": 9.988311557616346e-05, | |
| "loss": 2.684295654296875, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.046875, | |
| "grad_norm": 3.5662167072296143, | |
| "learning_rate": 9.987613345939935e-05, | |
| "loss": 2.7449139404296874, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.0475, | |
| "grad_norm": 3.681104898452759, | |
| "learning_rate": 9.986894906718328e-05, | |
| "loss": 2.6223516845703125, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.048125, | |
| "grad_norm": 3.008888006210327, | |
| "learning_rate": 9.986156242865192e-05, | |
| "loss": 2.566004333496094, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.04875, | |
| "grad_norm": 4.016690731048584, | |
| "learning_rate": 9.985397357376222e-05, | |
| "loss": 2.7363925170898438, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.049375, | |
| "grad_norm": 3.150728940963745, | |
| "learning_rate": 9.984618253329121e-05, | |
| "loss": 2.5361758422851564, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 3.397801160812378, | |
| "learning_rate": 9.983818933883586e-05, | |
| "loss": 2.7153076171875, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_loss": 0.5967366695404053, | |
| "eval_runtime": 15.9007, | |
| "eval_samples_per_second": 91.568, | |
| "eval_steps_per_second": 5.723, | |
| "eval_wer": 0.3696649418327278, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.050625, | |
| "grad_norm": 2.6956193447113037, | |
| "learning_rate": 9.982999402281306e-05, | |
| "loss": 2.545854339599609, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.05125, | |
| "grad_norm": 3.7128403186798096, | |
| "learning_rate": 9.982159661845934e-05, | |
| "loss": 2.6736044311523437, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.051875, | |
| "grad_norm": 3.2885775566101074, | |
| "learning_rate": 9.981299715983083e-05, | |
| "loss": 2.7458767700195312, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.0525, | |
| "grad_norm": 2.152069330215454, | |
| "learning_rate": 9.980419568180317e-05, | |
| "loss": 2.6045245361328124, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.053125, | |
| "grad_norm": 4.13241720199585, | |
| "learning_rate": 9.979519222007122e-05, | |
| "loss": 2.603431396484375, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.05375, | |
| "grad_norm": 2.568535327911377, | |
| "learning_rate": 9.978598681114898e-05, | |
| "loss": 2.567033386230469, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.054375, | |
| "grad_norm": 2.317216396331787, | |
| "learning_rate": 9.977657949236958e-05, | |
| "loss": 2.5691043090820314, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.055, | |
| "grad_norm": 3.7757654190063477, | |
| "learning_rate": 9.976697030188487e-05, | |
| "loss": 2.6709893798828124, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.055625, | |
| "grad_norm": 3.3160698413848877, | |
| "learning_rate": 9.97571592786655e-05, | |
| "loss": 2.5459054565429686, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.05625, | |
| "grad_norm": 2.971742630004883, | |
| "learning_rate": 9.974714646250057e-05, | |
| "loss": 2.728532409667969, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.056875, | |
| "grad_norm": 2.3057808876037598, | |
| "learning_rate": 9.973693189399766e-05, | |
| "loss": 2.4765863037109375, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.0575, | |
| "grad_norm": 2.3339619636535645, | |
| "learning_rate": 9.972651561458252e-05, | |
| "loss": 2.6263897705078123, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.058125, | |
| "grad_norm": 2.1193151473999023, | |
| "learning_rate": 9.971589766649896e-05, | |
| "loss": 2.466733093261719, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.05875, | |
| "grad_norm": 2.4383885860443115, | |
| "learning_rate": 9.970507809280865e-05, | |
| "loss": 2.599173889160156, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.059375, | |
| "grad_norm": 4.8630146980285645, | |
| "learning_rate": 9.969405693739096e-05, | |
| "loss": 2.4229290771484373, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 2.679741382598877, | |
| "learning_rate": 9.968283424494283e-05, | |
| "loss": 2.6939743041992186, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.060625, | |
| "grad_norm": 2.5089759826660156, | |
| "learning_rate": 9.967141006097851e-05, | |
| "loss": 2.314402313232422, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.06125, | |
| "grad_norm": 2.9203526973724365, | |
| "learning_rate": 9.965978443182942e-05, | |
| "loss": 2.536326446533203, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.061875, | |
| "grad_norm": 2.504847764968872, | |
| "learning_rate": 9.964795740464394e-05, | |
| "loss": 2.6262823486328126, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.0625, | |
| "grad_norm": 2.6516757011413574, | |
| "learning_rate": 9.963592902738725e-05, | |
| "loss": 2.5988543701171873, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.063125, | |
| "grad_norm": 2.7327418327331543, | |
| "learning_rate": 9.96236993488411e-05, | |
| "loss": 2.4847898864746094, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.06375, | |
| "grad_norm": 4.1733317375183105, | |
| "learning_rate": 9.961126841860366e-05, | |
| "loss": 2.4733262634277344, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.064375, | |
| "grad_norm": 2.833759307861328, | |
| "learning_rate": 9.959863628708922e-05, | |
| "loss": 2.4042843627929686, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.065, | |
| "grad_norm": 2.7223546504974365, | |
| "learning_rate": 9.958580300552815e-05, | |
| "loss": 2.612850341796875, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.065625, | |
| "grad_norm": 3.192142963409424, | |
| "learning_rate": 9.957276862596649e-05, | |
| "loss": 2.4569342041015627, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.06625, | |
| "grad_norm": 4.608920574188232, | |
| "learning_rate": 9.955953320126594e-05, | |
| "loss": 2.5547445678710936, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.066875, | |
| "grad_norm": 2.791745901107788, | |
| "learning_rate": 9.954609678510348e-05, | |
| "loss": 2.4209219360351564, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.0675, | |
| "grad_norm": 3.272300958633423, | |
| "learning_rate": 9.953245943197124e-05, | |
| "loss": 2.5017831420898435, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.068125, | |
| "grad_norm": 2.2691471576690674, | |
| "learning_rate": 9.951862119717629e-05, | |
| "loss": 2.4805477905273436, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.06875, | |
| "grad_norm": 3.448873996734619, | |
| "learning_rate": 9.950458213684035e-05, | |
| "loss": 2.445795440673828, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.069375, | |
| "grad_norm": 2.4772186279296875, | |
| "learning_rate": 9.949034230789965e-05, | |
| "loss": 2.4605558776855467, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 3.1712827682495117, | |
| "learning_rate": 9.947590176810456e-05, | |
| "loss": 2.5019020080566405, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.070625, | |
| "grad_norm": 3.1829514503479004, | |
| "learning_rate": 9.946126057601954e-05, | |
| "loss": 2.496224060058594, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.07125, | |
| "grad_norm": 2.0898146629333496, | |
| "learning_rate": 9.944641879102274e-05, | |
| "loss": 2.393035125732422, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.071875, | |
| "grad_norm": 2.449113607406616, | |
| "learning_rate": 9.943137647330588e-05, | |
| "loss": 2.6191220092773437, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.0725, | |
| "grad_norm": 1.8356566429138184, | |
| "learning_rate": 9.941613368387388e-05, | |
| "loss": 2.4733572387695313, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.073125, | |
| "grad_norm": 2.4358699321746826, | |
| "learning_rate": 9.940069048454476e-05, | |
| "loss": 2.5051329040527346, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.07375, | |
| "grad_norm": 2.275167942047119, | |
| "learning_rate": 9.938504693794927e-05, | |
| "loss": 2.4410516357421876, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.074375, | |
| "grad_norm": 2.6384241580963135, | |
| "learning_rate": 9.936920310753069e-05, | |
| "loss": 2.6117279052734377, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.075, | |
| "grad_norm": 1.924433708190918, | |
| "learning_rate": 9.935315905754454e-05, | |
| "loss": 2.446084442138672, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.075, | |
| "eval_loss": 0.5057567358016968, | |
| "eval_runtime": 24.6182, | |
| "eval_samples_per_second": 59.143, | |
| "eval_steps_per_second": 3.696, | |
| "eval_wer": 0.3102499174281625, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.075625, | |
| "grad_norm": 3.383068561553955, | |
| "learning_rate": 9.933691485305838e-05, | |
| "loss": 2.472366943359375, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.07625, | |
| "grad_norm": 3.4558751583099365, | |
| "learning_rate": 9.932047055995151e-05, | |
| "loss": 2.446722412109375, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.076875, | |
| "grad_norm": 2.9713714122772217, | |
| "learning_rate": 9.930382624491466e-05, | |
| "loss": 2.51492431640625, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.0775, | |
| "grad_norm": 2.5073225498199463, | |
| "learning_rate": 9.928698197544979e-05, | |
| "loss": 2.368467254638672, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.078125, | |
| "grad_norm": 4.28424596786499, | |
| "learning_rate": 9.926993781986975e-05, | |
| "loss": 2.288444061279297, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.07875, | |
| "grad_norm": 3.2725813388824463, | |
| "learning_rate": 9.925269384729812e-05, | |
| "loss": 2.4385311889648436, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.079375, | |
| "grad_norm": 3.6250274181365967, | |
| "learning_rate": 9.923525012766875e-05, | |
| "loss": 2.3601338195800783, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 3.5715112686157227, | |
| "learning_rate": 9.921760673172563e-05, | |
| "loss": 2.399430847167969, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.080625, | |
| "grad_norm": 2.888333797454834, | |
| "learning_rate": 9.919976373102257e-05, | |
| "loss": 2.343405914306641, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.08125, | |
| "grad_norm": 2.665381908416748, | |
| "learning_rate": 9.918172119792282e-05, | |
| "loss": 2.4569857788085936, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.081875, | |
| "grad_norm": 3.7903621196746826, | |
| "learning_rate": 9.916347920559891e-05, | |
| "loss": 2.537921142578125, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.0825, | |
| "grad_norm": 3.530862331390381, | |
| "learning_rate": 9.914503782803224e-05, | |
| "loss": 2.478512878417969, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.083125, | |
| "grad_norm": 3.4290575981140137, | |
| "learning_rate": 9.912639714001288e-05, | |
| "loss": 2.1560601806640625, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.08375, | |
| "grad_norm": 2.380321502685547, | |
| "learning_rate": 9.910755721713914e-05, | |
| "loss": 2.4317338562011717, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.084375, | |
| "grad_norm": 2.160979986190796, | |
| "learning_rate": 9.908851813581744e-05, | |
| "loss": 2.41833984375, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.085, | |
| "grad_norm": 2.911205768585205, | |
| "learning_rate": 9.906927997326177e-05, | |
| "loss": 2.4843405151367186, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.085625, | |
| "grad_norm": 2.503948211669922, | |
| "learning_rate": 9.904984280749363e-05, | |
| "loss": 2.47994873046875, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.08625, | |
| "grad_norm": 3.9231438636779785, | |
| "learning_rate": 9.903020671734152e-05, | |
| "loss": 2.478351898193359, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.086875, | |
| "grad_norm": 3.6701557636260986, | |
| "learning_rate": 9.90103717824407e-05, | |
| "loss": 2.4145101928710937, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.0875, | |
| "grad_norm": 3.4353156089782715, | |
| "learning_rate": 9.899033808323286e-05, | |
| "loss": 2.195841522216797, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.088125, | |
| "grad_norm": 2.3643858432769775, | |
| "learning_rate": 9.897010570096578e-05, | |
| "loss": 2.410039520263672, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.08875, | |
| "grad_norm": 2.947376251220703, | |
| "learning_rate": 9.894967471769303e-05, | |
| "loss": 2.5055174255371093, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.089375, | |
| "grad_norm": 3.526736259460449, | |
| "learning_rate": 9.892904521627361e-05, | |
| "loss": 2.52310546875, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 2.7263927459716797, | |
| "learning_rate": 9.89082172803716e-05, | |
| "loss": 2.515272216796875, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.090625, | |
| "grad_norm": 3.6337671279907227, | |
| "learning_rate": 9.888719099445589e-05, | |
| "loss": 2.548531799316406, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.09125, | |
| "grad_norm": 4.070236682891846, | |
| "learning_rate": 9.886596644379973e-05, | |
| "loss": 2.4597044372558594, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.091875, | |
| "grad_norm": 3.784086227416992, | |
| "learning_rate": 9.884454371448048e-05, | |
| "loss": 2.5076600646972658, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.0925, | |
| "grad_norm": 1.9711685180664062, | |
| "learning_rate": 9.882292289337924e-05, | |
| "loss": 2.44762939453125, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.093125, | |
| "grad_norm": 2.515859365463257, | |
| "learning_rate": 9.880110406818044e-05, | |
| "loss": 2.3809681701660157, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.09375, | |
| "grad_norm": 3.016655206680298, | |
| "learning_rate": 9.877908732737155e-05, | |
| "loss": 2.3009237670898437, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.094375, | |
| "grad_norm": 1.8320847749710083, | |
| "learning_rate": 9.875687276024269e-05, | |
| "loss": 2.3537553405761718, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.095, | |
| "grad_norm": 2.744502067565918, | |
| "learning_rate": 9.87344604568863e-05, | |
| "loss": 2.421930847167969, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.095625, | |
| "grad_norm": 3.465440034866333, | |
| "learning_rate": 9.871185050819672e-05, | |
| "loss": 2.3701936340332033, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.09625, | |
| "grad_norm": 2.384808301925659, | |
| "learning_rate": 9.868904300586985e-05, | |
| "loss": 2.38617431640625, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.096875, | |
| "grad_norm": 2.5891613960266113, | |
| "learning_rate": 9.866603804240281e-05, | |
| "loss": 2.469194793701172, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.0975, | |
| "grad_norm": 1.9982811212539673, | |
| "learning_rate": 9.864283571109352e-05, | |
| "loss": 2.581448669433594, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.098125, | |
| "grad_norm": 1.974596619606018, | |
| "learning_rate": 9.86194361060403e-05, | |
| "loss": 2.4108615112304688, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.09875, | |
| "grad_norm": 4.334737300872803, | |
| "learning_rate": 9.859583932214161e-05, | |
| "loss": 2.4004356384277346, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.099375, | |
| "grad_norm": 2.642355442047119, | |
| "learning_rate": 9.85720454550955e-05, | |
| "loss": 2.462643737792969, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 2.3971734046936035, | |
| "learning_rate": 9.854805460139934e-05, | |
| "loss": 2.4391294860839845, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_loss": 0.5427843928337097, | |
| "eval_runtime": 25.3762, | |
| "eval_samples_per_second": 57.377, | |
| "eval_steps_per_second": 3.586, | |
| "eval_wer": 0.29942383206723183, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.100625, | |
| "grad_norm": 3.7295141220092773, | |
| "learning_rate": 9.852386685834936e-05, | |
| "loss": 2.464426727294922, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.10125, | |
| "grad_norm": 1.8876482248306274, | |
| "learning_rate": 9.849948232404032e-05, | |
| "loss": 2.5142250061035156, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.101875, | |
| "grad_norm": 3.429696559906006, | |
| "learning_rate": 9.847490109736505e-05, | |
| "loss": 2.328139190673828, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.1025, | |
| "grad_norm": 2.30229115486145, | |
| "learning_rate": 9.845012327801413e-05, | |
| "loss": 2.5297906494140623, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.103125, | |
| "grad_norm": 4.078639507293701, | |
| "learning_rate": 9.842514896647537e-05, | |
| "loss": 2.5450367736816406, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.10375, | |
| "grad_norm": 1.965324878692627, | |
| "learning_rate": 9.839997826403346e-05, | |
| "loss": 2.372446746826172, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.104375, | |
| "grad_norm": 2.7977564334869385, | |
| "learning_rate": 9.837461127276966e-05, | |
| "loss": 2.4538418579101564, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.105, | |
| "grad_norm": 2.232367515563965, | |
| "learning_rate": 9.834904809556116e-05, | |
| "loss": 2.3611822509765625, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.105625, | |
| "grad_norm": 1.6842498779296875, | |
| "learning_rate": 9.832328883608088e-05, | |
| "loss": 2.281151123046875, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.10625, | |
| "grad_norm": 3.627267360687256, | |
| "learning_rate": 9.829733359879695e-05, | |
| "loss": 2.281341552734375, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.106875, | |
| "grad_norm": 4.500143527984619, | |
| "learning_rate": 9.827118248897227e-05, | |
| "loss": 2.3601921081542967, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.1075, | |
| "grad_norm": 2.101828098297119, | |
| "learning_rate": 9.824483561266415e-05, | |
| "loss": 2.3109523010253907, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.108125, | |
| "grad_norm": 2.2386698722839355, | |
| "learning_rate": 9.82182930767238e-05, | |
| "loss": 2.478082733154297, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.10875, | |
| "grad_norm": 2.753462314605713, | |
| "learning_rate": 9.819155498879598e-05, | |
| "loss": 2.5288987731933594, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.109375, | |
| "grad_norm": 2.5756032466888428, | |
| "learning_rate": 9.816462145731848e-05, | |
| "loss": 2.3797843933105467, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 1.967148780822754, | |
| "learning_rate": 9.813749259152177e-05, | |
| "loss": 2.3148460388183594, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.110625, | |
| "grad_norm": 3.878065586090088, | |
| "learning_rate": 9.811016850142848e-05, | |
| "loss": 2.3127200317382814, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.11125, | |
| "grad_norm": 1.919490098953247, | |
| "learning_rate": 9.808264929785297e-05, | |
| "loss": 2.433041839599609, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.111875, | |
| "grad_norm": 3.641495943069458, | |
| "learning_rate": 9.805493509240092e-05, | |
| "loss": 2.292813720703125, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.1125, | |
| "grad_norm": 2.2465927600860596, | |
| "learning_rate": 9.802702599746884e-05, | |
| "loss": 2.141736602783203, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.113125, | |
| "grad_norm": 2.242436170578003, | |
| "learning_rate": 9.799892212624366e-05, | |
| "loss": 2.301976776123047, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.11375, | |
| "grad_norm": 1.8059154748916626, | |
| "learning_rate": 9.797062359270215e-05, | |
| "loss": 2.3053215026855467, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.114375, | |
| "grad_norm": 4.8942790031433105, | |
| "learning_rate": 9.794213051161062e-05, | |
| "loss": 2.3062982177734375, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.115, | |
| "grad_norm": 2.4718339443206787, | |
| "learning_rate": 9.791344299852436e-05, | |
| "loss": 2.252259063720703, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.115625, | |
| "grad_norm": 2.0137128829956055, | |
| "learning_rate": 9.788456116978719e-05, | |
| "loss": 2.1838677978515624, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.11625, | |
| "grad_norm": 1.859391689300537, | |
| "learning_rate": 9.785548514253098e-05, | |
| "loss": 2.2040748596191406, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.116875, | |
| "grad_norm": 6.748249530792236, | |
| "learning_rate": 9.782621503467517e-05, | |
| "loss": 2.329748229980469, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.1175, | |
| "grad_norm": 2.274712562561035, | |
| "learning_rate": 9.779675096492635e-05, | |
| "loss": 2.393338317871094, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.118125, | |
| "grad_norm": 3.3160722255706787, | |
| "learning_rate": 9.776709305277767e-05, | |
| "loss": 2.3510055541992188, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.11875, | |
| "grad_norm": 2.128993272781372, | |
| "learning_rate": 9.773724141850849e-05, | |
| "loss": 2.4330215454101562, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.119375, | |
| "grad_norm": 4.108203411102295, | |
| "learning_rate": 9.770719618318376e-05, | |
| "loss": 2.387606353759766, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 2.7292726039886475, | |
| "learning_rate": 9.767695746865362e-05, | |
| "loss": 2.3424223327636717, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.120625, | |
| "grad_norm": 2.1014602184295654, | |
| "learning_rate": 9.764652539755286e-05, | |
| "loss": 2.162779541015625, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.12125, | |
| "grad_norm": 2.08599853515625, | |
| "learning_rate": 9.761590009330048e-05, | |
| "loss": 2.425719757080078, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.121875, | |
| "grad_norm": 2.942275047302246, | |
| "learning_rate": 9.75850816800991e-05, | |
| "loss": 2.2239019775390627, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.1225, | |
| "grad_norm": 1.82079017162323, | |
| "learning_rate": 9.755407028293448e-05, | |
| "loss": 2.3144776916503904, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.123125, | |
| "grad_norm": 4.4813032150268555, | |
| "learning_rate": 9.752286602757514e-05, | |
| "loss": 2.4774703979492188, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.12375, | |
| "grad_norm": 2.5883078575134277, | |
| "learning_rate": 9.749146904057166e-05, | |
| "loss": 2.397513427734375, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.124375, | |
| "grad_norm": 3.360262393951416, | |
| "learning_rate": 9.745987944925627e-05, | |
| "loss": 2.372333831787109, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.125, | |
| "grad_norm": 2.2018544673919678, | |
| "learning_rate": 9.742809738174235e-05, | |
| "loss": 2.450827789306641, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.125, | |
| "eval_loss": 0.6064216494560242, | |
| "eval_runtime": 22.6426, | |
| "eval_samples_per_second": 64.304, | |
| "eval_steps_per_second": 4.019, | |
| "eval_wer": 0.35828837755513965, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.125625, | |
| "grad_norm": 2.2889022827148438, | |
| "learning_rate": 9.739612296692382e-05, | |
| "loss": 2.6975006103515624, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 0.12625, | |
| "grad_norm": 2.9140493869781494, | |
| "learning_rate": 9.736395633447473e-05, | |
| "loss": 2.405331573486328, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.126875, | |
| "grad_norm": 2.3265461921691895, | |
| "learning_rate": 9.733159761484866e-05, | |
| "loss": 2.236347503662109, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 0.1275, | |
| "grad_norm": 2.084930419921875, | |
| "learning_rate": 9.729904693927821e-05, | |
| "loss": 2.243023681640625, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.128125, | |
| "grad_norm": 2.2969279289245605, | |
| "learning_rate": 9.726630443977444e-05, | |
| "loss": 2.3645864868164064, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 0.12875, | |
| "grad_norm": 2.2135696411132812, | |
| "learning_rate": 9.723337024912644e-05, | |
| "loss": 2.2532118225097655, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.129375, | |
| "grad_norm": 2.8939309120178223, | |
| "learning_rate": 9.720024450090064e-05, | |
| "loss": 2.2750958251953124, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 1.8961058855056763, | |
| "learning_rate": 9.716692732944035e-05, | |
| "loss": 2.1728594970703123, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.130625, | |
| "grad_norm": 2.3602912425994873, | |
| "learning_rate": 9.713341886986525e-05, | |
| "loss": 2.4110421752929687, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 0.13125, | |
| "grad_norm": 1.9523662328720093, | |
| "learning_rate": 9.709971925807076e-05, | |
| "loss": 2.116759033203125, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.131875, | |
| "grad_norm": 3.5182929039001465, | |
| "learning_rate": 9.706582863072753e-05, | |
| "loss": 2.174735260009766, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 0.1325, | |
| "grad_norm": 2.6362013816833496, | |
| "learning_rate": 9.703174712528094e-05, | |
| "loss": 2.1432125854492186, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.133125, | |
| "grad_norm": 2.188535690307617, | |
| "learning_rate": 9.699747487995039e-05, | |
| "loss": 2.2337890625, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 0.13375, | |
| "grad_norm": 2.5218043327331543, | |
| "learning_rate": 9.696301203372893e-05, | |
| "loss": 2.2242401123046873, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.134375, | |
| "grad_norm": 2.3167622089385986, | |
| "learning_rate": 9.692835872638255e-05, | |
| "loss": 2.2438348388671874, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 0.135, | |
| "grad_norm": 1.6925166845321655, | |
| "learning_rate": 9.689351509844966e-05, | |
| "loss": 2.1800050354003906, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.135625, | |
| "grad_norm": 2.118764638900757, | |
| "learning_rate": 9.685848129124057e-05, | |
| "loss": 2.279612274169922, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 0.13625, | |
| "grad_norm": 2.541534423828125, | |
| "learning_rate": 9.682325744683686e-05, | |
| "loss": 2.2910064697265624, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.136875, | |
| "grad_norm": 3.2209434509277344, | |
| "learning_rate": 9.678784370809076e-05, | |
| "loss": 2.255341796875, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 0.1375, | |
| "grad_norm": 2.286287784576416, | |
| "learning_rate": 9.675224021862471e-05, | |
| "loss": 2.2873764038085938, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.138125, | |
| "grad_norm": 2.6872918605804443, | |
| "learning_rate": 9.671644712283061e-05, | |
| "loss": 2.0835134887695315, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 0.13875, | |
| "grad_norm": 2.6672234535217285, | |
| "learning_rate": 9.66804645658694e-05, | |
| "loss": 2.1357687377929686, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.139375, | |
| "grad_norm": 2.4629364013671875, | |
| "learning_rate": 9.664429269367034e-05, | |
| "loss": 2.103699493408203, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 4.718692302703857, | |
| "learning_rate": 9.660793165293051e-05, | |
| "loss": 2.1635005187988283, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.140625, | |
| "grad_norm": 2.5984413623809814, | |
| "learning_rate": 9.65713815911141e-05, | |
| "loss": 2.2417706298828124, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 0.14125, | |
| "grad_norm": 2.926302671432495, | |
| "learning_rate": 9.653464265645197e-05, | |
| "loss": 2.2174407958984377, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.141875, | |
| "grad_norm": 2.5676918029785156, | |
| "learning_rate": 9.649771499794091e-05, | |
| "loss": 2.10097412109375, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 0.1425, | |
| "grad_norm": 1.7818841934204102, | |
| "learning_rate": 9.646059876534313e-05, | |
| "loss": 2.129024658203125, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.143125, | |
| "grad_norm": 1.8646036386489868, | |
| "learning_rate": 9.642329410918554e-05, | |
| "loss": 2.0129293823242187, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 0.14375, | |
| "grad_norm": 2.1912927627563477, | |
| "learning_rate": 9.638580118075932e-05, | |
| "loss": 2.202617340087891, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.144375, | |
| "grad_norm": 2.462655782699585, | |
| "learning_rate": 9.634812013211909e-05, | |
| "loss": 2.1457453918457032, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 0.145, | |
| "grad_norm": 2.062124013900757, | |
| "learning_rate": 9.63102511160825e-05, | |
| "loss": 2.299252166748047, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.145625, | |
| "grad_norm": 2.6305387020111084, | |
| "learning_rate": 9.627219428622943e-05, | |
| "loss": 2.2775376892089843, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 0.14625, | |
| "grad_norm": 1.9412083625793457, | |
| "learning_rate": 9.623394979690147e-05, | |
| "loss": 2.3565916442871093, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 0.146875, | |
| "grad_norm": 2.199693202972412, | |
| "learning_rate": 9.619551780320134e-05, | |
| "loss": 2.160765838623047, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 0.1475, | |
| "grad_norm": 2.2162418365478516, | |
| "learning_rate": 9.615689846099212e-05, | |
| "loss": 2.0144644165039063, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.148125, | |
| "grad_norm": 2.270930051803589, | |
| "learning_rate": 9.61180919268967e-05, | |
| "loss": 2.205564422607422, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 0.14875, | |
| "grad_norm": 1.832543969154358, | |
| "learning_rate": 9.607909835829716e-05, | |
| "loss": 2.1496823120117186, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 0.149375, | |
| "grad_norm": 2.0444109439849854, | |
| "learning_rate": 9.60399179133341e-05, | |
| "loss": 2.18775390625, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 1.747004508972168, | |
| "learning_rate": 9.600055075090601e-05, | |
| "loss": 2.0697125244140624, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_loss": 0.4449842572212219, | |
| "eval_runtime": 16.3655, | |
| "eval_samples_per_second": 88.967, | |
| "eval_steps_per_second": 5.56, | |
| "eval_wer": 0.254394656684649, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.150625, | |
| "grad_norm": 2.1939504146575928, | |
| "learning_rate": 9.596099703066863e-05, | |
| "loss": 2.129417724609375, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 0.15125, | |
| "grad_norm": 2.846012830734253, | |
| "learning_rate": 9.592125691303427e-05, | |
| "loss": 2.193754119873047, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 0.151875, | |
| "grad_norm": 2.117307424545288, | |
| "learning_rate": 9.588133055917122e-05, | |
| "loss": 2.3042280578613283, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 0.1525, | |
| "grad_norm": 2.637690305709839, | |
| "learning_rate": 9.584121813100304e-05, | |
| "loss": 2.20600341796875, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.153125, | |
| "grad_norm": 2.458855390548706, | |
| "learning_rate": 9.580091979120792e-05, | |
| "loss": 2.2941462707519533, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 0.15375, | |
| "grad_norm": 2.8159937858581543, | |
| "learning_rate": 9.576043570321804e-05, | |
| "loss": 2.127772216796875, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 0.154375, | |
| "grad_norm": 2.2587006092071533, | |
| "learning_rate": 9.571976603121888e-05, | |
| "loss": 2.0940127563476563, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 0.155, | |
| "grad_norm": 2.2611312866210938, | |
| "learning_rate": 9.567891094014856e-05, | |
| "loss": 2.0166357421875, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.155625, | |
| "grad_norm": 2.003955841064453, | |
| "learning_rate": 9.563787059569724e-05, | |
| "loss": 2.1084141540527344, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 0.15625, | |
| "grad_norm": 2.8468680381774902, | |
| "learning_rate": 9.559664516430628e-05, | |
| "loss": 2.1620755004882812, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.156875, | |
| "grad_norm": 2.2891390323638916, | |
| "learning_rate": 9.555523481316773e-05, | |
| "loss": 2.0871788024902345, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 0.1575, | |
| "grad_norm": 3.441826105117798, | |
| "learning_rate": 9.551363971022363e-05, | |
| "loss": 2.0544528198242187, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.158125, | |
| "grad_norm": 1.5288652181625366, | |
| "learning_rate": 9.547186002416521e-05, | |
| "loss": 2.08611572265625, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 0.15875, | |
| "grad_norm": 1.8317360877990723, | |
| "learning_rate": 9.542989592443235e-05, | |
| "loss": 2.1356245422363282, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 0.159375, | |
| "grad_norm": 2.2312066555023193, | |
| "learning_rate": 9.538774758121277e-05, | |
| "loss": 2.1984071350097656, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 1.9797857999801636, | |
| "learning_rate": 9.534541516544147e-05, | |
| "loss": 2.1279429626464843, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.160625, | |
| "grad_norm": 2.07370924949646, | |
| "learning_rate": 9.530289884879993e-05, | |
| "loss": 2.206969909667969, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 0.16125, | |
| "grad_norm": 2.1191327571868896, | |
| "learning_rate": 9.526019880371544e-05, | |
| "loss": 2.2596502685546875, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 0.161875, | |
| "grad_norm": 1.8480347394943237, | |
| "learning_rate": 9.521731520336044e-05, | |
| "loss": 2.1021141052246093, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 0.1625, | |
| "grad_norm": 2.238983631134033, | |
| "learning_rate": 9.517424822165175e-05, | |
| "loss": 2.101505126953125, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.163125, | |
| "grad_norm": 1.9658758640289307, | |
| "learning_rate": 9.513099803324994e-05, | |
| "loss": 2.2028057861328123, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 0.16375, | |
| "grad_norm": 1.738741159439087, | |
| "learning_rate": 9.508756481355856e-05, | |
| "loss": 2.1539662170410154, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 0.164375, | |
| "grad_norm": 2.2676610946655273, | |
| "learning_rate": 9.504394873872349e-05, | |
| "loss": 2.1724642944335937, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 0.165, | |
| "grad_norm": 2.54267954826355, | |
| "learning_rate": 9.500014998563214e-05, | |
| "loss": 2.2173406982421877, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.165625, | |
| "grad_norm": 2.314807653427124, | |
| "learning_rate": 9.49561687319128e-05, | |
| "loss": 2.1208013916015624, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 0.16625, | |
| "grad_norm": 3.054250717163086, | |
| "learning_rate": 9.491200515593394e-05, | |
| "loss": 2.2521047973632813, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 0.166875, | |
| "grad_norm": 2.711958885192871, | |
| "learning_rate": 9.486765943680338e-05, | |
| "loss": 2.1743328857421873, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 0.1675, | |
| "grad_norm": 2.922079086303711, | |
| "learning_rate": 9.482313175436768e-05, | |
| "loss": 2.197858428955078, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.168125, | |
| "grad_norm": 2.231698989868164, | |
| "learning_rate": 9.477842228921136e-05, | |
| "loss": 2.3998699951171876, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 0.16875, | |
| "grad_norm": 3.7652249336242676, | |
| "learning_rate": 9.473353122265614e-05, | |
| "loss": 2.6393746948242187, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 0.169375, | |
| "grad_norm": 2.0199086666107178, | |
| "learning_rate": 9.468845873676027e-05, | |
| "loss": 2.1917437744140624, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 2.3263399600982666, | |
| "learning_rate": 9.464320501431773e-05, | |
| "loss": 2.088440399169922, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.170625, | |
| "grad_norm": 2.06636118888855, | |
| "learning_rate": 9.459777023885755e-05, | |
| "loss": 2.064405670166016, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 0.17125, | |
| "grad_norm": 1.8440086841583252, | |
| "learning_rate": 9.455215459464297e-05, | |
| "loss": 1.9831680297851562, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 0.171875, | |
| "grad_norm": 1.9167689085006714, | |
| "learning_rate": 9.45063582666708e-05, | |
| "loss": 2.1471177673339845, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 0.1725, | |
| "grad_norm": 2.6260464191436768, | |
| "learning_rate": 9.446038144067062e-05, | |
| "loss": 2.166270446777344, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.173125, | |
| "grad_norm": 1.837060809135437, | |
| "learning_rate": 9.441422430310403e-05, | |
| "loss": 2.1496157836914063, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 0.17375, | |
| "grad_norm": 1.9264370203018188, | |
| "learning_rate": 9.436788704116386e-05, | |
| "loss": 2.0744915771484376, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 0.174375, | |
| "grad_norm": 2.0549418926239014, | |
| "learning_rate": 9.432136984277349e-05, | |
| "loss": 2.1053639221191407, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 0.175, | |
| "grad_norm": 1.947101354598999, | |
| "learning_rate": 9.427467289658603e-05, | |
| "loss": 2.0817604064941406, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.175, | |
| "eval_loss": 0.7903617024421692, | |
| "eval_runtime": 25.7513, | |
| "eval_samples_per_second": 56.541, | |
| "eval_steps_per_second": 3.534, | |
| "eval_wer": 0.3478659767330911, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.175625, | |
| "grad_norm": 2.2212603092193604, | |
| "learning_rate": 9.422779639198353e-05, | |
| "loss": 2.2741217041015624, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 0.17625, | |
| "grad_norm": 1.9873552322387695, | |
| "learning_rate": 9.418074051907633e-05, | |
| "loss": 2.0904864501953124, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 0.176875, | |
| "grad_norm": 2.7408645153045654, | |
| "learning_rate": 9.413350546870209e-05, | |
| "loss": 1.9595954895019532, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 0.1775, | |
| "grad_norm": 2.2390503883361816, | |
| "learning_rate": 9.408609143242526e-05, | |
| "loss": 2.0675558471679687, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.178125, | |
| "grad_norm": 2.1097710132598877, | |
| "learning_rate": 9.403849860253609e-05, | |
| "loss": 1.908758544921875, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 0.17875, | |
| "grad_norm": 2.0927650928497314, | |
| "learning_rate": 9.399072717204995e-05, | |
| "loss": 2.032224426269531, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 0.179375, | |
| "grad_norm": 2.0611257553100586, | |
| "learning_rate": 9.39427773347066e-05, | |
| "loss": 2.0971710205078127, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 2.6477508544921875, | |
| "learning_rate": 9.389464928496926e-05, | |
| "loss": 2.072987365722656, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.180625, | |
| "grad_norm": 2.9647626876831055, | |
| "learning_rate": 9.384634321802393e-05, | |
| "loss": 2.1512625122070315, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 0.18125, | |
| "grad_norm": 2.303650140762329, | |
| "learning_rate": 9.379785932977858e-05, | |
| "loss": 2.212369384765625, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 0.181875, | |
| "grad_norm": 2.2667808532714844, | |
| "learning_rate": 9.374919781686231e-05, | |
| "loss": 2.1247869873046876, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 0.1825, | |
| "grad_norm": 2.4079856872558594, | |
| "learning_rate": 9.370035887662465e-05, | |
| "loss": 2.0908448791503904, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.183125, | |
| "grad_norm": 1.9005861282348633, | |
| "learning_rate": 9.365134270713463e-05, | |
| "loss": 2.055008850097656, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 0.18375, | |
| "grad_norm": 2.5896031856536865, | |
| "learning_rate": 9.360214950718006e-05, | |
| "loss": 2.096898498535156, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 0.184375, | |
| "grad_norm": 2.2877445220947266, | |
| "learning_rate": 9.355277947626674e-05, | |
| "loss": 2.029735412597656, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 0.185, | |
| "grad_norm": 2.1667251586914062, | |
| "learning_rate": 9.350323281461759e-05, | |
| "loss": 1.9323843383789063, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.185625, | |
| "grad_norm": 1.9115021228790283, | |
| "learning_rate": 9.345350972317186e-05, | |
| "loss": 2.066067810058594, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 0.18625, | |
| "grad_norm": 2.01686692237854, | |
| "learning_rate": 9.340361040358434e-05, | |
| "loss": 2.1941493225097655, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 0.186875, | |
| "grad_norm": 1.412351369857788, | |
| "learning_rate": 9.33535350582245e-05, | |
| "loss": 1.9150367736816407, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 0.1875, | |
| "grad_norm": 2.7584800720214844, | |
| "learning_rate": 9.330328389017574e-05, | |
| "loss": 2.055033721923828, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.188125, | |
| "grad_norm": 1.7262470722198486, | |
| "learning_rate": 9.325285710323446e-05, | |
| "loss": 2.0305549621582033, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 0.18875, | |
| "grad_norm": 1.6967185735702515, | |
| "learning_rate": 9.320225490190932e-05, | |
| "loss": 2.110368347167969, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 0.189375, | |
| "grad_norm": 3.2678163051605225, | |
| "learning_rate": 9.315147749142039e-05, | |
| "loss": 2.2655294799804686, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 2.0790610313415527, | |
| "learning_rate": 9.310052507769828e-05, | |
| "loss": 2.0846192932128904, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.190625, | |
| "grad_norm": 2.0555837154388428, | |
| "learning_rate": 9.304939786738339e-05, | |
| "loss": 2.0940182495117186, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 0.19125, | |
| "grad_norm": 2.003434181213379, | |
| "learning_rate": 9.299809606782497e-05, | |
| "loss": 2.1488217163085936, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 0.191875, | |
| "grad_norm": 2.7323293685913086, | |
| "learning_rate": 9.294661988708032e-05, | |
| "loss": 2.064187774658203, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 0.1925, | |
| "grad_norm": 3.719198703765869, | |
| "learning_rate": 9.2894969533914e-05, | |
| "loss": 1.989752197265625, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.193125, | |
| "grad_norm": 2.450850009918213, | |
| "learning_rate": 9.28431452177969e-05, | |
| "loss": 2.042866516113281, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 0.19375, | |
| "grad_norm": 1.9355038404464722, | |
| "learning_rate": 9.279114714890543e-05, | |
| "loss": 2.0101048278808595, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 0.194375, | |
| "grad_norm": 2.764147996902466, | |
| "learning_rate": 9.273897553812068e-05, | |
| "loss": 2.1429312133789065, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 0.195, | |
| "grad_norm": 1.7017638683319092, | |
| "learning_rate": 9.268663059702753e-05, | |
| "loss": 2.110185089111328, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.195625, | |
| "grad_norm": 1.878867506980896, | |
| "learning_rate": 9.263411253791384e-05, | |
| "loss": 1.9916375732421876, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 0.19625, | |
| "grad_norm": 2.2714357376098633, | |
| "learning_rate": 9.258142157376952e-05, | |
| "loss": 2.04151123046875, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 0.196875, | |
| "grad_norm": 2.3166255950927734, | |
| "learning_rate": 9.252855791828571e-05, | |
| "loss": 1.9792080688476563, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 0.1975, | |
| "grad_norm": 2.0201849937438965, | |
| "learning_rate": 9.247552178585397e-05, | |
| "loss": 2.042154541015625, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.198125, | |
| "grad_norm": 2.234557867050171, | |
| "learning_rate": 9.242231339156527e-05, | |
| "loss": 2.1123789978027343, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 0.19875, | |
| "grad_norm": 2.220761775970459, | |
| "learning_rate": 9.236893295120923e-05, | |
| "loss": 2.0877952575683594, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 0.199375, | |
| "grad_norm": 1.8211852312088013, | |
| "learning_rate": 9.231538068127323e-05, | |
| "loss": 2.0735809326171877, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 2.3432486057281494, | |
| "learning_rate": 9.22616567989415e-05, | |
| "loss": 2.190612030029297, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.6346888542175293, | |
| "eval_runtime": 24.0078, | |
| "eval_samples_per_second": 60.647, | |
| "eval_steps_per_second": 3.79, | |
| "eval_wer": 0.31744284193915373, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.200625, | |
| "grad_norm": 1.9170923233032227, | |
| "learning_rate": 9.220776152209425e-05, | |
| "loss": 2.1188458251953124, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 0.20125, | |
| "grad_norm": 1.8902095556259155, | |
| "learning_rate": 9.21536950693068e-05, | |
| "loss": 1.9264413452148437, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 0.201875, | |
| "grad_norm": 2.5802721977233887, | |
| "learning_rate": 9.209945765984864e-05, | |
| "loss": 2.0863734436035157, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 0.2025, | |
| "grad_norm": 2.557882308959961, | |
| "learning_rate": 9.204504951368267e-05, | |
| "loss": 2.059514923095703, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.203125, | |
| "grad_norm": 2.689347267150879, | |
| "learning_rate": 9.199047085146415e-05, | |
| "loss": 2.0764089965820314, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 0.20375, | |
| "grad_norm": 1.9851337671279907, | |
| "learning_rate": 9.19357218945399e-05, | |
| "loss": 2.2245468139648437, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 0.204375, | |
| "grad_norm": 2.4927570819854736, | |
| "learning_rate": 9.188080286494738e-05, | |
| "loss": 2.1381333923339843, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 0.205, | |
| "grad_norm": 2.5442142486572266, | |
| "learning_rate": 9.18257139854138e-05, | |
| "loss": 1.9946365356445312, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.205625, | |
| "grad_norm": 3.00085186958313, | |
| "learning_rate": 9.177045547935518e-05, | |
| "loss": 2.0118177795410155, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 0.20625, | |
| "grad_norm": 2.2000389099121094, | |
| "learning_rate": 9.17150275708755e-05, | |
| "loss": 1.9428131103515625, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 0.206875, | |
| "grad_norm": 1.9186347723007202, | |
| "learning_rate": 9.165943048476574e-05, | |
| "loss": 1.9713262939453124, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 0.2075, | |
| "grad_norm": 2.9448137283325195, | |
| "learning_rate": 9.1603664446503e-05, | |
| "loss": 1.9461703491210938, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.208125, | |
| "grad_norm": 2.309593677520752, | |
| "learning_rate": 9.154772968224956e-05, | |
| "loss": 1.9409890747070313, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 0.20875, | |
| "grad_norm": 2.428645372390747, | |
| "learning_rate": 9.149162641885199e-05, | |
| "loss": 2.0932371520996096, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 0.209375, | |
| "grad_norm": 1.885838508605957, | |
| "learning_rate": 9.143535488384023e-05, | |
| "loss": 2.037969207763672, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 1.7821649312973022, | |
| "learning_rate": 9.137891530542663e-05, | |
| "loss": 2.072525634765625, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.210625, | |
| "grad_norm": 2.1600959300994873, | |
| "learning_rate": 9.132230791250507e-05, | |
| "loss": 2.152512969970703, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 0.21125, | |
| "grad_norm": 2.6756789684295654, | |
| "learning_rate": 9.126553293464998e-05, | |
| "loss": 2.0600608825683593, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 0.211875, | |
| "grad_norm": 2.103846311569214, | |
| "learning_rate": 9.120859060211548e-05, | |
| "loss": 2.1617041015625, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 0.2125, | |
| "grad_norm": 2.5558128356933594, | |
| "learning_rate": 9.115148114583439e-05, | |
| "loss": 1.95693603515625, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.213125, | |
| "grad_norm": 3.469834327697754, | |
| "learning_rate": 9.10942047974173e-05, | |
| "loss": 2.1383297729492186, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 0.21375, | |
| "grad_norm": 1.7859108448028564, | |
| "learning_rate": 9.103676178915166e-05, | |
| "loss": 2.0209396362304686, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 0.214375, | |
| "grad_norm": 2.7847373485565186, | |
| "learning_rate": 9.097915235400079e-05, | |
| "loss": 2.0980014038085937, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 0.215, | |
| "grad_norm": 1.914182186126709, | |
| "learning_rate": 9.092137672560298e-05, | |
| "loss": 1.9893528747558593, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.215625, | |
| "grad_norm": 2.0414235591888428, | |
| "learning_rate": 9.086343513827054e-05, | |
| "loss": 2.209757537841797, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 0.21625, | |
| "grad_norm": 2.1998109817504883, | |
| "learning_rate": 9.08053278269888e-05, | |
| "loss": 2.2378306579589844, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 0.216875, | |
| "grad_norm": 1.8598991632461548, | |
| "learning_rate": 9.074705502741525e-05, | |
| "loss": 2.043817443847656, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 0.2175, | |
| "grad_norm": 2.380380392074585, | |
| "learning_rate": 9.068861697587845e-05, | |
| "loss": 2.089021759033203, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.218125, | |
| "grad_norm": 2.4009857177734375, | |
| "learning_rate": 9.063001390937722e-05, | |
| "loss": 2.054262237548828, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 0.21875, | |
| "grad_norm": 2.0311126708984375, | |
| "learning_rate": 9.057124606557956e-05, | |
| "loss": 1.970128173828125, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 0.219375, | |
| "grad_norm": 2.1672239303588867, | |
| "learning_rate": 9.051231368282177e-05, | |
| "loss": 1.9935519409179687, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 2.0012753009796143, | |
| "learning_rate": 9.045321700010741e-05, | |
| "loss": 1.988118896484375, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.220625, | |
| "grad_norm": 1.8105686902999878, | |
| "learning_rate": 9.03939562571064e-05, | |
| "loss": 2.0694140625, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 0.22125, | |
| "grad_norm": 2.4012067317962646, | |
| "learning_rate": 9.0334531694154e-05, | |
| "loss": 2.083789520263672, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 0.221875, | |
| "grad_norm": 2.0857622623443604, | |
| "learning_rate": 9.027494355224984e-05, | |
| "loss": 2.0428839111328125, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 0.2225, | |
| "grad_norm": 2.792790174484253, | |
| "learning_rate": 9.021519207305697e-05, | |
| "loss": 2.0489065551757815, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.223125, | |
| "grad_norm": 2.5423669815063477, | |
| "learning_rate": 9.015527749890086e-05, | |
| "loss": 2.016570587158203, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 0.22375, | |
| "grad_norm": 1.7150146961212158, | |
| "learning_rate": 9.009520007276842e-05, | |
| "loss": 1.8667630004882811, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 0.224375, | |
| "grad_norm": 2.8952643871307373, | |
| "learning_rate": 9.003496003830703e-05, | |
| "loss": 1.9878672790527343, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 0.225, | |
| "grad_norm": 2.1229841709136963, | |
| "learning_rate": 8.99745576398235e-05, | |
| "loss": 2.1314736938476564, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.225, | |
| "eval_loss": 0.5940452814102173, | |
| "eval_runtime": 32.223, | |
| "eval_samples_per_second": 45.185, | |
| "eval_steps_per_second": 2.824, | |
| "eval_wer": 0.3113875738559213, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.225625, | |
| "grad_norm": 2.204343318939209, | |
| "learning_rate": 8.991399312228317e-05, | |
| "loss": 2.124026336669922, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 0.22625, | |
| "grad_norm": 2.272564172744751, | |
| "learning_rate": 8.985326673130879e-05, | |
| "loss": 1.9990528869628905, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 0.226875, | |
| "grad_norm": 3.719158411026001, | |
| "learning_rate": 8.979237871317967e-05, | |
| "loss": 2.0054283142089844, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 0.2275, | |
| "grad_norm": 2.5514683723449707, | |
| "learning_rate": 8.973132931483057e-05, | |
| "loss": 2.120261383056641, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.228125, | |
| "grad_norm": 2.807434320449829, | |
| "learning_rate": 8.967011878385074e-05, | |
| "loss": 1.9910020446777343, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 0.22875, | |
| "grad_norm": 2.172513246536255, | |
| "learning_rate": 8.960874736848292e-05, | |
| "loss": 2.2589642333984377, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 0.229375, | |
| "grad_norm": 3.1625287532806396, | |
| "learning_rate": 8.954721531762232e-05, | |
| "loss": 1.888506622314453, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 2.241400957107544, | |
| "learning_rate": 8.948552288081562e-05, | |
| "loss": 1.9426483154296874, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.230625, | |
| "grad_norm": 2.4364380836486816, | |
| "learning_rate": 8.942367030825994e-05, | |
| "loss": 2.0964259338378906, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 0.23125, | |
| "grad_norm": 2.825659990310669, | |
| "learning_rate": 8.936165785080186e-05, | |
| "loss": 1.9243450927734376, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 0.231875, | |
| "grad_norm": 1.7159090042114258, | |
| "learning_rate": 8.929948575993637e-05, | |
| "loss": 1.9655561828613282, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 0.2325, | |
| "grad_norm": 2.4356799125671387, | |
| "learning_rate": 8.923715428780588e-05, | |
| "loss": 1.9292584228515626, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.233125, | |
| "grad_norm": 1.9144049882888794, | |
| "learning_rate": 8.917466368719916e-05, | |
| "loss": 2.05948974609375, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 0.23375, | |
| "grad_norm": 2.614051580429077, | |
| "learning_rate": 8.911201421155031e-05, | |
| "loss": 1.9688128662109374, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 0.234375, | |
| "grad_norm": 1.9700573682785034, | |
| "learning_rate": 8.904920611493783e-05, | |
| "loss": 1.9391226196289062, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 0.235, | |
| "grad_norm": 2.6903114318847656, | |
| "learning_rate": 8.898623965208343e-05, | |
| "loss": 1.927160186767578, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.235625, | |
| "grad_norm": 1.8229823112487793, | |
| "learning_rate": 8.892311507835119e-05, | |
| "loss": 1.9136570739746093, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 0.23625, | |
| "grad_norm": 2.198741912841797, | |
| "learning_rate": 8.885983264974628e-05, | |
| "loss": 1.9460650634765626, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 0.236875, | |
| "grad_norm": 2.208942413330078, | |
| "learning_rate": 8.879639262291417e-05, | |
| "loss": 1.993262939453125, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 0.2375, | |
| "grad_norm": 2.250863790512085, | |
| "learning_rate": 8.873279525513943e-05, | |
| "loss": 1.8924490356445312, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.238125, | |
| "grad_norm": 2.1591739654541016, | |
| "learning_rate": 8.866904080434477e-05, | |
| "loss": 1.9552699279785157, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 0.23875, | |
| "grad_norm": 3.358647346496582, | |
| "learning_rate": 8.86051295290899e-05, | |
| "loss": 2.002980499267578, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 0.239375, | |
| "grad_norm": 1.877958059310913, | |
| "learning_rate": 8.85410616885706e-05, | |
| "loss": 2.0177581787109373, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 1.9018381834030151, | |
| "learning_rate": 8.847683754261757e-05, | |
| "loss": 1.9148069763183593, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.240625, | |
| "grad_norm": 1.8892041444778442, | |
| "learning_rate": 8.841245735169545e-05, | |
| "loss": 1.9308103942871093, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 0.24125, | |
| "grad_norm": 2.4355297088623047, | |
| "learning_rate": 8.83479213769017e-05, | |
| "loss": 1.9591015625, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 0.241875, | |
| "grad_norm": 2.590184450149536, | |
| "learning_rate": 8.828322987996556e-05, | |
| "loss": 2.022674560546875, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 0.2425, | |
| "grad_norm": 2.4130208492279053, | |
| "learning_rate": 8.821838312324704e-05, | |
| "loss": 2.028225555419922, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.243125, | |
| "grad_norm": 1.9873545169830322, | |
| "learning_rate": 8.81533813697358e-05, | |
| "loss": 2.0166253662109375, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 0.24375, | |
| "grad_norm": 2.1698150634765625, | |
| "learning_rate": 8.808822488305005e-05, | |
| "loss": 2.1305108642578126, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 0.244375, | |
| "grad_norm": 6.50513219833374, | |
| "learning_rate": 8.802291392743562e-05, | |
| "loss": 2.0857550048828126, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 0.245, | |
| "grad_norm": 2.437391757965088, | |
| "learning_rate": 8.795744876776471e-05, | |
| "loss": 2.0373057556152343, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.245625, | |
| "grad_norm": 1.8666143417358398, | |
| "learning_rate": 8.789182966953498e-05, | |
| "loss": 2.031862487792969, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 0.24625, | |
| "grad_norm": 2.410386085510254, | |
| "learning_rate": 8.78260568988683e-05, | |
| "loss": 2.2114854431152344, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 0.246875, | |
| "grad_norm": 2.5028300285339355, | |
| "learning_rate": 8.776013072250988e-05, | |
| "loss": 2.050908966064453, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 0.2475, | |
| "grad_norm": 3.532036542892456, | |
| "learning_rate": 8.769405140782699e-05, | |
| "loss": 2.154196472167969, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.248125, | |
| "grad_norm": 2.6913037300109863, | |
| "learning_rate": 8.7627819222808e-05, | |
| "loss": 2.111236572265625, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 0.24875, | |
| "grad_norm": 2.595747470855713, | |
| "learning_rate": 8.756143443606125e-05, | |
| "loss": 2.236370086669922, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 0.249375, | |
| "grad_norm": 3.8136496543884277, | |
| "learning_rate": 8.749489731681392e-05, | |
| "loss": 2.216940155029297, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 1.829177975654602, | |
| "learning_rate": 8.742820813491108e-05, | |
| "loss": 2.0302940368652345, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "eval_loss": 0.7854554057121277, | |
| "eval_runtime": 25.8739, | |
| "eval_samples_per_second": 56.273, | |
| "eval_steps_per_second": 3.517, | |
| "eval_wer": 0.3288561048111857, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.250625, | |
| "grad_norm": 2.1718060970306396, | |
| "learning_rate": 8.736136716081443e-05, | |
| "loss": 1.9573629760742188, | |
| "step": 10025 | |
| }, | |
| { | |
| "epoch": 0.25125, | |
| "grad_norm": 2.6805076599121094, | |
| "learning_rate": 8.729437466560125e-05, | |
| "loss": 2.042824249267578, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 0.251875, | |
| "grad_norm": 5.174410343170166, | |
| "learning_rate": 8.722723092096338e-05, | |
| "loss": 2.1021965026855467, | |
| "step": 10075 | |
| }, | |
| { | |
| "epoch": 0.2525, | |
| "grad_norm": 2.4342896938323975, | |
| "learning_rate": 8.715993619920603e-05, | |
| "loss": 2.3931375122070313, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 0.253125, | |
| "grad_norm": 2.0031697750091553, | |
| "learning_rate": 8.709249077324671e-05, | |
| "loss": 2.5030453491210936, | |
| "step": 10125 | |
| }, | |
| { | |
| "epoch": 0.25375, | |
| "grad_norm": 2.750117063522339, | |
| "learning_rate": 8.702489491661413e-05, | |
| "loss": 2.0430174255371094, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 0.254375, | |
| "grad_norm": 2.01338791847229, | |
| "learning_rate": 8.695714890344706e-05, | |
| "loss": 1.911546630859375, | |
| "step": 10175 | |
| }, | |
| { | |
| "epoch": 0.255, | |
| "grad_norm": 2.5122487545013428, | |
| "learning_rate": 8.688925300849327e-05, | |
| "loss": 1.9358489990234375, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.255625, | |
| "grad_norm": 2.6769537925720215, | |
| "learning_rate": 8.682120750710834e-05, | |
| "loss": 2.0621261596679688, | |
| "step": 10225 | |
| }, | |
| { | |
| "epoch": 0.25625, | |
| "grad_norm": 2.3083302974700928, | |
| "learning_rate": 8.67530126752546e-05, | |
| "loss": 2.0336347961425782, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 0.256875, | |
| "grad_norm": 1.7496837377548218, | |
| "learning_rate": 8.668466878950006e-05, | |
| "loss": 2.009535369873047, | |
| "step": 10275 | |
| }, | |
| { | |
| "epoch": 0.2575, | |
| "grad_norm": 2.3086769580841064, | |
| "learning_rate": 8.661617612701713e-05, | |
| "loss": 1.9052336120605469, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 0.258125, | |
| "grad_norm": 2.911654233932495, | |
| "learning_rate": 8.654753496558165e-05, | |
| "loss": 1.9193478393554688, | |
| "step": 10325 | |
| }, | |
| { | |
| "epoch": 0.25875, | |
| "grad_norm": 2.178903818130493, | |
| "learning_rate": 8.647874558357172e-05, | |
| "loss": 1.9950228881835939, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 0.259375, | |
| "grad_norm": 2.7750067710876465, | |
| "learning_rate": 8.640980825996649e-05, | |
| "loss": 2.0591200256347655, | |
| "step": 10375 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 2.5789103507995605, | |
| "learning_rate": 8.634072327434515e-05, | |
| "loss": 2.027841033935547, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 0.260625, | |
| "grad_norm": 2.0377888679504395, | |
| "learning_rate": 8.627149090688573e-05, | |
| "loss": 1.8890194702148437, | |
| "step": 10425 | |
| }, | |
| { | |
| "epoch": 0.26125, | |
| "grad_norm": 2.9035537242889404, | |
| "learning_rate": 8.620211143836399e-05, | |
| "loss": 2.076640777587891, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 0.261875, | |
| "grad_norm": 2.753873825073242, | |
| "learning_rate": 8.613258515015222e-05, | |
| "loss": 2.1916676330566407, | |
| "step": 10475 | |
| }, | |
| { | |
| "epoch": 0.2625, | |
| "grad_norm": 2.0765042304992676, | |
| "learning_rate": 8.606291232421817e-05, | |
| "loss": 2.0846051025390624, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.263125, | |
| "grad_norm": 2.840498924255371, | |
| "learning_rate": 8.599309324312392e-05, | |
| "loss": 1.9943656921386719, | |
| "step": 10525 | |
| }, | |
| { | |
| "epoch": 0.26375, | |
| "grad_norm": 1.9403188228607178, | |
| "learning_rate": 8.592312819002466e-05, | |
| "loss": 2.1683949279785155, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 0.264375, | |
| "grad_norm": 2.0462074279785156, | |
| "learning_rate": 8.585301744866754e-05, | |
| "loss": 1.9398512268066406, | |
| "step": 10575 | |
| }, | |
| { | |
| "epoch": 0.265, | |
| "grad_norm": 3.101336717605591, | |
| "learning_rate": 8.578276130339065e-05, | |
| "loss": 1.9195005798339844, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 0.265625, | |
| "grad_norm": 2.9250519275665283, | |
| "learning_rate": 8.571236003912168e-05, | |
| "loss": 2.1554660034179687, | |
| "step": 10625 | |
| }, | |
| { | |
| "epoch": 0.26625, | |
| "grad_norm": 1.8713876008987427, | |
| "learning_rate": 8.564181394137692e-05, | |
| "loss": 2.059755859375, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 0.266875, | |
| "grad_norm": 3.2304368019104004, | |
| "learning_rate": 8.557112329626003e-05, | |
| "loss": 1.9864997863769531, | |
| "step": 10675 | |
| }, | |
| { | |
| "epoch": 0.2675, | |
| "grad_norm": 1.734165072441101, | |
| "learning_rate": 8.550028839046085e-05, | |
| "loss": 1.9298501586914063, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 0.268125, | |
| "grad_norm": 2.250795841217041, | |
| "learning_rate": 8.542930951125432e-05, | |
| "loss": 1.9580307006835938, | |
| "step": 10725 | |
| }, | |
| { | |
| "epoch": 0.26875, | |
| "grad_norm": 2.1491730213165283, | |
| "learning_rate": 8.535818694649925e-05, | |
| "loss": 2.063010559082031, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 0.269375, | |
| "grad_norm": 1.8326674699783325, | |
| "learning_rate": 8.52869209846372e-05, | |
| "loss": 1.9176422119140626, | |
| "step": 10775 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 2.222517728805542, | |
| "learning_rate": 8.521551191469128e-05, | |
| "loss": 2.0425796508789062, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 0.270625, | |
| "grad_norm": 1.8573559522628784, | |
| "learning_rate": 8.514396002626493e-05, | |
| "loss": 2.162534027099609, | |
| "step": 10825 | |
| }, | |
| { | |
| "epoch": 0.27125, | |
| "grad_norm": 2.0595502853393555, | |
| "learning_rate": 8.507226560954087e-05, | |
| "loss": 2.046962890625, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 0.271875, | |
| "grad_norm": 1.7844104766845703, | |
| "learning_rate": 8.500042895527982e-05, | |
| "loss": 1.970653076171875, | |
| "step": 10875 | |
| }, | |
| { | |
| "epoch": 0.2725, | |
| "grad_norm": 2.583991765975952, | |
| "learning_rate": 8.492845035481935e-05, | |
| "loss": 1.99356201171875, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 0.273125, | |
| "grad_norm": 2.370809316635132, | |
| "learning_rate": 8.48563301000727e-05, | |
| "loss": 1.9609019470214843, | |
| "step": 10925 | |
| }, | |
| { | |
| "epoch": 0.27375, | |
| "grad_norm": 2.251582384109497, | |
| "learning_rate": 8.478406848352761e-05, | |
| "loss": 1.97093994140625, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 0.274375, | |
| "grad_norm": 1.7385530471801758, | |
| "learning_rate": 8.47116657982451e-05, | |
| "loss": 1.9887588500976563, | |
| "step": 10975 | |
| }, | |
| { | |
| "epoch": 0.275, | |
| "grad_norm": 2.855250120162964, | |
| "learning_rate": 8.463912233785834e-05, | |
| "loss": 1.9731906127929688, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.275, | |
| "eval_loss": 0.44069960713386536, | |
| "eval_runtime": 22.3314, | |
| "eval_samples_per_second": 65.2, | |
| "eval_steps_per_second": 4.075, | |
| "eval_wer": 0.22918272230173584, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.275625, | |
| "grad_norm": 3.3369674682617188, | |
| "learning_rate": 8.456643839657134e-05, | |
| "loss": 1.940252685546875, | |
| "step": 11025 | |
| }, | |
| { | |
| "epoch": 0.27625, | |
| "grad_norm": 2.3533928394317627, | |
| "learning_rate": 8.449361426915797e-05, | |
| "loss": 1.9291694641113282, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 0.276875, | |
| "grad_norm": 2.2795844078063965, | |
| "learning_rate": 8.44206502509605e-05, | |
| "loss": 2.004873046875, | |
| "step": 11075 | |
| }, | |
| { | |
| "epoch": 0.2775, | |
| "grad_norm": 2.0435261726379395, | |
| "learning_rate": 8.43475466378886e-05, | |
| "loss": 1.9740420532226564, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 0.278125, | |
| "grad_norm": 1.5292131900787354, | |
| "learning_rate": 8.427430372641808e-05, | |
| "loss": 1.9866587829589843, | |
| "step": 11125 | |
| }, | |
| { | |
| "epoch": 0.27875, | |
| "grad_norm": 3.218453884124756, | |
| "learning_rate": 8.420092181358967e-05, | |
| "loss": 1.9022023010253906, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 0.279375, | |
| "grad_norm": 2.010437250137329, | |
| "learning_rate": 8.412740119700781e-05, | |
| "loss": 1.881462860107422, | |
| "step": 11175 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 2.4542250633239746, | |
| "learning_rate": 8.405374217483949e-05, | |
| "loss": 1.8804313659667968, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.280625, | |
| "grad_norm": 1.5241410732269287, | |
| "learning_rate": 8.397994504581298e-05, | |
| "loss": 1.8881832885742187, | |
| "step": 11225 | |
| }, | |
| { | |
| "epoch": 0.28125, | |
| "grad_norm": 1.9741950035095215, | |
| "learning_rate": 8.39060101092167e-05, | |
| "loss": 1.7537062072753906, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 0.281875, | |
| "grad_norm": 2.614842414855957, | |
| "learning_rate": 8.383193766489792e-05, | |
| "loss": 1.8642085266113282, | |
| "step": 11275 | |
| }, | |
| { | |
| "epoch": 0.2825, | |
| "grad_norm": 3.313495635986328, | |
| "learning_rate": 8.375772801326156e-05, | |
| "loss": 1.9018295288085938, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 0.283125, | |
| "grad_norm": 1.7892448902130127, | |
| "learning_rate": 8.368338145526904e-05, | |
| "loss": 1.924811553955078, | |
| "step": 11325 | |
| }, | |
| { | |
| "epoch": 0.28375, | |
| "grad_norm": 2.3369665145874023, | |
| "learning_rate": 8.360889829243697e-05, | |
| "loss": 1.8512675476074218, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 0.284375, | |
| "grad_norm": 2.5320141315460205, | |
| "learning_rate": 8.3534278826836e-05, | |
| "loss": 1.9888609313964845, | |
| "step": 11375 | |
| }, | |
| { | |
| "epoch": 0.285, | |
| "grad_norm": 1.639463186264038, | |
| "learning_rate": 8.345952336108955e-05, | |
| "loss": 1.8716566467285156, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 0.285625, | |
| "grad_norm": 1.7238328456878662, | |
| "learning_rate": 8.338463219837262e-05, | |
| "loss": 1.9238868713378907, | |
| "step": 11425 | |
| }, | |
| { | |
| "epoch": 0.28625, | |
| "grad_norm": 2.405167818069458, | |
| "learning_rate": 8.330960564241046e-05, | |
| "loss": 2.0168470764160156, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 0.286875, | |
| "grad_norm": 2.518434762954712, | |
| "learning_rate": 8.323444399747752e-05, | |
| "loss": 2.033546142578125, | |
| "step": 11475 | |
| }, | |
| { | |
| "epoch": 0.2875, | |
| "grad_norm": 2.7626497745513916, | |
| "learning_rate": 8.315914756839603e-05, | |
| "loss": 2.000341491699219, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.288125, | |
| "grad_norm": 2.116691827774048, | |
| "learning_rate": 8.308371666053489e-05, | |
| "loss": 1.9898493957519532, | |
| "step": 11525 | |
| }, | |
| { | |
| "epoch": 0.28875, | |
| "grad_norm": 1.8617552518844604, | |
| "learning_rate": 8.300815157980835e-05, | |
| "loss": 2.0167723083496094, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 0.289375, | |
| "grad_norm": 2.133197069168091, | |
| "learning_rate": 8.293245263267486e-05, | |
| "loss": 2.0451339721679687, | |
| "step": 11575 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 1.948728322982788, | |
| "learning_rate": 8.285662012613568e-05, | |
| "loss": 2.022440948486328, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 0.290625, | |
| "grad_norm": 3.426309823989868, | |
| "learning_rate": 8.278065436773386e-05, | |
| "loss": 1.930302734375, | |
| "step": 11625 | |
| }, | |
| { | |
| "epoch": 0.29125, | |
| "grad_norm": 1.868631362915039, | |
| "learning_rate": 8.270455566555275e-05, | |
| "loss": 1.9666001892089844, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 0.291875, | |
| "grad_norm": 2.4487411975860596, | |
| "learning_rate": 8.262832432821489e-05, | |
| "loss": 1.9045167541503907, | |
| "step": 11675 | |
| }, | |
| { | |
| "epoch": 0.2925, | |
| "grad_norm": 1.806212306022644, | |
| "learning_rate": 8.255196066488075e-05, | |
| "loss": 1.893499755859375, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 0.293125, | |
| "grad_norm": 2.955141067504883, | |
| "learning_rate": 8.247546498524743e-05, | |
| "loss": 1.9781475830078126, | |
| "step": 11725 | |
| }, | |
| { | |
| "epoch": 0.29375, | |
| "grad_norm": 2.9367880821228027, | |
| "learning_rate": 8.239883759954745e-05, | |
| "loss": 2.0095152282714843, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 0.294375, | |
| "grad_norm": 2.740344762802124, | |
| "learning_rate": 8.232207881854748e-05, | |
| "loss": 2.0785810852050783, | |
| "step": 11775 | |
| }, | |
| { | |
| "epoch": 0.295, | |
| "grad_norm": 1.9431589841842651, | |
| "learning_rate": 8.224518895354702e-05, | |
| "loss": 1.9623390197753907, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 0.295625, | |
| "grad_norm": 4.219827175140381, | |
| "learning_rate": 8.216816831637726e-05, | |
| "loss": 2.0050750732421876, | |
| "step": 11825 | |
| }, | |
| { | |
| "epoch": 0.29625, | |
| "grad_norm": 2.83176326751709, | |
| "learning_rate": 8.209101721939966e-05, | |
| "loss": 2.0024925231933595, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 0.296875, | |
| "grad_norm": 2.0029642581939697, | |
| "learning_rate": 8.201373597550488e-05, | |
| "loss": 2.063137664794922, | |
| "step": 11875 | |
| }, | |
| { | |
| "epoch": 0.2975, | |
| "grad_norm": 3.0763931274414062, | |
| "learning_rate": 8.193632489811127e-05, | |
| "loss": 2.044586486816406, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 0.298125, | |
| "grad_norm": 2.1960737705230713, | |
| "learning_rate": 8.185878430116381e-05, | |
| "loss": 2.029036407470703, | |
| "step": 11925 | |
| }, | |
| { | |
| "epoch": 0.29875, | |
| "grad_norm": 2.070204257965088, | |
| "learning_rate": 8.178111449913275e-05, | |
| "loss": 1.8954537963867188, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 0.299375, | |
| "grad_norm": 3.1767876148223877, | |
| "learning_rate": 8.17033158070123e-05, | |
| "loss": 1.8598477172851562, | |
| "step": 11975 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 1.634502649307251, | |
| "learning_rate": 8.16253885403194e-05, | |
| "loss": 1.7688037109375, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "eval_loss": 0.9235685467720032, | |
| "eval_runtime": 24.0451, | |
| "eval_samples_per_second": 60.553, | |
| "eval_steps_per_second": 3.785, | |
| "eval_wer": 0.3608205805717641, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.300625, | |
| "grad_norm": 2.6982874870300293, | |
| "learning_rate": 8.154733301509248e-05, | |
| "loss": 1.8867987060546876, | |
| "step": 12025 | |
| }, | |
| { | |
| "epoch": 0.30125, | |
| "grad_norm": 3.5996062755584717, | |
| "learning_rate": 8.146914954789008e-05, | |
| "loss": 1.9817445373535156, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 0.301875, | |
| "grad_norm": 2.2620458602905273, | |
| "learning_rate": 8.139083845578961e-05, | |
| "loss": 1.7980337524414063, | |
| "step": 12075 | |
| }, | |
| { | |
| "epoch": 0.3025, | |
| "grad_norm": 1.638851523399353, | |
| "learning_rate": 8.131240005638611e-05, | |
| "loss": 1.8636534118652344, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 0.303125, | |
| "grad_norm": 2.6652605533599854, | |
| "learning_rate": 8.12338346677909e-05, | |
| "loss": 1.964078369140625, | |
| "step": 12125 | |
| }, | |
| { | |
| "epoch": 0.30375, | |
| "grad_norm": 2.1387414932250977, | |
| "learning_rate": 8.11551426086303e-05, | |
| "loss": 1.8092033386230468, | |
| "step": 12150 | |
| }, | |
| { | |
| "epoch": 0.304375, | |
| "grad_norm": 1.9871318340301514, | |
| "learning_rate": 8.107632419804437e-05, | |
| "loss": 1.9469615173339845, | |
| "step": 12175 | |
| }, | |
| { | |
| "epoch": 0.305, | |
| "grad_norm": 2.1735482215881348, | |
| "learning_rate": 8.099737975568558e-05, | |
| "loss": 1.8296586608886718, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 0.305625, | |
| "grad_norm": 14.150854110717773, | |
| "learning_rate": 8.091830960171754e-05, | |
| "loss": 1.886156005859375, | |
| "step": 12225 | |
| }, | |
| { | |
| "epoch": 0.30625, | |
| "grad_norm": 2.708627939224243, | |
| "learning_rate": 8.08391140568137e-05, | |
| "loss": 1.7853367614746094, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 0.306875, | |
| "grad_norm": 2.0412747859954834, | |
| "learning_rate": 8.075979344215602e-05, | |
| "loss": 1.834159393310547, | |
| "step": 12275 | |
| }, | |
| { | |
| "epoch": 0.3075, | |
| "grad_norm": 3.147289991378784, | |
| "learning_rate": 8.068034807943368e-05, | |
| "loss": 1.9557830810546875, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 0.308125, | |
| "grad_norm": 3.0972049236297607, | |
| "learning_rate": 8.06007782908418e-05, | |
| "loss": 1.9587071228027344, | |
| "step": 12325 | |
| }, | |
| { | |
| "epoch": 0.30875, | |
| "grad_norm": 3.056018352508545, | |
| "learning_rate": 8.052108439908013e-05, | |
| "loss": 1.9297842407226562, | |
| "step": 12350 | |
| }, | |
| { | |
| "epoch": 0.309375, | |
| "grad_norm": 3.0825061798095703, | |
| "learning_rate": 8.044126672735171e-05, | |
| "loss": 1.9639511108398438, | |
| "step": 12375 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 2.7678534984588623, | |
| "learning_rate": 8.036132559936154e-05, | |
| "loss": 1.922163848876953, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 0.310625, | |
| "grad_norm": 2.881577253341675, | |
| "learning_rate": 8.02812613393154e-05, | |
| "loss": 1.8822640991210937, | |
| "step": 12425 | |
| }, | |
| { | |
| "epoch": 0.31125, | |
| "grad_norm": 1.7253997325897217, | |
| "learning_rate": 8.020107427191831e-05, | |
| "loss": 1.911553955078125, | |
| "step": 12450 | |
| }, | |
| { | |
| "epoch": 0.311875, | |
| "grad_norm": 2.472416639328003, | |
| "learning_rate": 8.012076472237349e-05, | |
| "loss": 2.17446533203125, | |
| "step": 12475 | |
| }, | |
| { | |
| "epoch": 0.3125, | |
| "grad_norm": 1.9672973155975342, | |
| "learning_rate": 8.004033301638075e-05, | |
| "loss": 2.20583740234375, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.313125, | |
| "grad_norm": 2.044351577758789, | |
| "learning_rate": 7.99597794801354e-05, | |
| "loss": 1.9255964660644531, | |
| "step": 12525 | |
| }, | |
| { | |
| "epoch": 0.31375, | |
| "grad_norm": 2.3398983478546143, | |
| "learning_rate": 7.987910444032683e-05, | |
| "loss": 1.8248521423339843, | |
| "step": 12550 | |
| }, | |
| { | |
| "epoch": 0.314375, | |
| "grad_norm": 2.231580972671509, | |
| "learning_rate": 7.979830822413717e-05, | |
| "loss": 1.8463557434082032, | |
| "step": 12575 | |
| }, | |
| { | |
| "epoch": 0.315, | |
| "grad_norm": 2.2503669261932373, | |
| "learning_rate": 7.971739115923999e-05, | |
| "loss": 1.8358750915527344, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 0.315625, | |
| "grad_norm": 2.612706184387207, | |
| "learning_rate": 7.963635357379899e-05, | |
| "loss": 2.061701202392578, | |
| "step": 12625 | |
| }, | |
| { | |
| "epoch": 0.31625, | |
| "grad_norm": 3.270247220993042, | |
| "learning_rate": 7.955519579646664e-05, | |
| "loss": 1.9802578735351561, | |
| "step": 12650 | |
| }, | |
| { | |
| "epoch": 0.316875, | |
| "grad_norm": 2.07159423828125, | |
| "learning_rate": 7.947391815638284e-05, | |
| "loss": 1.9522477722167968, | |
| "step": 12675 | |
| }, | |
| { | |
| "epoch": 0.3175, | |
| "grad_norm": 2.17724609375, | |
| "learning_rate": 7.93925209831736e-05, | |
| "loss": 1.9776730346679687, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 0.318125, | |
| "grad_norm": 2.6313347816467285, | |
| "learning_rate": 7.931100460694974e-05, | |
| "loss": 2.030944061279297, | |
| "step": 12725 | |
| }, | |
| { | |
| "epoch": 0.31875, | |
| "grad_norm": 1.989803671836853, | |
| "learning_rate": 7.922936935830545e-05, | |
| "loss": 1.8712763977050781, | |
| "step": 12750 | |
| }, | |
| { | |
| "epoch": 0.319375, | |
| "grad_norm": 2.5703608989715576, | |
| "learning_rate": 7.91476155683171e-05, | |
| "loss": 1.8772845458984375, | |
| "step": 12775 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 2.570967435836792, | |
| "learning_rate": 7.906574356854173e-05, | |
| "loss": 2.0224600219726563, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.320625, | |
| "grad_norm": 2.3009731769561768, | |
| "learning_rate": 7.898375369101583e-05, | |
| "loss": 1.9474807739257813, | |
| "step": 12825 | |
| }, | |
| { | |
| "epoch": 0.32125, | |
| "grad_norm": 2.1485612392425537, | |
| "learning_rate": 7.890164626825393e-05, | |
| "loss": 1.9762039184570312, | |
| "step": 12850 | |
| }, | |
| { | |
| "epoch": 0.321875, | |
| "grad_norm": 3.6888511180877686, | |
| "learning_rate": 7.88194216332473e-05, | |
| "loss": 2.025465393066406, | |
| "step": 12875 | |
| }, | |
| { | |
| "epoch": 0.3225, | |
| "grad_norm": 2.779696226119995, | |
| "learning_rate": 7.873708011946253e-05, | |
| "loss": 1.9852255249023438, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 0.323125, | |
| "grad_norm": 2.1718077659606934, | |
| "learning_rate": 7.865462206084024e-05, | |
| "loss": 2.1210531616210937, | |
| "step": 12925 | |
| }, | |
| { | |
| "epoch": 0.32375, | |
| "grad_norm": 3.026165246963501, | |
| "learning_rate": 7.857204779179372e-05, | |
| "loss": 2.065125274658203, | |
| "step": 12950 | |
| }, | |
| { | |
| "epoch": 0.324375, | |
| "grad_norm": 1.970770239830017, | |
| "learning_rate": 7.848935764720754e-05, | |
| "loss": 1.9867930603027344, | |
| "step": 12975 | |
| }, | |
| { | |
| "epoch": 0.325, | |
| "grad_norm": 2.950838088989258, | |
| "learning_rate": 7.84065519624362e-05, | |
| "loss": 2.025637512207031, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.325, | |
| "eval_loss": 0.45645052194595337, | |
| "eval_runtime": 29.4227, | |
| "eval_samples_per_second": 49.486, | |
| "eval_steps_per_second": 3.093, | |
| "eval_wer": 0.24092627252376234, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.325625, | |
| "grad_norm": 2.7446556091308594, | |
| "learning_rate": 7.83236310733028e-05, | |
| "loss": 2.0482838439941404, | |
| "step": 13025 | |
| }, | |
| { | |
| "epoch": 0.32625, | |
| "grad_norm": 2.089822769165039, | |
| "learning_rate": 7.824059531609767e-05, | |
| "loss": 1.8909512329101563, | |
| "step": 13050 | |
| }, | |
| { | |
| "epoch": 0.326875, | |
| "grad_norm": 2.5270745754241943, | |
| "learning_rate": 7.81574450275769e-05, | |
| "loss": 1.9806925964355468, | |
| "step": 13075 | |
| }, | |
| { | |
| "epoch": 0.3275, | |
| "grad_norm": 2.3272225856781006, | |
| "learning_rate": 7.807418054496122e-05, | |
| "loss": 2.088828582763672, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 0.328125, | |
| "grad_norm": 2.0077786445617676, | |
| "learning_rate": 7.799080220593436e-05, | |
| "loss": 1.982272186279297, | |
| "step": 13125 | |
| }, | |
| { | |
| "epoch": 0.32875, | |
| "grad_norm": 2.1527907848358154, | |
| "learning_rate": 7.790731034864187e-05, | |
| "loss": 1.9985891723632812, | |
| "step": 13150 | |
| }, | |
| { | |
| "epoch": 0.329375, | |
| "grad_norm": 2.180427312850952, | |
| "learning_rate": 7.782370531168963e-05, | |
| "loss": 2.1555322265625, | |
| "step": 13175 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 2.9644904136657715, | |
| "learning_rate": 7.773998743414254e-05, | |
| "loss": 2.1708287048339843, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 0.330625, | |
| "grad_norm": 1.5731215476989746, | |
| "learning_rate": 7.765615705552317e-05, | |
| "loss": 2.1018193054199217, | |
| "step": 13225 | |
| }, | |
| { | |
| "epoch": 0.33125, | |
| "grad_norm": 2.096733331680298, | |
| "learning_rate": 7.757221451581029e-05, | |
| "loss": 1.9853762817382812, | |
| "step": 13250 | |
| }, | |
| { | |
| "epoch": 0.331875, | |
| "grad_norm": 3.8913395404815674, | |
| "learning_rate": 7.748816015543758e-05, | |
| "loss": 2.1734413146972655, | |
| "step": 13275 | |
| }, | |
| { | |
| "epoch": 0.3325, | |
| "grad_norm": 4.194560527801514, | |
| "learning_rate": 7.740399431529222e-05, | |
| "loss": 2.0609205627441405, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 0.333125, | |
| "grad_norm": 2.308342933654785, | |
| "learning_rate": 7.731971733671346e-05, | |
| "loss": 2.141099700927734, | |
| "step": 13325 | |
| }, | |
| { | |
| "epoch": 0.33375, | |
| "grad_norm": 2.1439101696014404, | |
| "learning_rate": 7.723532956149133e-05, | |
| "loss": 2.0589324951171877, | |
| "step": 13350 | |
| }, | |
| { | |
| "epoch": 0.334375, | |
| "grad_norm": 1.9901295900344849, | |
| "learning_rate": 7.715083133186514e-05, | |
| "loss": 2.0867576599121094, | |
| "step": 13375 | |
| }, | |
| { | |
| "epoch": 0.335, | |
| "grad_norm": 2.457578659057617, | |
| "learning_rate": 7.706622299052222e-05, | |
| "loss": 2.0497677612304686, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 0.335625, | |
| "grad_norm": 6.67117166519165, | |
| "learning_rate": 7.698150488059645e-05, | |
| "loss": 2.0057133483886718, | |
| "step": 13425 | |
| }, | |
| { | |
| "epoch": 0.33625, | |
| "grad_norm": 2.882488489151001, | |
| "learning_rate": 7.689667734566684e-05, | |
| "loss": 1.9257473754882812, | |
| "step": 13450 | |
| }, | |
| { | |
| "epoch": 0.336875, | |
| "grad_norm": 2.7604854106903076, | |
| "learning_rate": 7.681174072975621e-05, | |
| "loss": 2.0035945129394532, | |
| "step": 13475 | |
| }, | |
| { | |
| "epoch": 0.3375, | |
| "grad_norm": 4.429797172546387, | |
| "learning_rate": 7.67266953773298e-05, | |
| "loss": 1.9608985900878906, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.338125, | |
| "grad_norm": 2.813183307647705, | |
| "learning_rate": 7.664154163329373e-05, | |
| "loss": 2.143636016845703, | |
| "step": 13525 | |
| }, | |
| { | |
| "epoch": 0.33875, | |
| "grad_norm": 2.645264148712158, | |
| "learning_rate": 7.655627984299383e-05, | |
| "loss": 2.034876251220703, | |
| "step": 13550 | |
| }, | |
| { | |
| "epoch": 0.339375, | |
| "grad_norm": 2.545079469680786, | |
| "learning_rate": 7.647091035221408e-05, | |
| "loss": 1.9272055053710937, | |
| "step": 13575 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 4.414487361907959, | |
| "learning_rate": 7.638543350717515e-05, | |
| "loss": 1.862533416748047, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 0.340625, | |
| "grad_norm": 1.9178180694580078, | |
| "learning_rate": 7.629984965453326e-05, | |
| "loss": 1.9168106079101563, | |
| "step": 13625 | |
| }, | |
| { | |
| "epoch": 0.34125, | |
| "grad_norm": 2.3425705432891846, | |
| "learning_rate": 7.621415914137846e-05, | |
| "loss": 2.044597473144531, | |
| "step": 13650 | |
| }, | |
| { | |
| "epoch": 0.341875, | |
| "grad_norm": 2.9883460998535156, | |
| "learning_rate": 7.612836231523345e-05, | |
| "loss": 1.94831787109375, | |
| "step": 13675 | |
| }, | |
| { | |
| "epoch": 0.3425, | |
| "grad_norm": 3.396294593811035, | |
| "learning_rate": 7.604245952405204e-05, | |
| "loss": 2.188429718017578, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 0.343125, | |
| "grad_norm": 1.749306082725525, | |
| "learning_rate": 7.595645111621784e-05, | |
| "loss": 2.307034912109375, | |
| "step": 13725 | |
| }, | |
| { | |
| "epoch": 0.34375, | |
| "grad_norm": 3.2763068675994873, | |
| "learning_rate": 7.587033744054274e-05, | |
| "loss": 1.9605122375488282, | |
| "step": 13750 | |
| }, | |
| { | |
| "epoch": 0.344375, | |
| "grad_norm": 1.9773906469345093, | |
| "learning_rate": 7.57841188462656e-05, | |
| "loss": 2.006392364501953, | |
| "step": 13775 | |
| }, | |
| { | |
| "epoch": 0.345, | |
| "grad_norm": 2.8342483043670654, | |
| "learning_rate": 7.569779568305073e-05, | |
| "loss": 1.936324920654297, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 0.345625, | |
| "grad_norm": 3.287250518798828, | |
| "learning_rate": 7.561136830098654e-05, | |
| "loss": 1.9688336181640624, | |
| "step": 13825 | |
| }, | |
| { | |
| "epoch": 0.34625, | |
| "grad_norm": 3.9808924198150635, | |
| "learning_rate": 7.552483705058415e-05, | |
| "loss": 1.9187359619140625, | |
| "step": 13850 | |
| }, | |
| { | |
| "epoch": 0.346875, | |
| "grad_norm": 2.512099027633667, | |
| "learning_rate": 7.543820228277589e-05, | |
| "loss": 2.004053955078125, | |
| "step": 13875 | |
| }, | |
| { | |
| "epoch": 0.3475, | |
| "grad_norm": 2.283036231994629, | |
| "learning_rate": 7.53514643489139e-05, | |
| "loss": 1.9978959655761719, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 0.348125, | |
| "grad_norm": 2.088773727416992, | |
| "learning_rate": 7.526462360076871e-05, | |
| "loss": 1.9999542236328125, | |
| "step": 13925 | |
| }, | |
| { | |
| "epoch": 0.34875, | |
| "grad_norm": 2.2686469554901123, | |
| "learning_rate": 7.517768039052788e-05, | |
| "loss": 2.0053221130371095, | |
| "step": 13950 | |
| }, | |
| { | |
| "epoch": 0.349375, | |
| "grad_norm": 2.9147212505340576, | |
| "learning_rate": 7.509063507079443e-05, | |
| "loss": 2.1617582702636717, | |
| "step": 13975 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 3.376758575439453, | |
| "learning_rate": 7.500348799458554e-05, | |
| "loss": 2.1277151489257813, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "eval_loss": 0.6547609567642212, | |
| "eval_runtime": 31.7154, | |
| "eval_samples_per_second": 45.908, | |
| "eval_steps_per_second": 2.869, | |
| "eval_wer": 0.30962604132261734, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.350625, | |
| "grad_norm": 2.0201609134674072, | |
| "learning_rate": 7.491623951533106e-05, | |
| "loss": 2.035738220214844, | |
| "step": 14025 | |
| }, | |
| { | |
| "epoch": 0.35125, | |
| "grad_norm": 3.5170347690582275, | |
| "learning_rate": 7.482888998687208e-05, | |
| "loss": 1.9420370483398437, | |
| "step": 14050 | |
| }, | |
| { | |
| "epoch": 0.351875, | |
| "grad_norm": 2.0601696968078613, | |
| "learning_rate": 7.47414397634595e-05, | |
| "loss": 1.9036083984375, | |
| "step": 14075 | |
| }, | |
| { | |
| "epoch": 0.3525, | |
| "grad_norm": 3.323498010635376, | |
| "learning_rate": 7.46538891997526e-05, | |
| "loss": 1.925611572265625, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 0.353125, | |
| "grad_norm": 2.3221187591552734, | |
| "learning_rate": 7.45662386508176e-05, | |
| "loss": 1.8567727661132813, | |
| "step": 14125 | |
| }, | |
| { | |
| "epoch": 0.35375, | |
| "grad_norm": 3.476748466491699, | |
| "learning_rate": 7.44784884721262e-05, | |
| "loss": 1.9608552551269531, | |
| "step": 14150 | |
| }, | |
| { | |
| "epoch": 0.354375, | |
| "grad_norm": 1.93206787109375, | |
| "learning_rate": 7.439063901955417e-05, | |
| "loss": 1.8501052856445312, | |
| "step": 14175 | |
| }, | |
| { | |
| "epoch": 0.355, | |
| "grad_norm": 2.8160643577575684, | |
| "learning_rate": 7.43026906493799e-05, | |
| "loss": 1.8909588623046876, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 0.355625, | |
| "grad_norm": 2.585679292678833, | |
| "learning_rate": 7.421464371828288e-05, | |
| "loss": 1.883029022216797, | |
| "step": 14225 | |
| }, | |
| { | |
| "epoch": 0.35625, | |
| "grad_norm": 3.758340835571289, | |
| "learning_rate": 7.412649858334243e-05, | |
| "loss": 1.9194387817382812, | |
| "step": 14250 | |
| }, | |
| { | |
| "epoch": 0.356875, | |
| "grad_norm": 1.682645320892334, | |
| "learning_rate": 7.403825560203602e-05, | |
| "loss": 1.9219422912597657, | |
| "step": 14275 | |
| }, | |
| { | |
| "epoch": 0.3575, | |
| "grad_norm": 4.176120758056641, | |
| "learning_rate": 7.394991513223806e-05, | |
| "loss": 2.154660186767578, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 0.358125, | |
| "grad_norm": 4.0991597175598145, | |
| "learning_rate": 7.386147753221823e-05, | |
| "loss": 2.2225912475585936, | |
| "step": 14325 | |
| }, | |
| { | |
| "epoch": 0.35875, | |
| "grad_norm": 2.2694151401519775, | |
| "learning_rate": 7.377294316064019e-05, | |
| "loss": 1.8874559020996093, | |
| "step": 14350 | |
| }, | |
| { | |
| "epoch": 0.359375, | |
| "grad_norm": 2.3234915733337402, | |
| "learning_rate": 7.368431237656002e-05, | |
| "loss": 1.901197967529297, | |
| "step": 14375 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 2.327249765396118, | |
| "learning_rate": 7.359558553942485e-05, | |
| "loss": 1.871147918701172, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 0.360625, | |
| "grad_norm": 2.186589241027832, | |
| "learning_rate": 7.350676300907133e-05, | |
| "loss": 1.7369599914550782, | |
| "step": 14425 | |
| }, | |
| { | |
| "epoch": 0.36125, | |
| "grad_norm": 3.2321321964263916, | |
| "learning_rate": 7.34178451457242e-05, | |
| "loss": 1.9725926208496094, | |
| "step": 14450 | |
| }, | |
| { | |
| "epoch": 0.361875, | |
| "grad_norm": 2.7875306606292725, | |
| "learning_rate": 7.332883230999484e-05, | |
| "loss": 2.070450897216797, | |
| "step": 14475 | |
| }, | |
| { | |
| "epoch": 0.3625, | |
| "grad_norm": 3.5993967056274414, | |
| "learning_rate": 7.32397248628798e-05, | |
| "loss": 1.9798077392578124, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.363125, | |
| "grad_norm": 1.9958864450454712, | |
| "learning_rate": 7.315052316575929e-05, | |
| "loss": 1.822576446533203, | |
| "step": 14525 | |
| }, | |
| { | |
| "epoch": 0.36375, | |
| "grad_norm": 1.7966809272766113, | |
| "learning_rate": 7.30612275803958e-05, | |
| "loss": 1.7710894775390624, | |
| "step": 14550 | |
| }, | |
| { | |
| "epoch": 0.364375, | |
| "grad_norm": 2.1134510040283203, | |
| "learning_rate": 7.29718384689326e-05, | |
| "loss": 1.7678350830078124, | |
| "step": 14575 | |
| }, | |
| { | |
| "epoch": 0.365, | |
| "grad_norm": 3.246779203414917, | |
| "learning_rate": 7.28823561938922e-05, | |
| "loss": 1.7613668823242188, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 0.365625, | |
| "grad_norm": 1.8065413236618042, | |
| "learning_rate": 7.279278111817501e-05, | |
| "loss": 1.9480220031738282, | |
| "step": 14625 | |
| }, | |
| { | |
| "epoch": 0.36625, | |
| "grad_norm": 2.787346601486206, | |
| "learning_rate": 7.270311360505773e-05, | |
| "loss": 1.8117738342285157, | |
| "step": 14650 | |
| }, | |
| { | |
| "epoch": 0.366875, | |
| "grad_norm": 2.0104660987854004, | |
| "learning_rate": 7.261335401819202e-05, | |
| "loss": 1.9903297424316406, | |
| "step": 14675 | |
| }, | |
| { | |
| "epoch": 0.3675, | |
| "grad_norm": 3.308652639389038, | |
| "learning_rate": 7.252350272160288e-05, | |
| "loss": 1.934193115234375, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 0.368125, | |
| "grad_norm": 2.2588999271392822, | |
| "learning_rate": 7.243356007968729e-05, | |
| "loss": 1.9136190795898438, | |
| "step": 14725 | |
| }, | |
| { | |
| "epoch": 0.36875, | |
| "grad_norm": 2.5005884170532227, | |
| "learning_rate": 7.234352645721267e-05, | |
| "loss": 1.85359619140625, | |
| "step": 14750 | |
| }, | |
| { | |
| "epoch": 0.369375, | |
| "grad_norm": 1.9150588512420654, | |
| "learning_rate": 7.22534022193154e-05, | |
| "loss": 1.872948455810547, | |
| "step": 14775 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 2.4282660484313965, | |
| "learning_rate": 7.216318773149939e-05, | |
| "loss": 1.891831817626953, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 0.370625, | |
| "grad_norm": 2.173224449157715, | |
| "learning_rate": 7.207288335963456e-05, | |
| "loss": 2.002482147216797, | |
| "step": 14825 | |
| }, | |
| { | |
| "epoch": 0.37125, | |
| "grad_norm": 2.0919675827026367, | |
| "learning_rate": 7.198248946995532e-05, | |
| "loss": 1.9934518432617188, | |
| "step": 14850 | |
| }, | |
| { | |
| "epoch": 0.371875, | |
| "grad_norm": 2.4291343688964844, | |
| "learning_rate": 7.189200642905916e-05, | |
| "loss": 1.8515414428710937, | |
| "step": 14875 | |
| }, | |
| { | |
| "epoch": 0.3725, | |
| "grad_norm": 3.716540575027466, | |
| "learning_rate": 7.18014346039051e-05, | |
| "loss": 1.8777276611328124, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 0.373125, | |
| "grad_norm": 3.1523115634918213, | |
| "learning_rate": 7.171077436181228e-05, | |
| "loss": 1.9176744079589845, | |
| "step": 14925 | |
| }, | |
| { | |
| "epoch": 0.37375, | |
| "grad_norm": 2.2882704734802246, | |
| "learning_rate": 7.162002607045838e-05, | |
| "loss": 2.0611204528808593, | |
| "step": 14950 | |
| }, | |
| { | |
| "epoch": 0.374375, | |
| "grad_norm": 2.4674227237701416, | |
| "learning_rate": 7.152919009787817e-05, | |
| "loss": 1.9902169799804688, | |
| "step": 14975 | |
| }, | |
| { | |
| "epoch": 0.375, | |
| "grad_norm": 2.2611148357391357, | |
| "learning_rate": 7.143826681246203e-05, | |
| "loss": 1.9221986389160157, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.375, | |
| "eval_loss": 0.4132440984249115, | |
| "eval_runtime": 32.3222, | |
| "eval_samples_per_second": 45.046, | |
| "eval_steps_per_second": 2.815, | |
| "eval_wer": 0.23142133656280964, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.375625, | |
| "grad_norm": 2.3310348987579346, | |
| "learning_rate": 7.134725658295442e-05, | |
| "loss": 1.9514202880859375, | |
| "step": 15025 | |
| }, | |
| { | |
| "epoch": 0.37625, | |
| "grad_norm": 2.883848190307617, | |
| "learning_rate": 7.125615977845244e-05, | |
| "loss": 1.9112330627441407, | |
| "step": 15050 | |
| }, | |
| { | |
| "epoch": 0.376875, | |
| "grad_norm": 2.2497003078460693, | |
| "learning_rate": 7.116497676840428e-05, | |
| "loss": 1.925036163330078, | |
| "step": 15075 | |
| }, | |
| { | |
| "epoch": 0.3775, | |
| "grad_norm": 2.333568572998047, | |
| "learning_rate": 7.107370792260773e-05, | |
| "loss": 2.0763809204101564, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 0.378125, | |
| "grad_norm": 2.0863587856292725, | |
| "learning_rate": 7.09823536112087e-05, | |
| "loss": 1.9678395080566407, | |
| "step": 15125 | |
| }, | |
| { | |
| "epoch": 0.37875, | |
| "grad_norm": 3.7279679775238037, | |
| "learning_rate": 7.089091420469974e-05, | |
| "loss": 1.9590451049804687, | |
| "step": 15150 | |
| }, | |
| { | |
| "epoch": 0.379375, | |
| "grad_norm": 2.2482640743255615, | |
| "learning_rate": 7.079939007391848e-05, | |
| "loss": 1.787762451171875, | |
| "step": 15175 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 3.3935935497283936, | |
| "learning_rate": 7.070778159004613e-05, | |
| "loss": 1.7457099914550782, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 0.380625, | |
| "grad_norm": 1.8720303773880005, | |
| "learning_rate": 7.061608912460604e-05, | |
| "loss": 1.6555503845214843, | |
| "step": 15225 | |
| }, | |
| { | |
| "epoch": 0.38125, | |
| "grad_norm": 2.2656326293945312, | |
| "learning_rate": 7.052431304946216e-05, | |
| "loss": 1.8570997619628906, | |
| "step": 15250 | |
| }, | |
| { | |
| "epoch": 0.381875, | |
| "grad_norm": 3.670316457748413, | |
| "learning_rate": 7.043245373681747e-05, | |
| "loss": 1.9945068359375, | |
| "step": 15275 | |
| }, | |
| { | |
| "epoch": 0.3825, | |
| "grad_norm": 3.3342275619506836, | |
| "learning_rate": 7.034051155921255e-05, | |
| "loss": 1.9891676330566406, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 0.383125, | |
| "grad_norm": 2.2034950256347656, | |
| "learning_rate": 7.02484868895241e-05, | |
| "loss": 2.006198272705078, | |
| "step": 15325 | |
| }, | |
| { | |
| "epoch": 0.38375, | |
| "grad_norm": 2.857008218765259, | |
| "learning_rate": 7.015638010096326e-05, | |
| "loss": 1.9208688354492187, | |
| "step": 15350 | |
| }, | |
| { | |
| "epoch": 0.384375, | |
| "grad_norm": 2.5997228622436523, | |
| "learning_rate": 7.006419156707431e-05, | |
| "loss": 2.0539146423339845, | |
| "step": 15375 | |
| }, | |
| { | |
| "epoch": 0.385, | |
| "grad_norm": 2.93363094329834, | |
| "learning_rate": 6.9971921661733e-05, | |
| "loss": 2.182738952636719, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 0.385625, | |
| "grad_norm": 2.759763479232788, | |
| "learning_rate": 6.987957075914509e-05, | |
| "loss": 2.03933349609375, | |
| "step": 15425 | |
| }, | |
| { | |
| "epoch": 0.38625, | |
| "grad_norm": 3.8669092655181885, | |
| "learning_rate": 6.978713923384485e-05, | |
| "loss": 2.0265858459472654, | |
| "step": 15450 | |
| }, | |
| { | |
| "epoch": 0.386875, | |
| "grad_norm": 3.174175262451172, | |
| "learning_rate": 6.969462746069348e-05, | |
| "loss": 2.0877862548828126, | |
| "step": 15475 | |
| }, | |
| { | |
| "epoch": 0.3875, | |
| "grad_norm": 2.275374174118042, | |
| "learning_rate": 6.960203581487769e-05, | |
| "loss": 1.9313957214355468, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.388125, | |
| "grad_norm": 1.7126003503799438, | |
| "learning_rate": 6.950936467190806e-05, | |
| "loss": 2.001475067138672, | |
| "step": 15525 | |
| }, | |
| { | |
| "epoch": 0.38875, | |
| "grad_norm": 2.750932216644287, | |
| "learning_rate": 6.941661440761761e-05, | |
| "loss": 2.0474671936035156, | |
| "step": 15550 | |
| }, | |
| { | |
| "epoch": 0.389375, | |
| "grad_norm": 3.1161842346191406, | |
| "learning_rate": 6.932378539816022e-05, | |
| "loss": 2.0202691650390623, | |
| "step": 15575 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 3.031973123550415, | |
| "learning_rate": 6.923087802000916e-05, | |
| "loss": 2.1009735107421874, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 0.390625, | |
| "grad_norm": 5.697539329528809, | |
| "learning_rate": 6.913789264995548e-05, | |
| "loss": 1.982456512451172, | |
| "step": 15625 | |
| }, | |
| { | |
| "epoch": 0.39125, | |
| "grad_norm": 3.4139108657836914, | |
| "learning_rate": 6.904482966510655e-05, | |
| "loss": 1.8953237915039063, | |
| "step": 15650 | |
| }, | |
| { | |
| "epoch": 0.391875, | |
| "grad_norm": 2.7083370685577393, | |
| "learning_rate": 6.895168944288456e-05, | |
| "loss": 2.0133412170410154, | |
| "step": 15675 | |
| }, | |
| { | |
| "epoch": 0.3925, | |
| "grad_norm": 3.779400110244751, | |
| "learning_rate": 6.885847236102484e-05, | |
| "loss": 2.086522674560547, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 0.393125, | |
| "grad_norm": 4.286654949188232, | |
| "learning_rate": 6.876517879757452e-05, | |
| "loss": 1.9416719055175782, | |
| "step": 15725 | |
| }, | |
| { | |
| "epoch": 0.39375, | |
| "grad_norm": 2.672431468963623, | |
| "learning_rate": 6.867180913089085e-05, | |
| "loss": 1.9716299438476563, | |
| "step": 15750 | |
| }, | |
| { | |
| "epoch": 0.394375, | |
| "grad_norm": 6.711437225341797, | |
| "learning_rate": 6.857836373963975e-05, | |
| "loss": 1.9477621459960937, | |
| "step": 15775 | |
| }, | |
| { | |
| "epoch": 0.395, | |
| "grad_norm": 1.726197600364685, | |
| "learning_rate": 6.848484300279426e-05, | |
| "loss": 2.003711700439453, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 0.395625, | |
| "grad_norm": 7.441131591796875, | |
| "learning_rate": 6.839124729963292e-05, | |
| "loss": 2.0010731506347654, | |
| "step": 15825 | |
| }, | |
| { | |
| "epoch": 0.39625, | |
| "grad_norm": 5.97795295715332, | |
| "learning_rate": 6.829757700973838e-05, | |
| "loss": 1.9998374938964845, | |
| "step": 15850 | |
| }, | |
| { | |
| "epoch": 0.396875, | |
| "grad_norm": 2.555112361907959, | |
| "learning_rate": 6.820383251299574e-05, | |
| "loss": 2.098868865966797, | |
| "step": 15875 | |
| }, | |
| { | |
| "epoch": 0.3975, | |
| "grad_norm": 4.291688442230225, | |
| "learning_rate": 6.811001418959103e-05, | |
| "loss": 1.9726272583007813, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 0.398125, | |
| "grad_norm": 3.521483898162842, | |
| "learning_rate": 6.801612242000974e-05, | |
| "loss": 1.9330564880371093, | |
| "step": 15925 | |
| }, | |
| { | |
| "epoch": 0.39875, | |
| "grad_norm": 3.037064790725708, | |
| "learning_rate": 6.79221575850352e-05, | |
| "loss": 1.903232421875, | |
| "step": 15950 | |
| }, | |
| { | |
| "epoch": 0.399375, | |
| "grad_norm": 3.5307297706604004, | |
| "learning_rate": 6.782812006574703e-05, | |
| "loss": 1.9007217407226562, | |
| "step": 15975 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 3.3679850101470947, | |
| "learning_rate": 6.773401024351971e-05, | |
| "loss": 1.8986062622070312, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 0.36609700322151184, | |
| "eval_runtime": 37.3891, | |
| "eval_samples_per_second": 38.942, | |
| "eval_steps_per_second": 2.434, | |
| "eval_wer": 0.20738375720209917, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.400625, | |
| "grad_norm": 2.296968460083008, | |
| "learning_rate": 6.763982850002085e-05, | |
| "loss": 1.888848876953125, | |
| "step": 16025 | |
| }, | |
| { | |
| "epoch": 0.40125, | |
| "grad_norm": 2.0273826122283936, | |
| "learning_rate": 6.754557521720977e-05, | |
| "loss": 2.0307679748535157, | |
| "step": 16050 | |
| }, | |
| { | |
| "epoch": 0.401875, | |
| "grad_norm": 2.717820167541504, | |
| "learning_rate": 6.745125077733598e-05, | |
| "loss": 2.0951828002929687, | |
| "step": 16075 | |
| }, | |
| { | |
| "epoch": 0.4025, | |
| "grad_norm": 2.5354480743408203, | |
| "learning_rate": 6.735685556293751e-05, | |
| "loss": 2.1294482421875, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 0.403125, | |
| "grad_norm": 2.7916624546051025, | |
| "learning_rate": 6.726238995683942e-05, | |
| "loss": 2.037589416503906, | |
| "step": 16125 | |
| }, | |
| { | |
| "epoch": 0.40375, | |
| "grad_norm": 2.6104416847229004, | |
| "learning_rate": 6.716785434215228e-05, | |
| "loss": 1.9768368530273437, | |
| "step": 16150 | |
| }, | |
| { | |
| "epoch": 0.404375, | |
| "grad_norm": 3.585571765899658, | |
| "learning_rate": 6.707324910227059e-05, | |
| "loss": 1.8193942260742189, | |
| "step": 16175 | |
| }, | |
| { | |
| "epoch": 0.405, | |
| "grad_norm": 2.7855167388916016, | |
| "learning_rate": 6.697857462087116e-05, | |
| "loss": 2.0768838500976563, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 0.405625, | |
| "grad_norm": 5.477653980255127, | |
| "learning_rate": 6.688383128191166e-05, | |
| "loss": 3.293823547363281, | |
| "step": 16225 | |
| }, | |
| { | |
| "epoch": 0.40625, | |
| "grad_norm": 2.6667253971099854, | |
| "learning_rate": 6.678901946962903e-05, | |
| "loss": 2.1456167602539065, | |
| "step": 16250 | |
| }, | |
| { | |
| "epoch": 0.406875, | |
| "grad_norm": 2.2158877849578857, | |
| "learning_rate": 6.669413956853787e-05, | |
| "loss": 1.9632577514648437, | |
| "step": 16275 | |
| }, | |
| { | |
| "epoch": 0.4075, | |
| "grad_norm": 2.7362637519836426, | |
| "learning_rate": 6.659919196342892e-05, | |
| "loss": 1.8887863159179688, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 0.408125, | |
| "grad_norm": 2.202183246612549, | |
| "learning_rate": 6.650417703936753e-05, | |
| "loss": 2.1288677978515627, | |
| "step": 16325 | |
| }, | |
| { | |
| "epoch": 0.40875, | |
| "grad_norm": 3.1673545837402344, | |
| "learning_rate": 6.6409095181692e-05, | |
| "loss": 1.9041790771484375, | |
| "step": 16350 | |
| }, | |
| { | |
| "epoch": 0.409375, | |
| "grad_norm": 2.2806448936462402, | |
| "learning_rate": 6.631394677601219e-05, | |
| "loss": 1.9273904418945313, | |
| "step": 16375 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 3.8558311462402344, | |
| "learning_rate": 6.621873220820773e-05, | |
| "loss": 2.0246847534179686, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 0.410625, | |
| "grad_norm": 4.146193504333496, | |
| "learning_rate": 6.612345186442667e-05, | |
| "loss": 1.922226104736328, | |
| "step": 16425 | |
| }, | |
| { | |
| "epoch": 0.41125, | |
| "grad_norm": 2.7093636989593506, | |
| "learning_rate": 6.602810613108379e-05, | |
| "loss": 1.8548619079589843, | |
| "step": 16450 | |
| }, | |
| { | |
| "epoch": 0.411875, | |
| "grad_norm": 1.6062623262405396, | |
| "learning_rate": 6.5932695394859e-05, | |
| "loss": 1.882491912841797, | |
| "step": 16475 | |
| }, | |
| { | |
| "epoch": 0.4125, | |
| "grad_norm": 2.3171355724334717, | |
| "learning_rate": 6.583722004269592e-05, | |
| "loss": 1.9619441223144531, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.413125, | |
| "grad_norm": 3.6338376998901367, | |
| "learning_rate": 6.574168046180018e-05, | |
| "loss": 1.9642619323730468, | |
| "step": 16525 | |
| }, | |
| { | |
| "epoch": 0.41375, | |
| "grad_norm": 3.814567804336548, | |
| "learning_rate": 6.564607703963788e-05, | |
| "loss": 1.8580174255371094, | |
| "step": 16550 | |
| }, | |
| { | |
| "epoch": 0.414375, | |
| "grad_norm": 2.2072980403900146, | |
| "learning_rate": 6.55504101639341e-05, | |
| "loss": 1.835050048828125, | |
| "step": 16575 | |
| }, | |
| { | |
| "epoch": 0.415, | |
| "grad_norm": 2.4695193767547607, | |
| "learning_rate": 6.545468022267115e-05, | |
| "loss": 1.8095762634277344, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 0.415625, | |
| "grad_norm": 2.1555118560791016, | |
| "learning_rate": 6.535888760408721e-05, | |
| "loss": 1.9073040771484375, | |
| "step": 16625 | |
| }, | |
| { | |
| "epoch": 0.41625, | |
| "grad_norm": 1.855448603630066, | |
| "learning_rate": 6.52630326966746e-05, | |
| "loss": 1.9520869445800781, | |
| "step": 16650 | |
| }, | |
| { | |
| "epoch": 0.416875, | |
| "grad_norm": 7.8992791175842285, | |
| "learning_rate": 6.516711588917824e-05, | |
| "loss": 1.8091944885253906, | |
| "step": 16675 | |
| }, | |
| { | |
| "epoch": 0.4175, | |
| "grad_norm": 2.432709217071533, | |
| "learning_rate": 6.507113757059416e-05, | |
| "loss": 1.8939776611328125, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 0.418125, | |
| "grad_norm": 2.3021738529205322, | |
| "learning_rate": 6.497509813016775e-05, | |
| "loss": 1.9395411682128907, | |
| "step": 16725 | |
| }, | |
| { | |
| "epoch": 0.41875, | |
| "grad_norm": 3.1644694805145264, | |
| "learning_rate": 6.487899795739238e-05, | |
| "loss": 2.019146270751953, | |
| "step": 16750 | |
| }, | |
| { | |
| "epoch": 0.419375, | |
| "grad_norm": 3.328117609024048, | |
| "learning_rate": 6.478283744200767e-05, | |
| "loss": 1.9459144592285156, | |
| "step": 16775 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 2.72804594039917, | |
| "learning_rate": 6.468661697399798e-05, | |
| "loss": 1.9769361877441407, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 0.420625, | |
| "grad_norm": 2.722776174545288, | |
| "learning_rate": 6.459033694359078e-05, | |
| "loss": 1.9012591552734375, | |
| "step": 16825 | |
| }, | |
| { | |
| "epoch": 0.42125, | |
| "grad_norm": 1.6808875799179077, | |
| "learning_rate": 6.449399774125516e-05, | |
| "loss": 1.9437788391113282, | |
| "step": 16850 | |
| }, | |
| { | |
| "epoch": 0.421875, | |
| "grad_norm": 4.818302631378174, | |
| "learning_rate": 6.439759975770013e-05, | |
| "loss": 1.926787109375, | |
| "step": 16875 | |
| }, | |
| { | |
| "epoch": 0.4225, | |
| "grad_norm": 2.7799878120422363, | |
| "learning_rate": 6.43011433838731e-05, | |
| "loss": 1.948600311279297, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 0.423125, | |
| "grad_norm": 3.0944535732269287, | |
| "learning_rate": 6.420462901095831e-05, | |
| "loss": 1.862349090576172, | |
| "step": 16925 | |
| }, | |
| { | |
| "epoch": 0.42375, | |
| "grad_norm": 2.066444158554077, | |
| "learning_rate": 6.410805703037522e-05, | |
| "loss": 1.8931501770019532, | |
| "step": 16950 | |
| }, | |
| { | |
| "epoch": 0.424375, | |
| "grad_norm": 2.870826005935669, | |
| "learning_rate": 6.401142783377685e-05, | |
| "loss": 1.8763572692871093, | |
| "step": 16975 | |
| }, | |
| { | |
| "epoch": 0.425, | |
| "grad_norm": 3.2741239070892334, | |
| "learning_rate": 6.391474181304838e-05, | |
| "loss": 1.9325959777832031, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.425, | |
| "eval_loss": 0.34806883335113525, | |
| "eval_runtime": 33.7846, | |
| "eval_samples_per_second": 43.097, | |
| "eval_steps_per_second": 2.694, | |
| "eval_wer": 0.2039707879188227, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.425625, | |
| "grad_norm": 3.0472190380096436, | |
| "learning_rate": 6.381799936030534e-05, | |
| "loss": 1.7822146606445313, | |
| "step": 17025 | |
| }, | |
| { | |
| "epoch": 0.42625, | |
| "grad_norm": 3.8250162601470947, | |
| "learning_rate": 6.372120086789218e-05, | |
| "loss": 1.9976683044433594, | |
| "step": 17050 | |
| }, | |
| { | |
| "epoch": 0.426875, | |
| "grad_norm": 1.6779402494430542, | |
| "learning_rate": 6.362434672838063e-05, | |
| "loss": 2.000716552734375, | |
| "step": 17075 | |
| }, | |
| { | |
| "epoch": 0.4275, | |
| "grad_norm": 2.088564157485962, | |
| "learning_rate": 6.3527437334568e-05, | |
| "loss": 2.12877685546875, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 0.428125, | |
| "grad_norm": 2.766195774078369, | |
| "learning_rate": 6.343047307947586e-05, | |
| "loss": 1.988256072998047, | |
| "step": 17125 | |
| }, | |
| { | |
| "epoch": 0.42875, | |
| "grad_norm": 2.342296600341797, | |
| "learning_rate": 6.33334543563481e-05, | |
| "loss": 1.9589828491210937, | |
| "step": 17150 | |
| }, | |
| { | |
| "epoch": 0.429375, | |
| "grad_norm": 2.6841084957122803, | |
| "learning_rate": 6.323638155864961e-05, | |
| "loss": 1.979658966064453, | |
| "step": 17175 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 2.6437816619873047, | |
| "learning_rate": 6.313925508006458e-05, | |
| "loss": 1.9324618530273439, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 0.430625, | |
| "grad_norm": 3.6072323322296143, | |
| "learning_rate": 6.304207531449486e-05, | |
| "loss": 2.0506765747070315, | |
| "step": 17225 | |
| }, | |
| { | |
| "epoch": 0.43125, | |
| "grad_norm": 2.2150444984436035, | |
| "learning_rate": 6.294484265605843e-05, | |
| "loss": 1.8644593811035157, | |
| "step": 17250 | |
| }, | |
| { | |
| "epoch": 0.431875, | |
| "grad_norm": 2.2496914863586426, | |
| "learning_rate": 6.284755749908776e-05, | |
| "loss": 2.0084588623046873, | |
| "step": 17275 | |
| }, | |
| { | |
| "epoch": 0.4325, | |
| "grad_norm": 2.4838860034942627, | |
| "learning_rate": 6.275022023812829e-05, | |
| "loss": 1.83822509765625, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 0.433125, | |
| "grad_norm": 2.4967970848083496, | |
| "learning_rate": 6.265283126793671e-05, | |
| "loss": 2.0170634460449217, | |
| "step": 17325 | |
| }, | |
| { | |
| "epoch": 0.43375, | |
| "grad_norm": 4.760284900665283, | |
| "learning_rate": 6.255539098347944e-05, | |
| "loss": 2.009665679931641, | |
| "step": 17350 | |
| }, | |
| { | |
| "epoch": 0.434375, | |
| "grad_norm": 3.9792542457580566, | |
| "learning_rate": 6.245789977993105e-05, | |
| "loss": 2.047353973388672, | |
| "step": 17375 | |
| }, | |
| { | |
| "epoch": 0.435, | |
| "grad_norm": 2.1760575771331787, | |
| "learning_rate": 6.236035805267252e-05, | |
| "loss": 1.9648904418945312, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 0.435625, | |
| "grad_norm": 4.262086391448975, | |
| "learning_rate": 6.226276619728981e-05, | |
| "loss": 1.7895712280273437, | |
| "step": 17425 | |
| }, | |
| { | |
| "epoch": 0.43625, | |
| "grad_norm": 1.6788431406021118, | |
| "learning_rate": 6.216512460957216e-05, | |
| "loss": 1.932099609375, | |
| "step": 17450 | |
| }, | |
| { | |
| "epoch": 0.436875, | |
| "grad_norm": 2.0120656490325928, | |
| "learning_rate": 6.206743368551048e-05, | |
| "loss": 1.8715950012207032, | |
| "step": 17475 | |
| }, | |
| { | |
| "epoch": 0.4375, | |
| "grad_norm": 2.3570568561553955, | |
| "learning_rate": 6.19696938212958e-05, | |
| "loss": 1.8846690368652343, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.438125, | |
| "grad_norm": 3.909029960632324, | |
| "learning_rate": 6.18719054133176e-05, | |
| "loss": 2.0164616394042967, | |
| "step": 17525 | |
| }, | |
| { | |
| "epoch": 0.43875, | |
| "grad_norm": 2.2573139667510986, | |
| "learning_rate": 6.177406885816224e-05, | |
| "loss": 1.8774488830566407, | |
| "step": 17550 | |
| }, | |
| { | |
| "epoch": 0.439375, | |
| "grad_norm": 2.470564603805542, | |
| "learning_rate": 6.167618455261138e-05, | |
| "loss": 2.0835089111328124, | |
| "step": 17575 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 1.713579773902893, | |
| "learning_rate": 6.157825289364026e-05, | |
| "loss": 2.054410095214844, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 0.440625, | |
| "grad_norm": 3.8746795654296875, | |
| "learning_rate": 6.148027427841622e-05, | |
| "loss": 1.911707763671875, | |
| "step": 17625 | |
| }, | |
| { | |
| "epoch": 0.44125, | |
| "grad_norm": 2.102914810180664, | |
| "learning_rate": 6.138224910429703e-05, | |
| "loss": 1.9782815551757813, | |
| "step": 17650 | |
| }, | |
| { | |
| "epoch": 0.441875, | |
| "grad_norm": 4.503527641296387, | |
| "learning_rate": 6.128417776882927e-05, | |
| "loss": 1.965908966064453, | |
| "step": 17675 | |
| }, | |
| { | |
| "epoch": 0.4425, | |
| "grad_norm": 2.3784773349761963, | |
| "learning_rate": 6.118606066974673e-05, | |
| "loss": 1.9344596862792969, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 0.443125, | |
| "grad_norm": 2.994469404220581, | |
| "learning_rate": 6.108789820496878e-05, | |
| "loss": 1.941181640625, | |
| "step": 17725 | |
| }, | |
| { | |
| "epoch": 0.44375, | |
| "grad_norm": 2.103759765625, | |
| "learning_rate": 6.098969077259884e-05, | |
| "loss": 1.7989714050292969, | |
| "step": 17750 | |
| }, | |
| { | |
| "epoch": 0.444375, | |
| "grad_norm": 3.4272022247314453, | |
| "learning_rate": 6.089143877092261e-05, | |
| "loss": 1.9051060485839844, | |
| "step": 17775 | |
| }, | |
| { | |
| "epoch": 0.445, | |
| "grad_norm": 2.121569871902466, | |
| "learning_rate": 6.0793142598406586e-05, | |
| "loss": 1.837896728515625, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 0.445625, | |
| "grad_norm": 3.5464529991149902, | |
| "learning_rate": 6.0694802653696416e-05, | |
| "loss": 1.8274598693847657, | |
| "step": 17825 | |
| }, | |
| { | |
| "epoch": 0.44625, | |
| "grad_norm": 1.8415335416793823, | |
| "learning_rate": 6.0596419335615237e-05, | |
| "loss": 2.152740936279297, | |
| "step": 17850 | |
| }, | |
| { | |
| "epoch": 0.446875, | |
| "grad_norm": 3.0613346099853516, | |
| "learning_rate": 6.049799304316214e-05, | |
| "loss": 1.925858154296875, | |
| "step": 17875 | |
| }, | |
| { | |
| "epoch": 0.4475, | |
| "grad_norm": 2.645794153213501, | |
| "learning_rate": 6.039952417551041e-05, | |
| "loss": 1.9710356140136718, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 0.448125, | |
| "grad_norm": 2.181049108505249, | |
| "learning_rate": 6.030101313200609e-05, | |
| "loss": 1.9547206115722657, | |
| "step": 17925 | |
| }, | |
| { | |
| "epoch": 0.44875, | |
| "grad_norm": 2.5496935844421387, | |
| "learning_rate": 6.020246031216622e-05, | |
| "loss": 1.9724104309082031, | |
| "step": 17950 | |
| }, | |
| { | |
| "epoch": 0.449375, | |
| "grad_norm": 2.428947687149048, | |
| "learning_rate": 6.010386611567729e-05, | |
| "loss": 2.226745910644531, | |
| "step": 17975 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 4.553809642791748, | |
| "learning_rate": 6.000523094239358e-05, | |
| "loss": 1.9936044311523438, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "eval_loss": 0.5246478319168091, | |
| "eval_runtime": 27.8557, | |
| "eval_samples_per_second": 52.269, | |
| "eval_steps_per_second": 3.267, | |
| "eval_wer": 0.25791772175125693, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.450625, | |
| "grad_norm": 2.5203700065612793, | |
| "learning_rate": 5.9906555192335545e-05, | |
| "loss": 1.95156982421875, | |
| "step": 18025 | |
| }, | |
| { | |
| "epoch": 0.45125, | |
| "grad_norm": 1.628963828086853, | |
| "learning_rate": 5.980783926568825e-05, | |
| "loss": 1.9565031433105469, | |
| "step": 18050 | |
| }, | |
| { | |
| "epoch": 0.451875, | |
| "grad_norm": 1.757991909980774, | |
| "learning_rate": 5.970908356279962e-05, | |
| "loss": 1.7542300415039063, | |
| "step": 18075 | |
| }, | |
| { | |
| "epoch": 0.4525, | |
| "grad_norm": 1.6905653476715088, | |
| "learning_rate": 5.961028848417894e-05, | |
| "loss": 1.7848200988769531, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 0.453125, | |
| "grad_norm": 3.066129446029663, | |
| "learning_rate": 5.951145443049524e-05, | |
| "loss": 1.8734425354003905, | |
| "step": 18125 | |
| }, | |
| { | |
| "epoch": 0.45375, | |
| "grad_norm": 3.032662868499756, | |
| "learning_rate": 5.9412581802575484e-05, | |
| "loss": 1.9825595092773438, | |
| "step": 18150 | |
| }, | |
| { | |
| "epoch": 0.454375, | |
| "grad_norm": 3.2917351722717285, | |
| "learning_rate": 5.93136710014032e-05, | |
| "loss": 1.8081863403320313, | |
| "step": 18175 | |
| }, | |
| { | |
| "epoch": 0.455, | |
| "grad_norm": 2.1225852966308594, | |
| "learning_rate": 5.921472242811668e-05, | |
| "loss": 1.7902223205566405, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 0.455625, | |
| "grad_norm": 4.139352321624756, | |
| "learning_rate": 5.9115736484007364e-05, | |
| "loss": 1.7889869689941407, | |
| "step": 18225 | |
| }, | |
| { | |
| "epoch": 0.45625, | |
| "grad_norm": 3.4504477977752686, | |
| "learning_rate": 5.901671357051833e-05, | |
| "loss": 1.8598127746582032, | |
| "step": 18250 | |
| }, | |
| { | |
| "epoch": 0.456875, | |
| "grad_norm": 5.106365203857422, | |
| "learning_rate": 5.8917654089242525e-05, | |
| "loss": 1.9735256958007812, | |
| "step": 18275 | |
| }, | |
| { | |
| "epoch": 0.4575, | |
| "grad_norm": 1.8373972177505493, | |
| "learning_rate": 5.881855844192124e-05, | |
| "loss": 2.1968653869628905, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 0.458125, | |
| "grad_norm": 2.9612114429473877, | |
| "learning_rate": 5.871942703044241e-05, | |
| "loss": 1.9368475341796876, | |
| "step": 18325 | |
| }, | |
| { | |
| "epoch": 0.45875, | |
| "grad_norm": 2.0474605560302734, | |
| "learning_rate": 5.862026025683902e-05, | |
| "loss": 1.9962863159179687, | |
| "step": 18350 | |
| }, | |
| { | |
| "epoch": 0.459375, | |
| "grad_norm": 2.2999298572540283, | |
| "learning_rate": 5.85210585232875e-05, | |
| "loss": 1.839505157470703, | |
| "step": 18375 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 1.9650732278823853, | |
| "learning_rate": 5.8421822232106005e-05, | |
| "loss": 1.8152035522460936, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 0.460625, | |
| "grad_norm": 3.597658157348633, | |
| "learning_rate": 5.8322551785752874e-05, | |
| "loss": 1.7580177307128906, | |
| "step": 18425 | |
| }, | |
| { | |
| "epoch": 0.46125, | |
| "grad_norm": 1.5032695531845093, | |
| "learning_rate": 5.822324758682498e-05, | |
| "loss": 1.8031529235839843, | |
| "step": 18450 | |
| }, | |
| { | |
| "epoch": 0.461875, | |
| "grad_norm": 3.4458670616149902, | |
| "learning_rate": 5.812391003805604e-05, | |
| "loss": 1.8211289978027343, | |
| "step": 18475 | |
| }, | |
| { | |
| "epoch": 0.4625, | |
| "grad_norm": 2.088357448577881, | |
| "learning_rate": 5.802453954231506e-05, | |
| "loss": 1.7136880493164062, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.463125, | |
| "grad_norm": 2.680191993713379, | |
| "learning_rate": 5.792513650260465e-05, | |
| "loss": 1.8173138427734374, | |
| "step": 18525 | |
| }, | |
| { | |
| "epoch": 0.46375, | |
| "grad_norm": 2.9088022708892822, | |
| "learning_rate": 5.782570132205939e-05, | |
| "loss": 2.3521595764160157, | |
| "step": 18550 | |
| }, | |
| { | |
| "epoch": 0.464375, | |
| "grad_norm": 2.4535582065582275, | |
| "learning_rate": 5.772623440394422e-05, | |
| "loss": 2.2033631896972654, | |
| "step": 18575 | |
| }, | |
| { | |
| "epoch": 0.465, | |
| "grad_norm": 2.32865834236145, | |
| "learning_rate": 5.762673615165281e-05, | |
| "loss": 1.983341827392578, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 0.465625, | |
| "grad_norm": 3.0877511501312256, | |
| "learning_rate": 5.752720696870591e-05, | |
| "loss": 2.002871551513672, | |
| "step": 18625 | |
| }, | |
| { | |
| "epoch": 0.46625, | |
| "grad_norm": 2.984201192855835, | |
| "learning_rate": 5.7427647258749674e-05, | |
| "loss": 1.9693428039550782, | |
| "step": 18650 | |
| }, | |
| { | |
| "epoch": 0.466875, | |
| "grad_norm": 2.5635454654693604, | |
| "learning_rate": 5.732805742555407e-05, | |
| "loss": 2.216877288818359, | |
| "step": 18675 | |
| }, | |
| { | |
| "epoch": 0.4675, | |
| "grad_norm": 1.9964343309402466, | |
| "learning_rate": 5.7228437873011255e-05, | |
| "loss": 2.0852287292480467, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 0.468125, | |
| "grad_norm": 3.6324875354766846, | |
| "learning_rate": 5.71287890051339e-05, | |
| "loss": 2.0506072998046876, | |
| "step": 18725 | |
| }, | |
| { | |
| "epoch": 0.46875, | |
| "grad_norm": 1.771398901939392, | |
| "learning_rate": 5.702911122605358e-05, | |
| "loss": 1.8342919921875, | |
| "step": 18750 | |
| }, | |
| { | |
| "epoch": 0.469375, | |
| "grad_norm": 2.2910521030426025, | |
| "learning_rate": 5.692940494001908e-05, | |
| "loss": 1.7564373779296876, | |
| "step": 18775 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 2.2783095836639404, | |
| "learning_rate": 5.6829670551394854e-05, | |
| "loss": 1.7932437133789063, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 0.470625, | |
| "grad_norm": 2.8627684116363525, | |
| "learning_rate": 5.67299084646593e-05, | |
| "loss": 1.9440560913085938, | |
| "step": 18825 | |
| }, | |
| { | |
| "epoch": 0.47125, | |
| "grad_norm": 1.7908530235290527, | |
| "learning_rate": 5.6630119084403124e-05, | |
| "loss": 1.9596897888183593, | |
| "step": 18850 | |
| }, | |
| { | |
| "epoch": 0.471875, | |
| "grad_norm": 1.6843006610870361, | |
| "learning_rate": 5.6530302815327774e-05, | |
| "loss": 1.8270423889160157, | |
| "step": 18875 | |
| }, | |
| { | |
| "epoch": 0.4725, | |
| "grad_norm": 2.073862075805664, | |
| "learning_rate": 5.6430460062243686e-05, | |
| "loss": 1.9874969482421876, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 0.473125, | |
| "grad_norm": 3.3529820442199707, | |
| "learning_rate": 5.633059123006875e-05, | |
| "loss": 1.984564208984375, | |
| "step": 18925 | |
| }, | |
| { | |
| "epoch": 0.47375, | |
| "grad_norm": 1.948522686958313, | |
| "learning_rate": 5.6230696723826624e-05, | |
| "loss": 2.0087545776367186, | |
| "step": 18950 | |
| }, | |
| { | |
| "epoch": 0.474375, | |
| "grad_norm": 2.4165892601013184, | |
| "learning_rate": 5.613077694864505e-05, | |
| "loss": 1.9331997680664061, | |
| "step": 18975 | |
| }, | |
| { | |
| "epoch": 0.475, | |
| "grad_norm": 2.435976505279541, | |
| "learning_rate": 5.603083230975428e-05, | |
| "loss": 1.903342742919922, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.475, | |
| "eval_loss": 0.46978750824928284, | |
| "eval_runtime": 39.4973, | |
| "eval_samples_per_second": 36.863, | |
| "eval_steps_per_second": 2.304, | |
| "eval_wer": 0.23967852031267203, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.475625, | |
| "grad_norm": 7.1493024826049805, | |
| "learning_rate": 5.5930863212485386e-05, | |
| "loss": 1.9395419311523439, | |
| "step": 19025 | |
| }, | |
| { | |
| "epoch": 0.47625, | |
| "grad_norm": 2.653665065765381, | |
| "learning_rate": 5.583087006226865e-05, | |
| "loss": 1.9351077270507813, | |
| "step": 19050 | |
| }, | |
| { | |
| "epoch": 0.476875, | |
| "grad_norm": 4.570038795471191, | |
| "learning_rate": 5.573085326463189e-05, | |
| "loss": 1.712913818359375, | |
| "step": 19075 | |
| }, | |
| { | |
| "epoch": 0.4775, | |
| "grad_norm": 2.1461360454559326, | |
| "learning_rate": 5.5630813225198805e-05, | |
| "loss": 1.9174960327148438, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 0.478125, | |
| "grad_norm": 2.16264009475708, | |
| "learning_rate": 5.553075034968742e-05, | |
| "loss": 1.8888673400878906, | |
| "step": 19125 | |
| }, | |
| { | |
| "epoch": 0.47875, | |
| "grad_norm": 2.3397109508514404, | |
| "learning_rate": 5.543066504390828e-05, | |
| "loss": 1.986513671875, | |
| "step": 19150 | |
| }, | |
| { | |
| "epoch": 0.479375, | |
| "grad_norm": 1.982263445854187, | |
| "learning_rate": 5.5330557713763e-05, | |
| "loss": 1.9120260620117187, | |
| "step": 19175 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 2.3013627529144287, | |
| "learning_rate": 5.5230428765242425e-05, | |
| "loss": 1.7947451782226562, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 0.480625, | |
| "grad_norm": 2.3883516788482666, | |
| "learning_rate": 5.513027860442512e-05, | |
| "loss": 1.871494903564453, | |
| "step": 19225 | |
| }, | |
| { | |
| "epoch": 0.48125, | |
| "grad_norm": 2.542034387588501, | |
| "learning_rate": 5.503010763747568e-05, | |
| "loss": 1.7927992248535156, | |
| "step": 19250 | |
| }, | |
| { | |
| "epoch": 0.481875, | |
| "grad_norm": 2.7753939628601074, | |
| "learning_rate": 5.4929916270643054e-05, | |
| "loss": 1.7810481262207032, | |
| "step": 19275 | |
| }, | |
| { | |
| "epoch": 0.4825, | |
| "grad_norm": 2.444901704788208, | |
| "learning_rate": 5.4829704910258975e-05, | |
| "loss": 1.8474925231933594, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 0.483125, | |
| "grad_norm": 2.5080387592315674, | |
| "learning_rate": 5.472947396273617e-05, | |
| "loss": 1.9183151245117187, | |
| "step": 19325 | |
| }, | |
| { | |
| "epoch": 0.48375, | |
| "grad_norm": 2.0195651054382324, | |
| "learning_rate": 5.46292238345669e-05, | |
| "loss": 2.0274208068847654, | |
| "step": 19350 | |
| }, | |
| { | |
| "epoch": 0.484375, | |
| "grad_norm": 2.062363386154175, | |
| "learning_rate": 5.4528954932321176e-05, | |
| "loss": 2.1351564025878904, | |
| "step": 19375 | |
| }, | |
| { | |
| "epoch": 0.485, | |
| "grad_norm": 3.2832655906677246, | |
| "learning_rate": 5.4428667662645116e-05, | |
| "loss": 2.127137451171875, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 0.485625, | |
| "grad_norm": 3.39137601852417, | |
| "learning_rate": 5.4328362432259374e-05, | |
| "loss": 1.7727845764160157, | |
| "step": 19425 | |
| }, | |
| { | |
| "epoch": 0.48625, | |
| "grad_norm": 2.0740859508514404, | |
| "learning_rate": 5.422803964795744e-05, | |
| "loss": 1.772019805908203, | |
| "step": 19450 | |
| }, | |
| { | |
| "epoch": 0.486875, | |
| "grad_norm": 2.573438882827759, | |
| "learning_rate": 5.4127699716603965e-05, | |
| "loss": 1.8231376647949218, | |
| "step": 19475 | |
| }, | |
| { | |
| "epoch": 0.4875, | |
| "grad_norm": 2.4238181114196777, | |
| "learning_rate": 5.402734304513316e-05, | |
| "loss": 2.066461181640625, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.488125, | |
| "grad_norm": 4.753930568695068, | |
| "learning_rate": 5.392697004054714e-05, | |
| "loss": 1.9029042053222656, | |
| "step": 19525 | |
| }, | |
| { | |
| "epoch": 0.48875, | |
| "grad_norm": 2.0650699138641357, | |
| "learning_rate": 5.382658110991423e-05, | |
| "loss": 1.81725830078125, | |
| "step": 19550 | |
| }, | |
| { | |
| "epoch": 0.489375, | |
| "grad_norm": 3.0433406829833984, | |
| "learning_rate": 5.3726176660367365e-05, | |
| "loss": 1.8258647155761718, | |
| "step": 19575 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 2.2786693572998047, | |
| "learning_rate": 5.3625757099102414e-05, | |
| "loss": 1.857409210205078, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 0.490625, | |
| "grad_norm": 1.7537442445755005, | |
| "learning_rate": 5.352532283337655e-05, | |
| "loss": 2.0061961364746095, | |
| "step": 19625 | |
| }, | |
| { | |
| "epoch": 0.49125, | |
| "grad_norm": 4.444217205047607, | |
| "learning_rate": 5.342487427050651e-05, | |
| "loss": 1.9786724853515625, | |
| "step": 19650 | |
| }, | |
| { | |
| "epoch": 0.491875, | |
| "grad_norm": 1.8723256587982178, | |
| "learning_rate": 5.3324411817867107e-05, | |
| "loss": 1.8531193542480469, | |
| "step": 19675 | |
| }, | |
| { | |
| "epoch": 0.4925, | |
| "grad_norm": 2.9987857341766357, | |
| "learning_rate": 5.322393588288943e-05, | |
| "loss": 1.9916073608398437, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 0.493125, | |
| "grad_norm": 1.7448976039886475, | |
| "learning_rate": 5.312344687305926e-05, | |
| "loss": 1.872205810546875, | |
| "step": 19725 | |
| }, | |
| { | |
| "epoch": 0.49375, | |
| "grad_norm": 1.9898735284805298, | |
| "learning_rate": 5.302294519591541e-05, | |
| "loss": 1.9510159301757812, | |
| "step": 19750 | |
| }, | |
| { | |
| "epoch": 0.494375, | |
| "grad_norm": 7.86880350112915, | |
| "learning_rate": 5.292243125904803e-05, | |
| "loss": 2.0468690490722654, | |
| "step": 19775 | |
| }, | |
| { | |
| "epoch": 0.495, | |
| "grad_norm": 3.264361619949341, | |
| "learning_rate": 5.2821905470097076e-05, | |
| "loss": 1.9322613525390624, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 0.495625, | |
| "grad_norm": 2.9220287799835205, | |
| "learning_rate": 5.272136823675046e-05, | |
| "loss": 1.805190887451172, | |
| "step": 19825 | |
| }, | |
| { | |
| "epoch": 0.49625, | |
| "grad_norm": 1.72073495388031, | |
| "learning_rate": 5.2620819966742584e-05, | |
| "loss": 1.8109715270996094, | |
| "step": 19850 | |
| }, | |
| { | |
| "epoch": 0.496875, | |
| "grad_norm": 3.5298118591308594, | |
| "learning_rate": 5.25202610678526e-05, | |
| "loss": 1.796669464111328, | |
| "step": 19875 | |
| }, | |
| { | |
| "epoch": 0.4975, | |
| "grad_norm": 3.1181745529174805, | |
| "learning_rate": 5.2419691947902704e-05, | |
| "loss": 2.015735778808594, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 0.498125, | |
| "grad_norm": 1.7105647325515747, | |
| "learning_rate": 5.231911301475666e-05, | |
| "loss": 1.8832003784179687, | |
| "step": 19925 | |
| }, | |
| { | |
| "epoch": 0.49875, | |
| "grad_norm": 2.4317970275878906, | |
| "learning_rate": 5.2218524676317916e-05, | |
| "loss": 1.7895272827148438, | |
| "step": 19950 | |
| }, | |
| { | |
| "epoch": 0.499375, | |
| "grad_norm": 1.7500026226043701, | |
| "learning_rate": 5.2117927340528114e-05, | |
| "loss": 1.8329066467285156, | |
| "step": 19975 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 2.4562625885009766, | |
| "learning_rate": 5.2017321415365406e-05, | |
| "loss": 1.8331224060058593, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "eval_loss": 0.446895569562912, | |
| "eval_runtime": 29.0441, | |
| "eval_samples_per_second": 50.131, | |
| "eval_steps_per_second": 3.133, | |
| "eval_wer": 0.21890711585746267, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.500625, | |
| "grad_norm": 3.0509231090545654, | |
| "learning_rate": 5.191670730884274e-05, | |
| "loss": 1.792626953125, | |
| "step": 20025 | |
| }, | |
| { | |
| "epoch": 0.50125, | |
| "grad_norm": 2.143178939819336, | |
| "learning_rate": 5.1816085429006256e-05, | |
| "loss": 1.9094760131835937, | |
| "step": 20050 | |
| }, | |
| { | |
| "epoch": 0.501875, | |
| "grad_norm": 2.9646804332733154, | |
| "learning_rate": 5.171545618393363e-05, | |
| "loss": 1.970606231689453, | |
| "step": 20075 | |
| }, | |
| { | |
| "epoch": 0.5025, | |
| "grad_norm": 2.410238027572632, | |
| "learning_rate": 5.161481998173239e-05, | |
| "loss": 1.8851353454589843, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 0.503125, | |
| "grad_norm": 2.782562255859375, | |
| "learning_rate": 5.1514177230538305e-05, | |
| "loss": 1.8776033020019531, | |
| "step": 20125 | |
| }, | |
| { | |
| "epoch": 0.50375, | |
| "grad_norm": 2.069823980331421, | |
| "learning_rate": 5.141352833851367e-05, | |
| "loss": 1.9081240844726564, | |
| "step": 20150 | |
| }, | |
| { | |
| "epoch": 0.504375, | |
| "grad_norm": 3.1768569946289062, | |
| "learning_rate": 5.131287371384571e-05, | |
| "loss": 1.9029811096191407, | |
| "step": 20175 | |
| }, | |
| { | |
| "epoch": 0.505, | |
| "grad_norm": 2.905313730239868, | |
| "learning_rate": 5.1212213764744895e-05, | |
| "loss": 1.7380809020996093, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 0.505625, | |
| "grad_norm": 3.4937968254089355, | |
| "learning_rate": 5.111154889944327e-05, | |
| "loss": 1.8822885131835938, | |
| "step": 20225 | |
| }, | |
| { | |
| "epoch": 0.50625, | |
| "grad_norm": 3.9621078968048096, | |
| "learning_rate": 5.1010879526192865e-05, | |
| "loss": 2.070146179199219, | |
| "step": 20250 | |
| }, | |
| { | |
| "epoch": 0.506875, | |
| "grad_norm": 3.4634897708892822, | |
| "learning_rate": 5.091020605326392e-05, | |
| "loss": 2.0489952087402346, | |
| "step": 20275 | |
| }, | |
| { | |
| "epoch": 0.5075, | |
| "grad_norm": 1.9321340322494507, | |
| "learning_rate": 5.080952888894337e-05, | |
| "loss": 2.0453794860839842, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 0.508125, | |
| "grad_norm": 3.7005345821380615, | |
| "learning_rate": 5.0708848441533084e-05, | |
| "loss": 1.9638418579101562, | |
| "step": 20325 | |
| }, | |
| { | |
| "epoch": 0.50875, | |
| "grad_norm": 3.204824686050415, | |
| "learning_rate": 5.060816511934824e-05, | |
| "loss": 2.133916931152344, | |
| "step": 20350 | |
| }, | |
| { | |
| "epoch": 0.509375, | |
| "grad_norm": 3.6386449337005615, | |
| "learning_rate": 5.050747933071571e-05, | |
| "loss": 1.8050582885742188, | |
| "step": 20375 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 2.9880363941192627, | |
| "learning_rate": 5.040679148397233e-05, | |
| "loss": 1.8290841674804688, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 0.510625, | |
| "grad_norm": 2.7354419231414795, | |
| "learning_rate": 5.0306101987463324e-05, | |
| "loss": 1.9834516906738282, | |
| "step": 20425 | |
| }, | |
| { | |
| "epoch": 0.51125, | |
| "grad_norm": 1.7301700115203857, | |
| "learning_rate": 5.020541124954056e-05, | |
| "loss": 1.8870535278320313, | |
| "step": 20450 | |
| }, | |
| { | |
| "epoch": 0.511875, | |
| "grad_norm": 3.146287202835083, | |
| "learning_rate": 5.010471967856096e-05, | |
| "loss": 1.7696142578125, | |
| "step": 20475 | |
| }, | |
| { | |
| "epoch": 0.5125, | |
| "grad_norm": 1.9643445014953613, | |
| "learning_rate": 5.0004027682884865e-05, | |
| "loss": 1.7880020141601562, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.513125, | |
| "grad_norm": 4.82666540145874, | |
| "learning_rate": 4.9903335670874275e-05, | |
| "loss": 1.854762725830078, | |
| "step": 20525 | |
| }, | |
| { | |
| "epoch": 0.51375, | |
| "grad_norm": 2.774909019470215, | |
| "learning_rate": 4.980264405089127e-05, | |
| "loss": 1.7687403869628906, | |
| "step": 20550 | |
| }, | |
| { | |
| "epoch": 0.514375, | |
| "grad_norm": 2.035015106201172, | |
| "learning_rate": 4.970195323129638e-05, | |
| "loss": 1.725184783935547, | |
| "step": 20575 | |
| }, | |
| { | |
| "epoch": 0.515, | |
| "grad_norm": 2.2140982151031494, | |
| "learning_rate": 4.960126362044684e-05, | |
| "loss": 1.7654539489746093, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 0.515625, | |
| "grad_norm": 2.018460512161255, | |
| "learning_rate": 4.950057562669501e-05, | |
| "loss": 1.8940281677246094, | |
| "step": 20625 | |
| }, | |
| { | |
| "epoch": 0.51625, | |
| "grad_norm": 1.808265209197998, | |
| "learning_rate": 4.939988965838669e-05, | |
| "loss": 1.7670578002929687, | |
| "step": 20650 | |
| }, | |
| { | |
| "epoch": 0.516875, | |
| "grad_norm": 1.9414933919906616, | |
| "learning_rate": 4.929920612385946e-05, | |
| "loss": 1.820958251953125, | |
| "step": 20675 | |
| }, | |
| { | |
| "epoch": 0.5175, | |
| "grad_norm": 3.9117960929870605, | |
| "learning_rate": 4.919852543144103e-05, | |
| "loss": 1.74396484375, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 0.518125, | |
| "grad_norm": 1.8444843292236328, | |
| "learning_rate": 4.90978479894476e-05, | |
| "loss": 1.6656741333007812, | |
| "step": 20725 | |
| }, | |
| { | |
| "epoch": 0.51875, | |
| "grad_norm": 1.4104043245315552, | |
| "learning_rate": 4.899717420618218e-05, | |
| "loss": 1.7394448852539062, | |
| "step": 20750 | |
| }, | |
| { | |
| "epoch": 0.519375, | |
| "grad_norm": 2.0645296573638916, | |
| "learning_rate": 4.8896504489932896e-05, | |
| "loss": 1.8372018432617188, | |
| "step": 20775 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 2.501368522644043, | |
| "learning_rate": 4.879583924897146e-05, | |
| "loss": 1.8813632202148438, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 0.520625, | |
| "grad_norm": 2.634612798690796, | |
| "learning_rate": 4.869517889155136e-05, | |
| "loss": 1.853408966064453, | |
| "step": 20825 | |
| }, | |
| { | |
| "epoch": 0.52125, | |
| "grad_norm": 1.948259711265564, | |
| "learning_rate": 4.8594523825906326e-05, | |
| "loss": 1.938267059326172, | |
| "step": 20850 | |
| }, | |
| { | |
| "epoch": 0.521875, | |
| "grad_norm": 2.3289828300476074, | |
| "learning_rate": 4.849387446024862e-05, | |
| "loss": 1.7549496459960938, | |
| "step": 20875 | |
| }, | |
| { | |
| "epoch": 0.5225, | |
| "grad_norm": 2.1848928928375244, | |
| "learning_rate": 4.839323120276737e-05, | |
| "loss": 1.8034368896484374, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 0.523125, | |
| "grad_norm": 3.162576675415039, | |
| "learning_rate": 4.8292594461626955e-05, | |
| "loss": 1.8527127075195313, | |
| "step": 20925 | |
| }, | |
| { | |
| "epoch": 0.52375, | |
| "grad_norm": 1.9480483531951904, | |
| "learning_rate": 4.819196464496529e-05, | |
| "loss": 1.9780697631835937, | |
| "step": 20950 | |
| }, | |
| { | |
| "epoch": 0.524375, | |
| "grad_norm": 2.1704161167144775, | |
| "learning_rate": 4.809134216089225e-05, | |
| "loss": 2.045873718261719, | |
| "step": 20975 | |
| }, | |
| { | |
| "epoch": 0.525, | |
| "grad_norm": 2.1766655445098877, | |
| "learning_rate": 4.799072741748796e-05, | |
| "loss": 2.0718917846679688, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.525, | |
| "eval_loss": 0.6117398142814636, | |
| "eval_runtime": 42.9792, | |
| "eval_samples_per_second": 33.877, | |
| "eval_steps_per_second": 2.117, | |
| "eval_wer": 0.27010165510660944, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.525625, | |
| "grad_norm": 4.16271448135376, | |
| "learning_rate": 4.789012082280112e-05, | |
| "loss": 2.076334686279297, | |
| "step": 21025 | |
| }, | |
| { | |
| "epoch": 0.52625, | |
| "grad_norm": 2.1584086418151855, | |
| "learning_rate": 4.7789522784847435e-05, | |
| "loss": 1.7636541748046874, | |
| "step": 21050 | |
| }, | |
| { | |
| "epoch": 0.526875, | |
| "grad_norm": 1.5130492448806763, | |
| "learning_rate": 4.768893371160785e-05, | |
| "loss": 1.698475341796875, | |
| "step": 21075 | |
| }, | |
| { | |
| "epoch": 0.5275, | |
| "grad_norm": 2.353510856628418, | |
| "learning_rate": 4.7588354011027015e-05, | |
| "loss": 1.8005508422851562, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 0.528125, | |
| "grad_norm": 1.8366512060165405, | |
| "learning_rate": 4.748778409101153e-05, | |
| "loss": 1.7288040161132812, | |
| "step": 21125 | |
| }, | |
| { | |
| "epoch": 0.52875, | |
| "grad_norm": 1.7773289680480957, | |
| "learning_rate": 4.738722435942832e-05, | |
| "loss": 1.8257809448242188, | |
| "step": 21150 | |
| }, | |
| { | |
| "epoch": 0.529375, | |
| "grad_norm": 2.2035529613494873, | |
| "learning_rate": 4.7286675224103024e-05, | |
| "loss": 1.736931610107422, | |
| "step": 21175 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 2.286503314971924, | |
| "learning_rate": 4.7186137092818284e-05, | |
| "loss": 1.9492512512207032, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 0.530625, | |
| "grad_norm": 1.6430020332336426, | |
| "learning_rate": 4.7085610373312115e-05, | |
| "loss": 1.8533085632324218, | |
| "step": 21225 | |
| }, | |
| { | |
| "epoch": 0.53125, | |
| "grad_norm": 2.4420385360717773, | |
| "learning_rate": 4.6985095473276276e-05, | |
| "loss": 1.905338897705078, | |
| "step": 21250 | |
| }, | |
| { | |
| "epoch": 0.531875, | |
| "grad_norm": 2.19472336769104, | |
| "learning_rate": 4.688459280035455e-05, | |
| "loss": 1.9446469116210938, | |
| "step": 21275 | |
| }, | |
| { | |
| "epoch": 0.5325, | |
| "grad_norm": 3.2131330966949463, | |
| "learning_rate": 4.678410276214116e-05, | |
| "loss": 1.7598631286621094, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 0.533125, | |
| "grad_norm": 2.1369640827178955, | |
| "learning_rate": 4.668362576617911e-05, | |
| "loss": 1.8427285766601562, | |
| "step": 21325 | |
| }, | |
| { | |
| "epoch": 0.53375, | |
| "grad_norm": 1.9467891454696655, | |
| "learning_rate": 4.658316221995844e-05, | |
| "loss": 1.7026889038085937, | |
| "step": 21350 | |
| }, | |
| { | |
| "epoch": 0.534375, | |
| "grad_norm": 2.351487159729004, | |
| "learning_rate": 4.6482712530914716e-05, | |
| "loss": 1.7875502014160156, | |
| "step": 21375 | |
| }, | |
| { | |
| "epoch": 0.535, | |
| "grad_norm": 2.184201717376709, | |
| "learning_rate": 4.6382277106427266e-05, | |
| "loss": 1.8839862060546875, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 0.535625, | |
| "grad_norm": 1.89774751663208, | |
| "learning_rate": 4.628185635381757e-05, | |
| "loss": 1.726077117919922, | |
| "step": 21425 | |
| }, | |
| { | |
| "epoch": 0.53625, | |
| "grad_norm": 2.249199628829956, | |
| "learning_rate": 4.618145068034764e-05, | |
| "loss": 1.955907440185547, | |
| "step": 21450 | |
| }, | |
| { | |
| "epoch": 0.536875, | |
| "grad_norm": 1.8250970840454102, | |
| "learning_rate": 4.608106049321827e-05, | |
| "loss": 1.8917100524902344, | |
| "step": 21475 | |
| }, | |
| { | |
| "epoch": 0.5375, | |
| "grad_norm": 2.0015861988067627, | |
| "learning_rate": 4.598068619956752e-05, | |
| "loss": 1.7551390075683593, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.538125, | |
| "grad_norm": 2.294790744781494, | |
| "learning_rate": 4.5880328206468935e-05, | |
| "loss": 1.8828746032714845, | |
| "step": 21525 | |
| }, | |
| { | |
| "epoch": 0.53875, | |
| "grad_norm": 1.969814658164978, | |
| "learning_rate": 4.5779986920929986e-05, | |
| "loss": 1.76124755859375, | |
| "step": 21550 | |
| }, | |
| { | |
| "epoch": 0.539375, | |
| "grad_norm": 2.1102406978607178, | |
| "learning_rate": 4.5679662749890376e-05, | |
| "loss": 1.7957994079589843, | |
| "step": 21575 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 1.9850748777389526, | |
| "learning_rate": 4.557935610022038e-05, | |
| "loss": 1.8996784973144532, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 0.540625, | |
| "grad_norm": 1.9119586944580078, | |
| "learning_rate": 4.5479067378719265e-05, | |
| "loss": 1.9267098999023438, | |
| "step": 21625 | |
| }, | |
| { | |
| "epoch": 0.54125, | |
| "grad_norm": 2.546102285385132, | |
| "learning_rate": 4.5378796992113525e-05, | |
| "loss": 1.8885546875, | |
| "step": 21650 | |
| }, | |
| { | |
| "epoch": 0.541875, | |
| "grad_norm": 3.0255777835845947, | |
| "learning_rate": 4.527854534705534e-05, | |
| "loss": 1.8068614196777344, | |
| "step": 21675 | |
| }, | |
| { | |
| "epoch": 0.5425, | |
| "grad_norm": 5.405459403991699, | |
| "learning_rate": 4.5178312850120884e-05, | |
| "loss": 1.7694587707519531, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 0.543125, | |
| "grad_norm": 2.104973077774048, | |
| "learning_rate": 4.507809990780863e-05, | |
| "loss": 1.6944454956054686, | |
| "step": 21725 | |
| }, | |
| { | |
| "epoch": 0.54375, | |
| "grad_norm": 1.6939712762832642, | |
| "learning_rate": 4.49779069265378e-05, | |
| "loss": 1.7271163940429688, | |
| "step": 21750 | |
| }, | |
| { | |
| "epoch": 0.544375, | |
| "grad_norm": 2.5886075496673584, | |
| "learning_rate": 4.487773431264664e-05, | |
| "loss": 1.882836151123047, | |
| "step": 21775 | |
| }, | |
| { | |
| "epoch": 0.545, | |
| "grad_norm": 2.734792470932007, | |
| "learning_rate": 4.4777582472390777e-05, | |
| "loss": 1.8201031494140625, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 0.545625, | |
| "grad_norm": 2.0589780807495117, | |
| "learning_rate": 4.467745181194163e-05, | |
| "loss": 1.8243490600585937, | |
| "step": 21825 | |
| }, | |
| { | |
| "epoch": 0.54625, | |
| "grad_norm": 1.72324800491333, | |
| "learning_rate": 4.457734273738468e-05, | |
| "loss": 1.7165113830566405, | |
| "step": 21850 | |
| }, | |
| { | |
| "epoch": 0.546875, | |
| "grad_norm": 2.2243459224700928, | |
| "learning_rate": 4.447725565471789e-05, | |
| "loss": 1.9087004089355468, | |
| "step": 21875 | |
| }, | |
| { | |
| "epoch": 0.5475, | |
| "grad_norm": 2.5698037147521973, | |
| "learning_rate": 4.437719096985001e-05, | |
| "loss": 1.8424166870117187, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 0.548125, | |
| "grad_norm": 3.057349920272827, | |
| "learning_rate": 4.427714908859903e-05, | |
| "loss": 1.756394805908203, | |
| "step": 21925 | |
| }, | |
| { | |
| "epoch": 0.54875, | |
| "grad_norm": 2.7070930004119873, | |
| "learning_rate": 4.417713041669034e-05, | |
| "loss": 1.7627694702148438, | |
| "step": 21950 | |
| }, | |
| { | |
| "epoch": 0.549375, | |
| "grad_norm": 2.6163322925567627, | |
| "learning_rate": 4.407713535975529e-05, | |
| "loss": 1.731389617919922, | |
| "step": 21975 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 2.1468441486358643, | |
| "learning_rate": 4.397716432332944e-05, | |
| "loss": 1.8486198425292968, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "eval_loss": 0.48784032464027405, | |
| "eval_runtime": 24.9825, | |
| "eval_samples_per_second": 58.281, | |
| "eval_steps_per_second": 3.643, | |
| "eval_wer": 0.2329259789350068, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.550625, | |
| "grad_norm": 1.8794786930084229, | |
| "learning_rate": 4.38772177128509e-05, | |
| "loss": 1.861927490234375, | |
| "step": 22025 | |
| }, | |
| { | |
| "epoch": 0.55125, | |
| "grad_norm": 2.141366720199585, | |
| "learning_rate": 4.377729593365879e-05, | |
| "loss": 1.695946044921875, | |
| "step": 22050 | |
| }, | |
| { | |
| "epoch": 0.551875, | |
| "grad_norm": 9.498908996582031, | |
| "learning_rate": 4.367739939099142e-05, | |
| "loss": 1.6303024291992188, | |
| "step": 22075 | |
| }, | |
| { | |
| "epoch": 0.5525, | |
| "grad_norm": 2.454974889755249, | |
| "learning_rate": 4.3577528489984854e-05, | |
| "loss": 1.8029104614257812, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 0.553125, | |
| "grad_norm": 2.0638253688812256, | |
| "learning_rate": 4.347768363567112e-05, | |
| "loss": 1.8980154418945312, | |
| "step": 22125 | |
| }, | |
| { | |
| "epoch": 0.55375, | |
| "grad_norm": 1.9089977741241455, | |
| "learning_rate": 4.33778652329766e-05, | |
| "loss": 1.9254740905761718, | |
| "step": 22150 | |
| }, | |
| { | |
| "epoch": 0.554375, | |
| "grad_norm": 2.857422351837158, | |
| "learning_rate": 4.327807368672043e-05, | |
| "loss": 1.7814622497558594, | |
| "step": 22175 | |
| }, | |
| { | |
| "epoch": 0.555, | |
| "grad_norm": 2.265509843826294, | |
| "learning_rate": 4.31783094016128e-05, | |
| "loss": 1.8639312744140626, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 0.555625, | |
| "grad_norm": 2.2425179481506348, | |
| "learning_rate": 4.307857278225336e-05, | |
| "loss": 1.7818797302246094, | |
| "step": 22225 | |
| }, | |
| { | |
| "epoch": 0.55625, | |
| "grad_norm": 2.193908214569092, | |
| "learning_rate": 4.297886423312957e-05, | |
| "loss": 1.9276255798339843, | |
| "step": 22250 | |
| }, | |
| { | |
| "epoch": 0.556875, | |
| "grad_norm": 3.2965497970581055, | |
| "learning_rate": 4.287918415861501e-05, | |
| "loss": 1.8666990661621095, | |
| "step": 22275 | |
| }, | |
| { | |
| "epoch": 0.5575, | |
| "grad_norm": 2.3592846393585205, | |
| "learning_rate": 4.277953296296782e-05, | |
| "loss": 1.9179656982421875, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 0.558125, | |
| "grad_norm": 3.051501750946045, | |
| "learning_rate": 4.2679911050328985e-05, | |
| "loss": 1.796717071533203, | |
| "step": 22325 | |
| }, | |
| { | |
| "epoch": 0.55875, | |
| "grad_norm": 1.99236261844635, | |
| "learning_rate": 4.258031882472076e-05, | |
| "loss": 1.8888055419921874, | |
| "step": 22350 | |
| }, | |
| { | |
| "epoch": 0.559375, | |
| "grad_norm": 1.938774824142456, | |
| "learning_rate": 4.248075669004501e-05, | |
| "loss": 1.7764259338378907, | |
| "step": 22375 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 2.437443971633911, | |
| "learning_rate": 4.2381225050081516e-05, | |
| "loss": 1.8339962768554687, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 0.560625, | |
| "grad_norm": 1.8378268480300903, | |
| "learning_rate": 4.228172430848644e-05, | |
| "loss": 1.9310797119140626, | |
| "step": 22425 | |
| }, | |
| { | |
| "epoch": 0.56125, | |
| "grad_norm": 1.8813570737838745, | |
| "learning_rate": 4.218225486879061e-05, | |
| "loss": 1.875206298828125, | |
| "step": 22450 | |
| }, | |
| { | |
| "epoch": 0.561875, | |
| "grad_norm": 2.2281298637390137, | |
| "learning_rate": 4.2082817134397894e-05, | |
| "loss": 1.8059898376464845, | |
| "step": 22475 | |
| }, | |
| { | |
| "epoch": 0.5625, | |
| "grad_norm": 2.3510892391204834, | |
| "learning_rate": 4.198341150858361e-05, | |
| "loss": 2.0157070922851563, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.563125, | |
| "grad_norm": 2.8363654613494873, | |
| "learning_rate": 4.188403839449281e-05, | |
| "loss": 2.1007691955566408, | |
| "step": 22525 | |
| }, | |
| { | |
| "epoch": 0.56375, | |
| "grad_norm": 1.9100419282913208, | |
| "learning_rate": 4.178469819513875e-05, | |
| "loss": 2.174064483642578, | |
| "step": 22550 | |
| }, | |
| { | |
| "epoch": 0.564375, | |
| "grad_norm": 4.716004371643066, | |
| "learning_rate": 4.1685391313401175e-05, | |
| "loss": 1.8730838012695312, | |
| "step": 22575 | |
| }, | |
| { | |
| "epoch": 0.565, | |
| "grad_norm": 2.5461723804473877, | |
| "learning_rate": 4.1586118152024675e-05, | |
| "loss": 1.918593292236328, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 0.565625, | |
| "grad_norm": 2.789107322692871, | |
| "learning_rate": 4.148687911361714e-05, | |
| "loss": 2.014490966796875, | |
| "step": 22625 | |
| }, | |
| { | |
| "epoch": 0.56625, | |
| "grad_norm": 2.1704745292663574, | |
| "learning_rate": 4.138767460064803e-05, | |
| "loss": 2.0495640563964845, | |
| "step": 22650 | |
| }, | |
| { | |
| "epoch": 0.566875, | |
| "grad_norm": 3.1955759525299072, | |
| "learning_rate": 4.128850501544681e-05, | |
| "loss": 1.9131755065917968, | |
| "step": 22675 | |
| }, | |
| { | |
| "epoch": 0.5675, | |
| "grad_norm": 2.6322648525238037, | |
| "learning_rate": 4.11893707602013e-05, | |
| "loss": 1.9004020690917969, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 0.568125, | |
| "grad_norm": 2.262258291244507, | |
| "learning_rate": 4.1090272236955995e-05, | |
| "loss": 1.7801530456542969, | |
| "step": 22725 | |
| }, | |
| { | |
| "epoch": 0.56875, | |
| "grad_norm": 2.216627359390259, | |
| "learning_rate": 4.0991209847610535e-05, | |
| "loss": 1.90194580078125, | |
| "step": 22750 | |
| }, | |
| { | |
| "epoch": 0.569375, | |
| "grad_norm": 2.4264588356018066, | |
| "learning_rate": 4.0892183993917955e-05, | |
| "loss": 1.8057530212402344, | |
| "step": 22775 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 2.169029474258423, | |
| "learning_rate": 4.0793195077483166e-05, | |
| "loss": 1.9591471862792968, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 0.570625, | |
| "grad_norm": 1.7696682214736938, | |
| "learning_rate": 4.069424349976127e-05, | |
| "loss": 1.829831085205078, | |
| "step": 22825 | |
| }, | |
| { | |
| "epoch": 0.57125, | |
| "grad_norm": 2.081209897994995, | |
| "learning_rate": 4.0595329662055914e-05, | |
| "loss": 1.6822006225585937, | |
| "step": 22850 | |
| }, | |
| { | |
| "epoch": 0.571875, | |
| "grad_norm": 2.01155686378479, | |
| "learning_rate": 4.0496453965517724e-05, | |
| "loss": 1.8477774047851563, | |
| "step": 22875 | |
| }, | |
| { | |
| "epoch": 0.5725, | |
| "grad_norm": 1.713213324546814, | |
| "learning_rate": 4.0397616811142594e-05, | |
| "loss": 1.748631134033203, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 0.573125, | |
| "grad_norm": 2.531766891479492, | |
| "learning_rate": 4.029881859977017e-05, | |
| "loss": 1.7651852416992186, | |
| "step": 22925 | |
| }, | |
| { | |
| "epoch": 0.57375, | |
| "grad_norm": 1.8059101104736328, | |
| "learning_rate": 4.0200059732082116e-05, | |
| "loss": 1.7520913696289062, | |
| "step": 22950 | |
| }, | |
| { | |
| "epoch": 0.574375, | |
| "grad_norm": 1.8047270774841309, | |
| "learning_rate": 4.010134060860054e-05, | |
| "loss": 1.6803910827636719, | |
| "step": 22975 | |
| }, | |
| { | |
| "epoch": 0.575, | |
| "grad_norm": 2.000319004058838, | |
| "learning_rate": 4.000266162968639e-05, | |
| "loss": 1.7070904541015626, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.575, | |
| "eval_loss": 0.665295422077179, | |
| "eval_runtime": 29.0696, | |
| "eval_samples_per_second": 50.087, | |
| "eval_steps_per_second": 3.13, | |
| "eval_wer": 0.2782487430731403, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.575625, | |
| "grad_norm": 3.405639410018921, | |
| "learning_rate": 3.99040231955378e-05, | |
| "loss": 1.6465997314453125, | |
| "step": 23025 | |
| }, | |
| { | |
| "epoch": 0.57625, | |
| "grad_norm": 1.4138330221176147, | |
| "learning_rate": 3.980542570618843e-05, | |
| "loss": 1.7943145751953125, | |
| "step": 23050 | |
| }, | |
| { | |
| "epoch": 0.576875, | |
| "grad_norm": 1.4225877523422241, | |
| "learning_rate": 3.9706869561505946e-05, | |
| "loss": 1.8432101440429687, | |
| "step": 23075 | |
| }, | |
| { | |
| "epoch": 0.5775, | |
| "grad_norm": 2.28407621383667, | |
| "learning_rate": 3.9608355161190304e-05, | |
| "loss": 1.8984259033203126, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 0.578125, | |
| "grad_norm": 3.603550672531128, | |
| "learning_rate": 3.950988290477215e-05, | |
| "loss": 1.8615032958984374, | |
| "step": 23125 | |
| }, | |
| { | |
| "epoch": 0.57875, | |
| "grad_norm": 2.479234457015991, | |
| "learning_rate": 3.9411453191611265e-05, | |
| "loss": 1.8217715454101562, | |
| "step": 23150 | |
| }, | |
| { | |
| "epoch": 0.579375, | |
| "grad_norm": 2.031212329864502, | |
| "learning_rate": 3.931306642089486e-05, | |
| "loss": 1.9374737548828125, | |
| "step": 23175 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 2.8353798389434814, | |
| "learning_rate": 3.9214722991635975e-05, | |
| "loss": 1.8460720825195311, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 0.580625, | |
| "grad_norm": 2.416569232940674, | |
| "learning_rate": 3.91164233026719e-05, | |
| "loss": 1.8669856262207032, | |
| "step": 23225 | |
| }, | |
| { | |
| "epoch": 0.58125, | |
| "grad_norm": 2.0849156379699707, | |
| "learning_rate": 3.9018167752662535e-05, | |
| "loss": 1.8928665161132812, | |
| "step": 23250 | |
| }, | |
| { | |
| "epoch": 0.581875, | |
| "grad_norm": 2.3874728679656982, | |
| "learning_rate": 3.8919956740088756e-05, | |
| "loss": 1.8264442443847657, | |
| "step": 23275 | |
| }, | |
| { | |
| "epoch": 0.5825, | |
| "grad_norm": 1.9824775457382202, | |
| "learning_rate": 3.8821790663250834e-05, | |
| "loss": 1.7081889343261718, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 0.583125, | |
| "grad_norm": 2.45715069770813, | |
| "learning_rate": 3.872366992026678e-05, | |
| "loss": 1.6332972717285157, | |
| "step": 23325 | |
| }, | |
| { | |
| "epoch": 0.58375, | |
| "grad_norm": 2.7627036571502686, | |
| "learning_rate": 3.8625594909070754e-05, | |
| "loss": 1.7343116760253907, | |
| "step": 23350 | |
| }, | |
| { | |
| "epoch": 0.584375, | |
| "grad_norm": 2.2500531673431396, | |
| "learning_rate": 3.8527566027411484e-05, | |
| "loss": 1.8161436462402343, | |
| "step": 23375 | |
| }, | |
| { | |
| "epoch": 0.585, | |
| "grad_norm": 2.4250290393829346, | |
| "learning_rate": 3.842958367285056e-05, | |
| "loss": 1.7605868530273439, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 0.585625, | |
| "grad_norm": 1.944176435470581, | |
| "learning_rate": 3.833164824276092e-05, | |
| "loss": 1.9290882873535156, | |
| "step": 23425 | |
| }, | |
| { | |
| "epoch": 0.58625, | |
| "grad_norm": 1.9780529737472534, | |
| "learning_rate": 3.823376013432517e-05, | |
| "loss": 1.7536163330078125, | |
| "step": 23450 | |
| }, | |
| { | |
| "epoch": 0.586875, | |
| "grad_norm": 2.1542751789093018, | |
| "learning_rate": 3.813591974453404e-05, | |
| "loss": 1.834794921875, | |
| "step": 23475 | |
| }, | |
| { | |
| "epoch": 0.5875, | |
| "grad_norm": 1.810913324356079, | |
| "learning_rate": 3.80381274701847e-05, | |
| "loss": 1.7412551879882812, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.588125, | |
| "grad_norm": 2.1213550567626953, | |
| "learning_rate": 3.7940383707879177e-05, | |
| "loss": 1.7780479431152343, | |
| "step": 23525 | |
| }, | |
| { | |
| "epoch": 0.58875, | |
| "grad_norm": 2.1620070934295654, | |
| "learning_rate": 3.78426888540228e-05, | |
| "loss": 1.7828038024902344, | |
| "step": 23550 | |
| }, | |
| { | |
| "epoch": 0.589375, | |
| "grad_norm": 1.9433670043945312, | |
| "learning_rate": 3.774504330482249e-05, | |
| "loss": 1.8109051513671874, | |
| "step": 23575 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 3.386655569076538, | |
| "learning_rate": 3.7647447456285254e-05, | |
| "loss": 1.9007333374023438, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 0.590625, | |
| "grad_norm": 2.147420883178711, | |
| "learning_rate": 3.754990170421654e-05, | |
| "loss": 1.8475360107421874, | |
| "step": 23625 | |
| }, | |
| { | |
| "epoch": 0.59125, | |
| "grad_norm": 1.8655669689178467, | |
| "learning_rate": 3.7452406444218566e-05, | |
| "loss": 1.7568521118164062, | |
| "step": 23650 | |
| }, | |
| { | |
| "epoch": 0.591875, | |
| "grad_norm": 3.003970146179199, | |
| "learning_rate": 3.735496207168885e-05, | |
| "loss": 1.8325260925292968, | |
| "step": 23675 | |
| }, | |
| { | |
| "epoch": 0.5925, | |
| "grad_norm": 2.215942859649658, | |
| "learning_rate": 3.7257568981818494e-05, | |
| "loss": 2.0860830688476564, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 0.593125, | |
| "grad_norm": 2.724135637283325, | |
| "learning_rate": 3.716022756959061e-05, | |
| "loss": 2.035277099609375, | |
| "step": 23725 | |
| }, | |
| { | |
| "epoch": 0.59375, | |
| "grad_norm": 2.5257678031921387, | |
| "learning_rate": 3.706293822977878e-05, | |
| "loss": 2.158614044189453, | |
| "step": 23750 | |
| }, | |
| { | |
| "epoch": 0.594375, | |
| "grad_norm": 2.1608970165252686, | |
| "learning_rate": 3.6965701356945336e-05, | |
| "loss": 1.9955123901367187, | |
| "step": 23775 | |
| }, | |
| { | |
| "epoch": 0.595, | |
| "grad_norm": 2.263308048248291, | |
| "learning_rate": 3.686851734543987e-05, | |
| "loss": 1.8430841064453125, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 0.595625, | |
| "grad_norm": 2.849299907684326, | |
| "learning_rate": 3.6771386589397604e-05, | |
| "loss": 1.9462979125976563, | |
| "step": 23825 | |
| }, | |
| { | |
| "epoch": 0.59625, | |
| "grad_norm": 2.50551176071167, | |
| "learning_rate": 3.6674309482737724e-05, | |
| "loss": 1.8574705505371094, | |
| "step": 23850 | |
| }, | |
| { | |
| "epoch": 0.596875, | |
| "grad_norm": 2.032573938369751, | |
| "learning_rate": 3.657728641916189e-05, | |
| "loss": 1.8497027587890624, | |
| "step": 23875 | |
| }, | |
| { | |
| "epoch": 0.5975, | |
| "grad_norm": 3.0877480506896973, | |
| "learning_rate": 3.648031779215257e-05, | |
| "loss": 1.776781768798828, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 0.598125, | |
| "grad_norm": 3.14872145652771, | |
| "learning_rate": 3.6383403994971456e-05, | |
| "loss": 1.8453164672851563, | |
| "step": 23925 | |
| }, | |
| { | |
| "epoch": 0.59875, | |
| "grad_norm": 1.8324261903762817, | |
| "learning_rate": 3.628654542065788e-05, | |
| "loss": 1.7086810302734374, | |
| "step": 23950 | |
| }, | |
| { | |
| "epoch": 0.599375, | |
| "grad_norm": 2.2475340366363525, | |
| "learning_rate": 3.618974246202721e-05, | |
| "loss": 1.6697857666015625, | |
| "step": 23975 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 1.5676270723342896, | |
| "learning_rate": 3.6092995511669274e-05, | |
| "loss": 1.7643623352050781, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_loss": 0.6699950695037842, | |
| "eval_runtime": 23.6327, | |
| "eval_samples_per_second": 61.61, | |
| "eval_steps_per_second": 3.851, | |
| "eval_wer": 0.28459759991192335, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.600625, | |
| "grad_norm": 2.5892279148101807, | |
| "learning_rate": 3.599630496194673e-05, | |
| "loss": 1.6844937133789062, | |
| "step": 24025 | |
| }, | |
| { | |
| "epoch": 0.60125, | |
| "grad_norm": 2.220022439956665, | |
| "learning_rate": 3.589967120499353e-05, | |
| "loss": 1.7022825622558593, | |
| "step": 24050 | |
| }, | |
| { | |
| "epoch": 0.601875, | |
| "grad_norm": 2.0049166679382324, | |
| "learning_rate": 3.5803094632713296e-05, | |
| "loss": 1.7939231872558594, | |
| "step": 24075 | |
| }, | |
| { | |
| "epoch": 0.6025, | |
| "grad_norm": 3.158663272857666, | |
| "learning_rate": 3.5706575636777706e-05, | |
| "loss": 1.8239601135253907, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 0.603125, | |
| "grad_norm": 2.075392961502075, | |
| "learning_rate": 3.5610114608624965e-05, | |
| "loss": 1.841759033203125, | |
| "step": 24125 | |
| }, | |
| { | |
| "epoch": 0.60375, | |
| "grad_norm": 2.3904948234558105, | |
| "learning_rate": 3.551371193945815e-05, | |
| "loss": 1.698827362060547, | |
| "step": 24150 | |
| }, | |
| { | |
| "epoch": 0.604375, | |
| "grad_norm": 2.668787956237793, | |
| "learning_rate": 3.5417368020243704e-05, | |
| "loss": 1.715019989013672, | |
| "step": 24175 | |
| }, | |
| { | |
| "epoch": 0.605, | |
| "grad_norm": 3.015451431274414, | |
| "learning_rate": 3.532108324170979e-05, | |
| "loss": 1.7710948181152344, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 0.605625, | |
| "grad_norm": 2.4660584926605225, | |
| "learning_rate": 3.5224857994344714e-05, | |
| "loss": 1.7959405517578124, | |
| "step": 24225 | |
| }, | |
| { | |
| "epoch": 0.60625, | |
| "grad_norm": 2.016432523727417, | |
| "learning_rate": 3.5128692668395345e-05, | |
| "loss": 1.7882905578613282, | |
| "step": 24250 | |
| }, | |
| { | |
| "epoch": 0.606875, | |
| "grad_norm": 3.4150125980377197, | |
| "learning_rate": 3.5032587653865564e-05, | |
| "loss": 1.7204815673828124, | |
| "step": 24275 | |
| }, | |
| { | |
| "epoch": 0.6075, | |
| "grad_norm": 3.173091173171997, | |
| "learning_rate": 3.4936543340514615e-05, | |
| "loss": 1.8169828796386718, | |
| "step": 24300 | |
| }, | |
| { | |
| "epoch": 0.608125, | |
| "grad_norm": 2.5396976470947266, | |
| "learning_rate": 3.484056011785562e-05, | |
| "loss": 1.7763470458984374, | |
| "step": 24325 | |
| }, | |
| { | |
| "epoch": 0.60875, | |
| "grad_norm": 1.8674589395523071, | |
| "learning_rate": 3.474463837515386e-05, | |
| "loss": 1.7911065673828126, | |
| "step": 24350 | |
| }, | |
| { | |
| "epoch": 0.609375, | |
| "grad_norm": 1.4210426807403564, | |
| "learning_rate": 3.4648778501425405e-05, | |
| "loss": 1.8412464904785155, | |
| "step": 24375 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 3.335195302963257, | |
| "learning_rate": 3.4552980885435285e-05, | |
| "loss": 1.8631402587890624, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 0.610625, | |
| "grad_norm": 2.155010938644409, | |
| "learning_rate": 3.445724591569614e-05, | |
| "loss": 1.8997987365722657, | |
| "step": 24425 | |
| }, | |
| { | |
| "epoch": 0.61125, | |
| "grad_norm": 2.1205575466156006, | |
| "learning_rate": 3.4361573980466446e-05, | |
| "loss": 1.7354130554199219, | |
| "step": 24450 | |
| }, | |
| { | |
| "epoch": 0.611875, | |
| "grad_norm": 2.228691816329956, | |
| "learning_rate": 3.426596546774914e-05, | |
| "loss": 1.7464279174804687, | |
| "step": 24475 | |
| }, | |
| { | |
| "epoch": 0.6125, | |
| "grad_norm": 2.0042595863342285, | |
| "learning_rate": 3.417042076528988e-05, | |
| "loss": 1.7940188598632814, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.613125, | |
| "grad_norm": 2.696664571762085, | |
| "learning_rate": 3.4074940260575555e-05, | |
| "loss": 1.5943063354492188, | |
| "step": 24525 | |
| }, | |
| { | |
| "epoch": 0.61375, | |
| "grad_norm": 3.257740020751953, | |
| "learning_rate": 3.397952434083272e-05, | |
| "loss": 1.679752197265625, | |
| "step": 24550 | |
| }, | |
| { | |
| "epoch": 0.614375, | |
| "grad_norm": 2.6322507858276367, | |
| "learning_rate": 3.388417339302593e-05, | |
| "loss": 1.6827996826171876, | |
| "step": 24575 | |
| }, | |
| { | |
| "epoch": 0.615, | |
| "grad_norm": 2.9432754516601562, | |
| "learning_rate": 3.378888780385634e-05, | |
| "loss": 1.7791423034667968, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 0.615625, | |
| "grad_norm": 2.477353572845459, | |
| "learning_rate": 3.369366795975998e-05, | |
| "loss": 1.6179205322265624, | |
| "step": 24625 | |
| }, | |
| { | |
| "epoch": 0.61625, | |
| "grad_norm": 2.6826860904693604, | |
| "learning_rate": 3.359851424690625e-05, | |
| "loss": 1.7498304748535156, | |
| "step": 24650 | |
| }, | |
| { | |
| "epoch": 0.616875, | |
| "grad_norm": 1.9868218898773193, | |
| "learning_rate": 3.3503427051196375e-05, | |
| "loss": 1.9038162231445312, | |
| "step": 24675 | |
| }, | |
| { | |
| "epoch": 0.6175, | |
| "grad_norm": 2.116527795791626, | |
| "learning_rate": 3.340840675826178e-05, | |
| "loss": 1.7065106201171876, | |
| "step": 24700 | |
| }, | |
| { | |
| "epoch": 0.618125, | |
| "grad_norm": 1.9596235752105713, | |
| "learning_rate": 3.33134537534626e-05, | |
| "loss": 1.8208642578125, | |
| "step": 24725 | |
| }, | |
| { | |
| "epoch": 0.61875, | |
| "grad_norm": 1.9185731410980225, | |
| "learning_rate": 3.3218568421886064e-05, | |
| "loss": 1.7175167846679686, | |
| "step": 24750 | |
| }, | |
| { | |
| "epoch": 0.619375, | |
| "grad_norm": 2.750948667526245, | |
| "learning_rate": 3.312375114834494e-05, | |
| "loss": 1.686314239501953, | |
| "step": 24775 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 2.953357458114624, | |
| "learning_rate": 3.3029002317376e-05, | |
| "loss": 1.7556192016601562, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 0.620625, | |
| "grad_norm": 1.440243124961853, | |
| "learning_rate": 3.293432231323842e-05, | |
| "loss": 1.8063711547851562, | |
| "step": 24825 | |
| }, | |
| { | |
| "epoch": 0.62125, | |
| "grad_norm": 1.9147006273269653, | |
| "learning_rate": 3.283971151991225e-05, | |
| "loss": 1.7902940368652345, | |
| "step": 24850 | |
| }, | |
| { | |
| "epoch": 0.621875, | |
| "grad_norm": 2.003424882888794, | |
| "learning_rate": 3.27451703210969e-05, | |
| "loss": 1.7132737731933594, | |
| "step": 24875 | |
| }, | |
| { | |
| "epoch": 0.6225, | |
| "grad_norm": 3.51619553565979, | |
| "learning_rate": 3.2650699100209445e-05, | |
| "loss": 1.9095518493652344, | |
| "step": 24900 | |
| }, | |
| { | |
| "epoch": 0.623125, | |
| "grad_norm": 2.542562484741211, | |
| "learning_rate": 3.255629824038324e-05, | |
| "loss": 1.8072769165039062, | |
| "step": 24925 | |
| }, | |
| { | |
| "epoch": 0.62375, | |
| "grad_norm": 3.504107713699341, | |
| "learning_rate": 3.246196812446627e-05, | |
| "loss": 1.7472404479980468, | |
| "step": 24950 | |
| }, | |
| { | |
| "epoch": 0.624375, | |
| "grad_norm": 1.561985969543457, | |
| "learning_rate": 3.2367709135019576e-05, | |
| "loss": 1.731632080078125, | |
| "step": 24975 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 1.9944498538970947, | |
| "learning_rate": 3.227352165431577e-05, | |
| "loss": 1.6879058837890626, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "eval_loss": 0.7341573238372803, | |
| "eval_runtime": 35.5012, | |
| "eval_samples_per_second": 41.013, | |
| "eval_steps_per_second": 2.563, | |
| "eval_wer": 0.28907482843407095, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.625625, | |
| "grad_norm": 2.0819621086120605, | |
| "learning_rate": 3.217940606433747e-05, | |
| "loss": 1.6297799682617187, | |
| "step": 25025 | |
| }, | |
| { | |
| "epoch": 0.62625, | |
| "grad_norm": 1.4772917032241821, | |
| "learning_rate": 3.208536274677571e-05, | |
| "loss": 1.750072021484375, | |
| "step": 25050 | |
| }, | |
| { | |
| "epoch": 0.626875, | |
| "grad_norm": 1.7171095609664917, | |
| "learning_rate": 3.1991392083028424e-05, | |
| "loss": 1.692607421875, | |
| "step": 25075 | |
| }, | |
| { | |
| "epoch": 0.6275, | |
| "grad_norm": 3.3012447357177734, | |
| "learning_rate": 3.1897494454198915e-05, | |
| "loss": 1.7129794311523439, | |
| "step": 25100 | |
| }, | |
| { | |
| "epoch": 0.628125, | |
| "grad_norm": 2.9091410636901855, | |
| "learning_rate": 3.1803670241094274e-05, | |
| "loss": 1.7479110717773438, | |
| "step": 25125 | |
| }, | |
| { | |
| "epoch": 0.62875, | |
| "grad_norm": 2.5538108348846436, | |
| "learning_rate": 3.170991982422383e-05, | |
| "loss": 1.690687255859375, | |
| "step": 25150 | |
| }, | |
| { | |
| "epoch": 0.629375, | |
| "grad_norm": 1.8337676525115967, | |
| "learning_rate": 3.161624358379767e-05, | |
| "loss": 1.7536323547363282, | |
| "step": 25175 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 1.9157038927078247, | |
| "learning_rate": 3.152264189972503e-05, | |
| "loss": 1.7153205871582031, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 0.630625, | |
| "grad_norm": 2.3431942462921143, | |
| "learning_rate": 3.1429115151612766e-05, | |
| "loss": 1.8436314392089843, | |
| "step": 25225 | |
| }, | |
| { | |
| "epoch": 0.63125, | |
| "grad_norm": 2.1703617572784424, | |
| "learning_rate": 3.1335663718763866e-05, | |
| "loss": 1.7437437438964845, | |
| "step": 25250 | |
| }, | |
| { | |
| "epoch": 0.631875, | |
| "grad_norm": 2.3940718173980713, | |
| "learning_rate": 3.1242287980175815e-05, | |
| "loss": 1.858861846923828, | |
| "step": 25275 | |
| }, | |
| { | |
| "epoch": 0.6325, | |
| "grad_norm": 3.592817544937134, | |
| "learning_rate": 3.1148988314539176e-05, | |
| "loss": 1.8692318725585937, | |
| "step": 25300 | |
| }, | |
| { | |
| "epoch": 0.633125, | |
| "grad_norm": 2.8741536140441895, | |
| "learning_rate": 3.105576510023597e-05, | |
| "loss": 1.754287567138672, | |
| "step": 25325 | |
| }, | |
| { | |
| "epoch": 0.63375, | |
| "grad_norm": 2.2680890560150146, | |
| "learning_rate": 3.096261871533813e-05, | |
| "loss": 1.8966297912597656, | |
| "step": 25350 | |
| }, | |
| { | |
| "epoch": 0.634375, | |
| "grad_norm": 1.9580203294754028, | |
| "learning_rate": 3.086954953760608e-05, | |
| "loss": 1.7048651123046874, | |
| "step": 25375 | |
| }, | |
| { | |
| "epoch": 0.635, | |
| "grad_norm": 2.508688449859619, | |
| "learning_rate": 3.077655794448705e-05, | |
| "loss": 1.696322479248047, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 0.635625, | |
| "grad_norm": 1.6616238355636597, | |
| "learning_rate": 3.0683644313113656e-05, | |
| "loss": 1.7671804809570313, | |
| "step": 25425 | |
| }, | |
| { | |
| "epoch": 0.63625, | |
| "grad_norm": 2.7090115547180176, | |
| "learning_rate": 3.0590809020302346e-05, | |
| "loss": 1.6990518188476562, | |
| "step": 25450 | |
| }, | |
| { | |
| "epoch": 0.636875, | |
| "grad_norm": 2.2064476013183594, | |
| "learning_rate": 3.049805244255184e-05, | |
| "loss": 1.7052224731445313, | |
| "step": 25475 | |
| }, | |
| { | |
| "epoch": 0.6375, | |
| "grad_norm": 2.7846221923828125, | |
| "learning_rate": 3.0405374956041626e-05, | |
| "loss": 1.8141432189941407, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.638125, | |
| "grad_norm": 2.3835833072662354, | |
| "learning_rate": 3.031277693663046e-05, | |
| "loss": 1.7668656921386718, | |
| "step": 25525 | |
| }, | |
| { | |
| "epoch": 0.63875, | |
| "grad_norm": 2.583956718444824, | |
| "learning_rate": 3.022025875985476e-05, | |
| "loss": 1.854783477783203, | |
| "step": 25550 | |
| }, | |
| { | |
| "epoch": 0.639375, | |
| "grad_norm": 2.1010680198669434, | |
| "learning_rate": 3.012782080092721e-05, | |
| "loss": 1.7635113525390624, | |
| "step": 25575 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 2.0765833854675293, | |
| "learning_rate": 3.0035463434735106e-05, | |
| "loss": 1.8019525146484374, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.640625, | |
| "grad_norm": 2.1619768142700195, | |
| "learning_rate": 2.994318703583894e-05, | |
| "loss": 1.8397747802734374, | |
| "step": 25625 | |
| }, | |
| { | |
| "epoch": 0.64125, | |
| "grad_norm": 2.1369686126708984, | |
| "learning_rate": 2.985099197847078e-05, | |
| "loss": 1.5887631225585936, | |
| "step": 25650 | |
| }, | |
| { | |
| "epoch": 0.641875, | |
| "grad_norm": 3.028017520904541, | |
| "learning_rate": 2.9758878636532883e-05, | |
| "loss": 1.825139617919922, | |
| "step": 25675 | |
| }, | |
| { | |
| "epoch": 0.6425, | |
| "grad_norm": 3.987513303756714, | |
| "learning_rate": 2.966684738359603e-05, | |
| "loss": 1.7027491760253906, | |
| "step": 25700 | |
| }, | |
| { | |
| "epoch": 0.643125, | |
| "grad_norm": 1.7240517139434814, | |
| "learning_rate": 2.9574898592898138e-05, | |
| "loss": 1.640452880859375, | |
| "step": 25725 | |
| }, | |
| { | |
| "epoch": 0.64375, | |
| "grad_norm": 4.836753845214844, | |
| "learning_rate": 2.9483032637342677e-05, | |
| "loss": 1.7357638549804688, | |
| "step": 25750 | |
| }, | |
| { | |
| "epoch": 0.644375, | |
| "grad_norm": 5.317409038543701, | |
| "learning_rate": 2.9391249889497163e-05, | |
| "loss": 1.7839738464355468, | |
| "step": 25775 | |
| }, | |
| { | |
| "epoch": 0.645, | |
| "grad_norm": 4.556936740875244, | |
| "learning_rate": 2.929955072159169e-05, | |
| "loss": 1.905267333984375, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 0.645625, | |
| "grad_norm": 2.5061700344085693, | |
| "learning_rate": 2.920793550551734e-05, | |
| "loss": 1.9363316345214843, | |
| "step": 25825 | |
| }, | |
| { | |
| "epoch": 0.64625, | |
| "grad_norm": 2.9503777027130127, | |
| "learning_rate": 2.9116404612824775e-05, | |
| "loss": 1.951572265625, | |
| "step": 25850 | |
| }, | |
| { | |
| "epoch": 0.646875, | |
| "grad_norm": 2.187394380569458, | |
| "learning_rate": 2.902495841472264e-05, | |
| "loss": 1.892421875, | |
| "step": 25875 | |
| }, | |
| { | |
| "epoch": 0.6475, | |
| "grad_norm": 2.206587791442871, | |
| "learning_rate": 2.893359728207612e-05, | |
| "loss": 1.7759242248535156, | |
| "step": 25900 | |
| }, | |
| { | |
| "epoch": 0.648125, | |
| "grad_norm": 2.526491165161133, | |
| "learning_rate": 2.8842321585405396e-05, | |
| "loss": 1.769729461669922, | |
| "step": 25925 | |
| }, | |
| { | |
| "epoch": 0.64875, | |
| "grad_norm": 3.713080644607544, | |
| "learning_rate": 2.875113169488417e-05, | |
| "loss": 1.8624252319335937, | |
| "step": 25950 | |
| }, | |
| { | |
| "epoch": 0.649375, | |
| "grad_norm": 2.3911359310150146, | |
| "learning_rate": 2.8660027980338177e-05, | |
| "loss": 1.9479046630859376, | |
| "step": 25975 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 2.115046262741089, | |
| "learning_rate": 2.856901081124359e-05, | |
| "loss": 1.9840438842773438, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "eval_loss": 0.827679455280304, | |
| "eval_runtime": 21.4437, | |
| "eval_samples_per_second": 67.899, | |
| "eval_steps_per_second": 4.244, | |
| "eval_wer": 0.30995632867261186, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.650625, | |
| "grad_norm": 1.836103081703186, | |
| "learning_rate": 2.8478080556725652e-05, | |
| "loss": 1.8618795776367187, | |
| "step": 26025 | |
| }, | |
| { | |
| "epoch": 0.65125, | |
| "grad_norm": 4.286990165710449, | |
| "learning_rate": 2.838723758555709e-05, | |
| "loss": 1.850547332763672, | |
| "step": 26050 | |
| }, | |
| { | |
| "epoch": 0.651875, | |
| "grad_norm": 2.2095324993133545, | |
| "learning_rate": 2.829648226615668e-05, | |
| "loss": 1.9137684631347656, | |
| "step": 26075 | |
| }, | |
| { | |
| "epoch": 0.6525, | |
| "grad_norm": 2.2275943756103516, | |
| "learning_rate": 2.820581496658772e-05, | |
| "loss": 1.9542002868652344, | |
| "step": 26100 | |
| }, | |
| { | |
| "epoch": 0.653125, | |
| "grad_norm": 2.6691653728485107, | |
| "learning_rate": 2.811523605455646e-05, | |
| "loss": 1.8251153564453124, | |
| "step": 26125 | |
| }, | |
| { | |
| "epoch": 0.65375, | |
| "grad_norm": 3.110239267349243, | |
| "learning_rate": 2.8024745897410787e-05, | |
| "loss": 1.920596923828125, | |
| "step": 26150 | |
| }, | |
| { | |
| "epoch": 0.654375, | |
| "grad_norm": 2.1628358364105225, | |
| "learning_rate": 2.79343448621386e-05, | |
| "loss": 1.9261192321777343, | |
| "step": 26175 | |
| }, | |
| { | |
| "epoch": 0.655, | |
| "grad_norm": 2.831921339035034, | |
| "learning_rate": 2.7844033315366358e-05, | |
| "loss": 1.8953877258300782, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 0.655625, | |
| "grad_norm": 2.8279213905334473, | |
| "learning_rate": 2.7753811623357606e-05, | |
| "loss": 1.96121337890625, | |
| "step": 26225 | |
| }, | |
| { | |
| "epoch": 0.65625, | |
| "grad_norm": 2.546863317489624, | |
| "learning_rate": 2.766368015201144e-05, | |
| "loss": 1.8773268127441407, | |
| "step": 26250 | |
| }, | |
| { | |
| "epoch": 0.656875, | |
| "grad_norm": 2.338458299636841, | |
| "learning_rate": 2.75736392668611e-05, | |
| "loss": 2.0234925842285154, | |
| "step": 26275 | |
| }, | |
| { | |
| "epoch": 0.6575, | |
| "grad_norm": 3.0420849323272705, | |
| "learning_rate": 2.7483689333072443e-05, | |
| "loss": 1.8467626953125, | |
| "step": 26300 | |
| }, | |
| { | |
| "epoch": 0.658125, | |
| "grad_norm": 2.235715389251709, | |
| "learning_rate": 2.739383071544246e-05, | |
| "loss": 1.9796235656738281, | |
| "step": 26325 | |
| }, | |
| { | |
| "epoch": 0.65875, | |
| "grad_norm": 2.7309744358062744, | |
| "learning_rate": 2.7304063778397827e-05, | |
| "loss": 1.902901153564453, | |
| "step": 26350 | |
| }, | |
| { | |
| "epoch": 0.659375, | |
| "grad_norm": 2.214077949523926, | |
| "learning_rate": 2.7214388885993348e-05, | |
| "loss": 1.9776980590820312, | |
| "step": 26375 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 3.80493426322937, | |
| "learning_rate": 2.712480640191058e-05, | |
| "loss": 1.8304763793945313, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 0.660625, | |
| "grad_norm": 2.5608863830566406, | |
| "learning_rate": 2.7035316689456314e-05, | |
| "loss": 1.779889373779297, | |
| "step": 26425 | |
| }, | |
| { | |
| "epoch": 0.66125, | |
| "grad_norm": 2.7427258491516113, | |
| "learning_rate": 2.6945920111561073e-05, | |
| "loss": 1.9308509826660156, | |
| "step": 26450 | |
| }, | |
| { | |
| "epoch": 0.661875, | |
| "grad_norm": 1.943137526512146, | |
| "learning_rate": 2.685661703077773e-05, | |
| "loss": 1.8887556457519532, | |
| "step": 26475 | |
| }, | |
| { | |
| "epoch": 0.6625, | |
| "grad_norm": 3.2523791790008545, | |
| "learning_rate": 2.6767407809279865e-05, | |
| "loss": 1.8608908081054687, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.663125, | |
| "grad_norm": 2.2498466968536377, | |
| "learning_rate": 2.6678292808860506e-05, | |
| "loss": 1.970323028564453, | |
| "step": 26525 | |
| }, | |
| { | |
| "epoch": 0.66375, | |
| "grad_norm": 4.641754150390625, | |
| "learning_rate": 2.658927239093052e-05, | |
| "loss": 1.825666961669922, | |
| "step": 26550 | |
| }, | |
| { | |
| "epoch": 0.664375, | |
| "grad_norm": 2.3471319675445557, | |
| "learning_rate": 2.6500346916517205e-05, | |
| "loss": 1.772312774658203, | |
| "step": 26575 | |
| }, | |
| { | |
| "epoch": 0.665, | |
| "grad_norm": 2.5294320583343506, | |
| "learning_rate": 2.641151674626281e-05, | |
| "loss": 1.8614224243164061, | |
| "step": 26600 | |
| }, | |
| { | |
| "epoch": 0.665625, | |
| "grad_norm": 2.1994786262512207, | |
| "learning_rate": 2.6322782240423027e-05, | |
| "loss": 1.8166519165039063, | |
| "step": 26625 | |
| }, | |
| { | |
| "epoch": 0.66625, | |
| "grad_norm": 4.34580135345459, | |
| "learning_rate": 2.6234143758865638e-05, | |
| "loss": 1.8048843383789062, | |
| "step": 26650 | |
| }, | |
| { | |
| "epoch": 0.666875, | |
| "grad_norm": 2.131984233856201, | |
| "learning_rate": 2.614560166106897e-05, | |
| "loss": 1.9018528747558594, | |
| "step": 26675 | |
| }, | |
| { | |
| "epoch": 0.6675, | |
| "grad_norm": 2.473973274230957, | |
| "learning_rate": 2.605715630612045e-05, | |
| "loss": 1.827135009765625, | |
| "step": 26700 | |
| }, | |
| { | |
| "epoch": 0.668125, | |
| "grad_norm": 2.944078207015991, | |
| "learning_rate": 2.5968808052715193e-05, | |
| "loss": 1.8921070861816407, | |
| "step": 26725 | |
| }, | |
| { | |
| "epoch": 0.66875, | |
| "grad_norm": 2.477018117904663, | |
| "learning_rate": 2.5880557259154446e-05, | |
| "loss": 1.894139862060547, | |
| "step": 26750 | |
| }, | |
| { | |
| "epoch": 0.669375, | |
| "grad_norm": 4.036627292633057, | |
| "learning_rate": 2.5792404283344245e-05, | |
| "loss": 1.7410826110839843, | |
| "step": 26775 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 2.961561679840088, | |
| "learning_rate": 2.5704349482793955e-05, | |
| "loss": 1.8269784545898438, | |
| "step": 26800 | |
| }, | |
| { | |
| "epoch": 0.670625, | |
| "grad_norm": 2.2599332332611084, | |
| "learning_rate": 2.561639321461476e-05, | |
| "loss": 1.8808795166015626, | |
| "step": 26825 | |
| }, | |
| { | |
| "epoch": 0.67125, | |
| "grad_norm": 3.052338123321533, | |
| "learning_rate": 2.5528535835518175e-05, | |
| "loss": 1.8674078369140625, | |
| "step": 26850 | |
| }, | |
| { | |
| "epoch": 0.671875, | |
| "grad_norm": 2.3569788932800293, | |
| "learning_rate": 2.5440777701814755e-05, | |
| "loss": 1.7916732788085938, | |
| "step": 26875 | |
| }, | |
| { | |
| "epoch": 0.6725, | |
| "grad_norm": 2.763045310974121, | |
| "learning_rate": 2.535311916941252e-05, | |
| "loss": 1.849720458984375, | |
| "step": 26900 | |
| }, | |
| { | |
| "epoch": 0.673125, | |
| "grad_norm": 2.0208351612091064, | |
| "learning_rate": 2.5265560593815556e-05, | |
| "loss": 1.8993585205078125, | |
| "step": 26925 | |
| }, | |
| { | |
| "epoch": 0.67375, | |
| "grad_norm": 3.555210828781128, | |
| "learning_rate": 2.5178102330122588e-05, | |
| "loss": 1.94588134765625, | |
| "step": 26950 | |
| }, | |
| { | |
| "epoch": 0.674375, | |
| "grad_norm": 2.8405017852783203, | |
| "learning_rate": 2.509074473302546e-05, | |
| "loss": 1.9300169372558593, | |
| "step": 26975 | |
| }, | |
| { | |
| "epoch": 0.675, | |
| "grad_norm": 3.01963210105896, | |
| "learning_rate": 2.500348815680783e-05, | |
| "loss": 1.751306915283203, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.675, | |
| "eval_loss": 0.6867467761039734, | |
| "eval_runtime": 25.5797, | |
| "eval_samples_per_second": 56.92, | |
| "eval_steps_per_second": 3.558, | |
| "eval_wer": 0.28323975191750156, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.675625, | |
| "grad_norm": 1.9335639476776123, | |
| "learning_rate": 2.4916332955343614e-05, | |
| "loss": 1.8480300903320312, | |
| "step": 27025 | |
| }, | |
| { | |
| "epoch": 0.67625, | |
| "grad_norm": 2.7521188259124756, | |
| "learning_rate": 2.4829279482095614e-05, | |
| "loss": 1.8683226013183594, | |
| "step": 27050 | |
| }, | |
| { | |
| "epoch": 0.676875, | |
| "grad_norm": 3.2874202728271484, | |
| "learning_rate": 2.474232809011407e-05, | |
| "loss": 1.8606121826171875, | |
| "step": 27075 | |
| }, | |
| { | |
| "epoch": 0.6775, | |
| "grad_norm": 2.4721004962921143, | |
| "learning_rate": 2.4655479132035242e-05, | |
| "loss": 1.9349343872070313, | |
| "step": 27100 | |
| }, | |
| { | |
| "epoch": 0.678125, | |
| "grad_norm": 1.3699150085449219, | |
| "learning_rate": 2.4568732960079897e-05, | |
| "loss": 1.9246978759765625, | |
| "step": 27125 | |
| }, | |
| { | |
| "epoch": 0.67875, | |
| "grad_norm": 4.582528591156006, | |
| "learning_rate": 2.4482089926052014e-05, | |
| "loss": 1.7549371337890625, | |
| "step": 27150 | |
| }, | |
| { | |
| "epoch": 0.679375, | |
| "grad_norm": 1.8700374364852905, | |
| "learning_rate": 2.4395550381337273e-05, | |
| "loss": 1.879395751953125, | |
| "step": 27175 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 1.7924561500549316, | |
| "learning_rate": 2.4309114676901646e-05, | |
| "loss": 1.6684425354003907, | |
| "step": 27200 | |
| }, | |
| { | |
| "epoch": 0.680625, | |
| "grad_norm": 2.1560938358306885, | |
| "learning_rate": 2.422278316328998e-05, | |
| "loss": 1.7351014709472656, | |
| "step": 27225 | |
| }, | |
| { | |
| "epoch": 0.68125, | |
| "grad_norm": 2.2467801570892334, | |
| "learning_rate": 2.4136556190624526e-05, | |
| "loss": 1.7653411865234374, | |
| "step": 27250 | |
| }, | |
| { | |
| "epoch": 0.681875, | |
| "grad_norm": 2.138880729675293, | |
| "learning_rate": 2.4050434108603633e-05, | |
| "loss": 1.7972740173339843, | |
| "step": 27275 | |
| }, | |
| { | |
| "epoch": 0.6825, | |
| "grad_norm": 3.2346391677856445, | |
| "learning_rate": 2.396441726650021e-05, | |
| "loss": 1.755618896484375, | |
| "step": 27300 | |
| }, | |
| { | |
| "epoch": 0.683125, | |
| "grad_norm": 4.690395832061768, | |
| "learning_rate": 2.3878506013160383e-05, | |
| "loss": 1.7412397766113281, | |
| "step": 27325 | |
| }, | |
| { | |
| "epoch": 0.68375, | |
| "grad_norm": 3.0066471099853516, | |
| "learning_rate": 2.3792700697002074e-05, | |
| "loss": 1.820947265625, | |
| "step": 27350 | |
| }, | |
| { | |
| "epoch": 0.684375, | |
| "grad_norm": 2.272731304168701, | |
| "learning_rate": 2.37070016660135e-05, | |
| "loss": 1.803546142578125, | |
| "step": 27375 | |
| }, | |
| { | |
| "epoch": 0.685, | |
| "grad_norm": 1.9828675985336304, | |
| "learning_rate": 2.362140926775191e-05, | |
| "loss": 1.8425863647460938, | |
| "step": 27400 | |
| }, | |
| { | |
| "epoch": 0.685625, | |
| "grad_norm": 1.9557849168777466, | |
| "learning_rate": 2.3535923849342056e-05, | |
| "loss": 1.8902931213378906, | |
| "step": 27425 | |
| }, | |
| { | |
| "epoch": 0.68625, | |
| "grad_norm": 2.228438138961792, | |
| "learning_rate": 2.3450545757474844e-05, | |
| "loss": 1.7733322143554688, | |
| "step": 27450 | |
| }, | |
| { | |
| "epoch": 0.686875, | |
| "grad_norm": 2.263880968093872, | |
| "learning_rate": 2.336527533840593e-05, | |
| "loss": 1.762078094482422, | |
| "step": 27475 | |
| }, | |
| { | |
| "epoch": 0.6875, | |
| "grad_norm": 6.303035736083984, | |
| "learning_rate": 2.328011293795423e-05, | |
| "loss": 1.72776123046875, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.688125, | |
| "grad_norm": 2.3560125827789307, | |
| "learning_rate": 2.3195058901500645e-05, | |
| "loss": 1.8493244934082032, | |
| "step": 27525 | |
| }, | |
| { | |
| "epoch": 0.68875, | |
| "grad_norm": 2.6900041103363037, | |
| "learning_rate": 2.3110113573986588e-05, | |
| "loss": 1.77732177734375, | |
| "step": 27550 | |
| }, | |
| { | |
| "epoch": 0.689375, | |
| "grad_norm": 1.8254210948944092, | |
| "learning_rate": 2.3025277299912583e-05, | |
| "loss": 1.7760365295410157, | |
| "step": 27575 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 2.400632381439209, | |
| "learning_rate": 2.2940550423336915e-05, | |
| "loss": 1.6966725158691407, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 0.690625, | |
| "grad_norm": 1.8421419858932495, | |
| "learning_rate": 2.2855933287874138e-05, | |
| "loss": 1.870133819580078, | |
| "step": 27625 | |
| }, | |
| { | |
| "epoch": 0.69125, | |
| "grad_norm": 3.155991554260254, | |
| "learning_rate": 2.27714262366938e-05, | |
| "loss": 1.7450431823730468, | |
| "step": 27650 | |
| }, | |
| { | |
| "epoch": 0.691875, | |
| "grad_norm": 1.9954915046691895, | |
| "learning_rate": 2.2687029612518978e-05, | |
| "loss": 1.9261622619628906, | |
| "step": 27675 | |
| }, | |
| { | |
| "epoch": 0.6925, | |
| "grad_norm": 1.9402109384536743, | |
| "learning_rate": 2.260274375762491e-05, | |
| "loss": 1.7911344909667968, | |
| "step": 27700 | |
| }, | |
| { | |
| "epoch": 0.693125, | |
| "grad_norm": 1.8402515649795532, | |
| "learning_rate": 2.2518569013837615e-05, | |
| "loss": 1.7637394714355468, | |
| "step": 27725 | |
| }, | |
| { | |
| "epoch": 0.69375, | |
| "grad_norm": 2.1865599155426025, | |
| "learning_rate": 2.243450572253245e-05, | |
| "loss": 1.7278936767578126, | |
| "step": 27750 | |
| }, | |
| { | |
| "epoch": 0.694375, | |
| "grad_norm": 2.1965880393981934, | |
| "learning_rate": 2.235055422463281e-05, | |
| "loss": 1.5395339965820312, | |
| "step": 27775 | |
| }, | |
| { | |
| "epoch": 0.695, | |
| "grad_norm": 2.4809231758117676, | |
| "learning_rate": 2.2266714860608696e-05, | |
| "loss": 1.8909536743164062, | |
| "step": 27800 | |
| }, | |
| { | |
| "epoch": 0.695625, | |
| "grad_norm": 2.097362518310547, | |
| "learning_rate": 2.2182987970475343e-05, | |
| "loss": 1.8717481994628906, | |
| "step": 27825 | |
| }, | |
| { | |
| "epoch": 0.69625, | |
| "grad_norm": 2.9639782905578613, | |
| "learning_rate": 2.209937389379187e-05, | |
| "loss": 1.833187255859375, | |
| "step": 27850 | |
| }, | |
| { | |
| "epoch": 0.696875, | |
| "grad_norm": 1.578460454940796, | |
| "learning_rate": 2.20158729696598e-05, | |
| "loss": 1.871891326904297, | |
| "step": 27875 | |
| }, | |
| { | |
| "epoch": 0.6975, | |
| "grad_norm": 2.9352340698242188, | |
| "learning_rate": 2.193248553672183e-05, | |
| "loss": 1.8578401184082032, | |
| "step": 27900 | |
| }, | |
| { | |
| "epoch": 0.698125, | |
| "grad_norm": 2.3028459548950195, | |
| "learning_rate": 2.1849211933160356e-05, | |
| "loss": 1.9112147521972656, | |
| "step": 27925 | |
| }, | |
| { | |
| "epoch": 0.69875, | |
| "grad_norm": 3.4521656036376953, | |
| "learning_rate": 2.1766052496696153e-05, | |
| "loss": 1.9455368041992187, | |
| "step": 27950 | |
| }, | |
| { | |
| "epoch": 0.699375, | |
| "grad_norm": 3.0296742916107178, | |
| "learning_rate": 2.1683007564586977e-05, | |
| "loss": 1.917243194580078, | |
| "step": 27975 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 4.724900245666504, | |
| "learning_rate": 2.160007747362615e-05, | |
| "loss": 1.7917201232910156, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_loss": 0.5828148722648621, | |
| "eval_runtime": 26.0047, | |
| "eval_samples_per_second": 55.99, | |
| "eval_steps_per_second": 3.499, | |
| "eval_wer": 0.26298212778450586, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.700625, | |
| "grad_norm": 2.025892496109009, | |
| "learning_rate": 2.151726256014136e-05, | |
| "loss": 1.7265559387207032, | |
| "step": 28025 | |
| }, | |
| { | |
| "epoch": 0.70125, | |
| "grad_norm": 2.0473244190216064, | |
| "learning_rate": 2.1434563159993097e-05, | |
| "loss": 1.6775180053710939, | |
| "step": 28050 | |
| }, | |
| { | |
| "epoch": 0.701875, | |
| "grad_norm": 2.3128581047058105, | |
| "learning_rate": 2.1351979608573438e-05, | |
| "loss": 1.7089291381835938, | |
| "step": 28075 | |
| }, | |
| { | |
| "epoch": 0.7025, | |
| "grad_norm": 4.393394470214844, | |
| "learning_rate": 2.1269512240804558e-05, | |
| "loss": 1.7424867248535156, | |
| "step": 28100 | |
| }, | |
| { | |
| "epoch": 0.703125, | |
| "grad_norm": 1.9760547876358032, | |
| "learning_rate": 2.11871613911375e-05, | |
| "loss": 1.7922735595703125, | |
| "step": 28125 | |
| }, | |
| { | |
| "epoch": 0.70375, | |
| "grad_norm": 2.7836265563964844, | |
| "learning_rate": 2.110492739355075e-05, | |
| "loss": 1.894503631591797, | |
| "step": 28150 | |
| }, | |
| { | |
| "epoch": 0.704375, | |
| "grad_norm": 2.5852749347686768, | |
| "learning_rate": 2.1022810581548887e-05, | |
| "loss": 1.788990478515625, | |
| "step": 28175 | |
| }, | |
| { | |
| "epoch": 0.705, | |
| "grad_norm": 3.10060453414917, | |
| "learning_rate": 2.094081128816126e-05, | |
| "loss": 1.8902725219726562, | |
| "step": 28200 | |
| }, | |
| { | |
| "epoch": 0.705625, | |
| "grad_norm": 2.3038454055786133, | |
| "learning_rate": 2.085892984594056e-05, | |
| "loss": 1.8767176818847657, | |
| "step": 28225 | |
| }, | |
| { | |
| "epoch": 0.70625, | |
| "grad_norm": 3.018732786178589, | |
| "learning_rate": 2.0777166586961576e-05, | |
| "loss": 1.8796475219726563, | |
| "step": 28250 | |
| }, | |
| { | |
| "epoch": 0.706875, | |
| "grad_norm": 2.4248321056365967, | |
| "learning_rate": 2.0695521842819788e-05, | |
| "loss": 1.7550675964355469, | |
| "step": 28275 | |
| }, | |
| { | |
| "epoch": 0.7075, | |
| "grad_norm": 4.1902666091918945, | |
| "learning_rate": 2.0613995944630017e-05, | |
| "loss": 1.7355245971679687, | |
| "step": 28300 | |
| }, | |
| { | |
| "epoch": 0.708125, | |
| "grad_norm": 1.5900154113769531, | |
| "learning_rate": 2.0532589223025117e-05, | |
| "loss": 1.8640727233886718, | |
| "step": 28325 | |
| }, | |
| { | |
| "epoch": 0.70875, | |
| "grad_norm": 3.142465829849243, | |
| "learning_rate": 2.0451302008154627e-05, | |
| "loss": 1.7659039306640625, | |
| "step": 28350 | |
| }, | |
| { | |
| "epoch": 0.709375, | |
| "grad_norm": 2.7777488231658936, | |
| "learning_rate": 2.037013462968335e-05, | |
| "loss": 1.8307470703125, | |
| "step": 28375 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 2.978544235229492, | |
| "learning_rate": 2.0289087416790164e-05, | |
| "loss": 1.7319793701171875, | |
| "step": 28400 | |
| }, | |
| { | |
| "epoch": 0.710625, | |
| "grad_norm": 2.3036937713623047, | |
| "learning_rate": 2.0208160698166573e-05, | |
| "loss": 1.7220335388183594, | |
| "step": 28425 | |
| }, | |
| { | |
| "epoch": 0.71125, | |
| "grad_norm": 3.40317440032959, | |
| "learning_rate": 2.0127354802015413e-05, | |
| "loss": 1.8039266967773437, | |
| "step": 28450 | |
| }, | |
| { | |
| "epoch": 0.711875, | |
| "grad_norm": 3.2229976654052734, | |
| "learning_rate": 2.0046670056049543e-05, | |
| "loss": 1.7953263854980468, | |
| "step": 28475 | |
| }, | |
| { | |
| "epoch": 0.7125, | |
| "grad_norm": 2.764338254928589, | |
| "learning_rate": 1.9966106787490436e-05, | |
| "loss": 1.8304457092285156, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.713125, | |
| "grad_norm": 2.370849847793579, | |
| "learning_rate": 1.9885665323066948e-05, | |
| "loss": 1.823063507080078, | |
| "step": 28525 | |
| }, | |
| { | |
| "epoch": 0.71375, | |
| "grad_norm": 4.724574089050293, | |
| "learning_rate": 1.9805345989013946e-05, | |
| "loss": 1.8434736633300781, | |
| "step": 28550 | |
| }, | |
| { | |
| "epoch": 0.714375, | |
| "grad_norm": 2.887899160385132, | |
| "learning_rate": 1.972514911107099e-05, | |
| "loss": 1.9003230285644532, | |
| "step": 28575 | |
| }, | |
| { | |
| "epoch": 0.715, | |
| "grad_norm": 3.3361716270446777, | |
| "learning_rate": 1.9645075014481022e-05, | |
| "loss": 1.8826466369628907, | |
| "step": 28600 | |
| }, | |
| { | |
| "epoch": 0.715625, | |
| "grad_norm": 2.7780508995056152, | |
| "learning_rate": 1.956512402398899e-05, | |
| "loss": 1.74138916015625, | |
| "step": 28625 | |
| }, | |
| { | |
| "epoch": 0.71625, | |
| "grad_norm": 4.50689697265625, | |
| "learning_rate": 1.948529646384063e-05, | |
| "loss": 1.7760557556152343, | |
| "step": 28650 | |
| }, | |
| { | |
| "epoch": 0.716875, | |
| "grad_norm": 2.1600148677825928, | |
| "learning_rate": 1.9405592657781085e-05, | |
| "loss": 1.7644316101074218, | |
| "step": 28675 | |
| }, | |
| { | |
| "epoch": 0.7175, | |
| "grad_norm": 2.3051178455352783, | |
| "learning_rate": 1.93260129290536e-05, | |
| "loss": 1.8100906372070313, | |
| "step": 28700 | |
| }, | |
| { | |
| "epoch": 0.718125, | |
| "grad_norm": 3.156888961791992, | |
| "learning_rate": 1.924655760039824e-05, | |
| "loss": 1.8510385131835938, | |
| "step": 28725 | |
| }, | |
| { | |
| "epoch": 0.71875, | |
| "grad_norm": 2.6495463848114014, | |
| "learning_rate": 1.9167226994050504e-05, | |
| "loss": 1.885066375732422, | |
| "step": 28750 | |
| }, | |
| { | |
| "epoch": 0.719375, | |
| "grad_norm": 2.384624719619751, | |
| "learning_rate": 1.908802143174012e-05, | |
| "loss": 1.84552490234375, | |
| "step": 28775 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 2.8937699794769287, | |
| "learning_rate": 1.900894123468969e-05, | |
| "loss": 1.8889472961425782, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 0.720625, | |
| "grad_norm": 2.001762628555298, | |
| "learning_rate": 1.892998672361338e-05, | |
| "loss": 1.8564105224609375, | |
| "step": 28825 | |
| }, | |
| { | |
| "epoch": 0.72125, | |
| "grad_norm": 4.232551574707031, | |
| "learning_rate": 1.8851158218715642e-05, | |
| "loss": 1.8606890869140624, | |
| "step": 28850 | |
| }, | |
| { | |
| "epoch": 0.721875, | |
| "grad_norm": 2.124354600906372, | |
| "learning_rate": 1.8772456039689863e-05, | |
| "loss": 1.9090719604492188, | |
| "step": 28875 | |
| }, | |
| { | |
| "epoch": 0.7225, | |
| "grad_norm": 2.9052963256835938, | |
| "learning_rate": 1.8693880505717148e-05, | |
| "loss": 1.7463037109375, | |
| "step": 28900 | |
| }, | |
| { | |
| "epoch": 0.723125, | |
| "grad_norm": 2.1502320766448975, | |
| "learning_rate": 1.8615431935464982e-05, | |
| "loss": 1.8378799438476563, | |
| "step": 28925 | |
| }, | |
| { | |
| "epoch": 0.72375, | |
| "grad_norm": 2.4353716373443604, | |
| "learning_rate": 1.8537110647085927e-05, | |
| "loss": 1.6792488098144531, | |
| "step": 28950 | |
| }, | |
| { | |
| "epoch": 0.724375, | |
| "grad_norm": 1.8831502199172974, | |
| "learning_rate": 1.8458916958216377e-05, | |
| "loss": 1.787216033935547, | |
| "step": 28975 | |
| }, | |
| { | |
| "epoch": 0.725, | |
| "grad_norm": 3.8031959533691406, | |
| "learning_rate": 1.8380851185975172e-05, | |
| "loss": 1.9620639038085939, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.725, | |
| "eval_loss": 0.449940949678421, | |
| "eval_runtime": 23.729, | |
| "eval_samples_per_second": 61.359, | |
| "eval_steps_per_second": 3.835, | |
| "eval_wer": 0.22426511064626226, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.725625, | |
| "grad_norm": 1.5311033725738525, | |
| "learning_rate": 1.830291364696245e-05, | |
| "loss": 1.8388400268554688, | |
| "step": 29025 | |
| }, | |
| { | |
| "epoch": 0.72625, | |
| "grad_norm": 2.860775947570801, | |
| "learning_rate": 1.822510465725826e-05, | |
| "loss": 1.8175730895996094, | |
| "step": 29050 | |
| }, | |
| { | |
| "epoch": 0.726875, | |
| "grad_norm": 3.2072203159332275, | |
| "learning_rate": 1.814742453242132e-05, | |
| "loss": 1.7896888732910157, | |
| "step": 29075 | |
| }, | |
| { | |
| "epoch": 0.7275, | |
| "grad_norm": 2.612344980239868, | |
| "learning_rate": 1.8069873587487735e-05, | |
| "loss": 1.771622772216797, | |
| "step": 29100 | |
| }, | |
| { | |
| "epoch": 0.728125, | |
| "grad_norm": 4.4283833503723145, | |
| "learning_rate": 1.7992452136969672e-05, | |
| "loss": 1.8493814086914062, | |
| "step": 29125 | |
| }, | |
| { | |
| "epoch": 0.72875, | |
| "grad_norm": 3.086913824081421, | |
| "learning_rate": 1.791516049485419e-05, | |
| "loss": 1.7472283935546875, | |
| "step": 29150 | |
| }, | |
| { | |
| "epoch": 0.729375, | |
| "grad_norm": 2.3418471813201904, | |
| "learning_rate": 1.783799897460185e-05, | |
| "loss": 1.7702195739746094, | |
| "step": 29175 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 1.799540400505066, | |
| "learning_rate": 1.776096788914553e-05, | |
| "loss": 1.6539627075195313, | |
| "step": 29200 | |
| }, | |
| { | |
| "epoch": 0.730625, | |
| "grad_norm": 3.156853437423706, | |
| "learning_rate": 1.768406755088911e-05, | |
| "loss": 1.905420379638672, | |
| "step": 29225 | |
| }, | |
| { | |
| "epoch": 0.73125, | |
| "grad_norm": 2.70528244972229, | |
| "learning_rate": 1.760729827170622e-05, | |
| "loss": 1.8957766723632812, | |
| "step": 29250 | |
| }, | |
| { | |
| "epoch": 0.731875, | |
| "grad_norm": 2.209939479827881, | |
| "learning_rate": 1.753066036293896e-05, | |
| "loss": 1.7784126281738282, | |
| "step": 29275 | |
| }, | |
| { | |
| "epoch": 0.7325, | |
| "grad_norm": 2.672243595123291, | |
| "learning_rate": 1.7454154135396672e-05, | |
| "loss": 1.8713038635253907, | |
| "step": 29300 | |
| }, | |
| { | |
| "epoch": 0.733125, | |
| "grad_norm": 2.45483660697937, | |
| "learning_rate": 1.737777989935465e-05, | |
| "loss": 1.8648719787597656, | |
| "step": 29325 | |
| }, | |
| { | |
| "epoch": 0.73375, | |
| "grad_norm": 2.733436346054077, | |
| "learning_rate": 1.7301537964552865e-05, | |
| "loss": 1.819281005859375, | |
| "step": 29350 | |
| }, | |
| { | |
| "epoch": 0.734375, | |
| "grad_norm": 3.501919984817505, | |
| "learning_rate": 1.722542864019476e-05, | |
| "loss": 1.8372895812988281, | |
| "step": 29375 | |
| }, | |
| { | |
| "epoch": 0.735, | |
| "grad_norm": 2.9574389457702637, | |
| "learning_rate": 1.7149452234945968e-05, | |
| "loss": 1.938941192626953, | |
| "step": 29400 | |
| }, | |
| { | |
| "epoch": 0.735625, | |
| "grad_norm": 2.8631019592285156, | |
| "learning_rate": 1.707360905693304e-05, | |
| "loss": 1.968040771484375, | |
| "step": 29425 | |
| }, | |
| { | |
| "epoch": 0.73625, | |
| "grad_norm": 3.750969886779785, | |
| "learning_rate": 1.699789941374225e-05, | |
| "loss": 1.8375764465332032, | |
| "step": 29450 | |
| }, | |
| { | |
| "epoch": 0.736875, | |
| "grad_norm": 1.6876264810562134, | |
| "learning_rate": 1.6922323612418307e-05, | |
| "loss": 1.776588134765625, | |
| "step": 29475 | |
| }, | |
| { | |
| "epoch": 0.7375, | |
| "grad_norm": 2.239081859588623, | |
| "learning_rate": 1.6846881959463067e-05, | |
| "loss": 1.8600083923339843, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.738125, | |
| "grad_norm": 3.5036659240722656, | |
| "learning_rate": 1.677157476083439e-05, | |
| "loss": 1.7588369750976562, | |
| "step": 29525 | |
| }, | |
| { | |
| "epoch": 0.73875, | |
| "grad_norm": 3.6509110927581787, | |
| "learning_rate": 1.6696402321944847e-05, | |
| "loss": 1.767943115234375, | |
| "step": 29550 | |
| }, | |
| { | |
| "epoch": 0.739375, | |
| "grad_norm": 2.497973680496216, | |
| "learning_rate": 1.6621364947660473e-05, | |
| "loss": 1.7164718627929687, | |
| "step": 29575 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 2.525029182434082, | |
| "learning_rate": 1.6546462942299547e-05, | |
| "loss": 1.776769256591797, | |
| "step": 29600 | |
| }, | |
| { | |
| "epoch": 0.740625, | |
| "grad_norm": 2.0380594730377197, | |
| "learning_rate": 1.647169660963132e-05, | |
| "loss": 1.6442259216308595, | |
| "step": 29625 | |
| }, | |
| { | |
| "epoch": 0.74125, | |
| "grad_norm": 2.036754846572876, | |
| "learning_rate": 1.639706625287486e-05, | |
| "loss": 1.6836285400390625, | |
| "step": 29650 | |
| }, | |
| { | |
| "epoch": 0.741875, | |
| "grad_norm": 2.36887788772583, | |
| "learning_rate": 1.6322572174697752e-05, | |
| "loss": 1.7694798278808594, | |
| "step": 29675 | |
| }, | |
| { | |
| "epoch": 0.7425, | |
| "grad_norm": 3.849665403366089, | |
| "learning_rate": 1.6248214677214903e-05, | |
| "loss": 1.79074462890625, | |
| "step": 29700 | |
| }, | |
| { | |
| "epoch": 0.743125, | |
| "grad_norm": 2.272274971008301, | |
| "learning_rate": 1.6173994061987323e-05, | |
| "loss": 1.7998681640625, | |
| "step": 29725 | |
| }, | |
| { | |
| "epoch": 0.74375, | |
| "grad_norm": 3.529465436935425, | |
| "learning_rate": 1.6099910630020842e-05, | |
| "loss": 1.856277313232422, | |
| "step": 29750 | |
| }, | |
| { | |
| "epoch": 0.744375, | |
| "grad_norm": 1.9301955699920654, | |
| "learning_rate": 1.602596468176498e-05, | |
| "loss": 1.828474578857422, | |
| "step": 29775 | |
| }, | |
| { | |
| "epoch": 0.745, | |
| "grad_norm": 6.633857727050781, | |
| "learning_rate": 1.595215651711168e-05, | |
| "loss": 1.8854193115234374, | |
| "step": 29800 | |
| }, | |
| { | |
| "epoch": 0.745625, | |
| "grad_norm": 2.964223861694336, | |
| "learning_rate": 1.587848643539407e-05, | |
| "loss": 1.9634402465820313, | |
| "step": 29825 | |
| }, | |
| { | |
| "epoch": 0.74625, | |
| "grad_norm": 3.6228153705596924, | |
| "learning_rate": 1.580495473538532e-05, | |
| "loss": 1.9821005249023438, | |
| "step": 29850 | |
| }, | |
| { | |
| "epoch": 0.746875, | |
| "grad_norm": 2.8341407775878906, | |
| "learning_rate": 1.5731561715297322e-05, | |
| "loss": 1.9044583129882813, | |
| "step": 29875 | |
| }, | |
| { | |
| "epoch": 0.7475, | |
| "grad_norm": 1.762088656425476, | |
| "learning_rate": 1.5658307672779593e-05, | |
| "loss": 1.9889335632324219, | |
| "step": 29900 | |
| }, | |
| { | |
| "epoch": 0.748125, | |
| "grad_norm": 2.5662949085235596, | |
| "learning_rate": 1.5585192904918006e-05, | |
| "loss": 1.8690359497070312, | |
| "step": 29925 | |
| }, | |
| { | |
| "epoch": 0.74875, | |
| "grad_norm": 5.102180480957031, | |
| "learning_rate": 1.5512217708233595e-05, | |
| "loss": 1.853144073486328, | |
| "step": 29950 | |
| }, | |
| { | |
| "epoch": 0.749375, | |
| "grad_norm": 2.1142396926879883, | |
| "learning_rate": 1.543938237868137e-05, | |
| "loss": 1.8663597106933594, | |
| "step": 29975 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 2.418893337249756, | |
| "learning_rate": 1.5366687211649062e-05, | |
| "loss": 1.8371963500976562, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "eval_loss": 0.5036048293113708, | |
| "eval_runtime": 22.6329, | |
| "eval_samples_per_second": 64.331, | |
| "eval_steps_per_second": 4.021, | |
| "eval_wer": 0.23967852031267203, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.750625, | |
| "grad_norm": 3.334658622741699, | |
| "learning_rate": 1.5294132501955998e-05, | |
| "loss": 1.7670906066894532, | |
| "step": 30025 | |
| }, | |
| { | |
| "epoch": 0.75125, | |
| "grad_norm": 2.8523125648498535, | |
| "learning_rate": 1.5221718543851881e-05, | |
| "loss": 1.7615585327148438, | |
| "step": 30050 | |
| }, | |
| { | |
| "epoch": 0.751875, | |
| "grad_norm": 3.096256732940674, | |
| "learning_rate": 1.5149445631015557e-05, | |
| "loss": 1.735417022705078, | |
| "step": 30075 | |
| }, | |
| { | |
| "epoch": 0.7525, | |
| "grad_norm": 3.688579559326172, | |
| "learning_rate": 1.5077314056553893e-05, | |
| "loss": 1.7880709838867188, | |
| "step": 30100 | |
| }, | |
| { | |
| "epoch": 0.753125, | |
| "grad_norm": 3.3371503353118896, | |
| "learning_rate": 1.5005324113000501e-05, | |
| "loss": 1.8693490600585938, | |
| "step": 30125 | |
| }, | |
| { | |
| "epoch": 0.75375, | |
| "grad_norm": 2.1416015625, | |
| "learning_rate": 1.4933476092314636e-05, | |
| "loss": 1.984261474609375, | |
| "step": 30150 | |
| }, | |
| { | |
| "epoch": 0.754375, | |
| "grad_norm": 2.997878313064575, | |
| "learning_rate": 1.4861770285879973e-05, | |
| "loss": 1.7618035888671875, | |
| "step": 30175 | |
| }, | |
| { | |
| "epoch": 0.755, | |
| "grad_norm": 1.7867478132247925, | |
| "learning_rate": 1.4790206984503424e-05, | |
| "loss": 1.7961968994140625, | |
| "step": 30200 | |
| }, | |
| { | |
| "epoch": 0.755625, | |
| "grad_norm": 2.6427416801452637, | |
| "learning_rate": 1.4718786478413982e-05, | |
| "loss": 1.6146282958984375, | |
| "step": 30225 | |
| }, | |
| { | |
| "epoch": 0.75625, | |
| "grad_norm": 2.7268593311309814, | |
| "learning_rate": 1.4647509057261477e-05, | |
| "loss": 1.7259321594238282, | |
| "step": 30250 | |
| }, | |
| { | |
| "epoch": 0.756875, | |
| "grad_norm": 2.222780704498291, | |
| "learning_rate": 1.4576375010115494e-05, | |
| "loss": 1.6407740783691407, | |
| "step": 30275 | |
| }, | |
| { | |
| "epoch": 0.7575, | |
| "grad_norm": 1.7338247299194336, | |
| "learning_rate": 1.4505384625464147e-05, | |
| "loss": 1.7334445190429688, | |
| "step": 30300 | |
| }, | |
| { | |
| "epoch": 0.758125, | |
| "grad_norm": 2.2542569637298584, | |
| "learning_rate": 1.4434538191212915e-05, | |
| "loss": 1.7945254516601563, | |
| "step": 30325 | |
| }, | |
| { | |
| "epoch": 0.75875, | |
| "grad_norm": 2.1305809020996094, | |
| "learning_rate": 1.4363835994683494e-05, | |
| "loss": 1.803955078125, | |
| "step": 30350 | |
| }, | |
| { | |
| "epoch": 0.759375, | |
| "grad_norm": 5.294459819793701, | |
| "learning_rate": 1.4293278322612575e-05, | |
| "loss": 1.8314285278320312, | |
| "step": 30375 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 2.845045566558838, | |
| "learning_rate": 1.4222865461150759e-05, | |
| "loss": 1.8424786376953124, | |
| "step": 30400 | |
| }, | |
| { | |
| "epoch": 0.760625, | |
| "grad_norm": 3.2333109378814697, | |
| "learning_rate": 1.415259769586133e-05, | |
| "loss": 1.760749053955078, | |
| "step": 30425 | |
| }, | |
| { | |
| "epoch": 0.76125, | |
| "grad_norm": 2.41461443901062, | |
| "learning_rate": 1.4082475311719195e-05, | |
| "loss": 1.6901165771484374, | |
| "step": 30450 | |
| }, | |
| { | |
| "epoch": 0.761875, | |
| "grad_norm": 3.1380228996276855, | |
| "learning_rate": 1.4012498593109562e-05, | |
| "loss": 1.8321327209472655, | |
| "step": 30475 | |
| }, | |
| { | |
| "epoch": 0.7625, | |
| "grad_norm": 1.8883472681045532, | |
| "learning_rate": 1.394266782382695e-05, | |
| "loss": 1.7628610229492188, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.763125, | |
| "grad_norm": 2.9469122886657715, | |
| "learning_rate": 1.3872983287073943e-05, | |
| "loss": 1.705959014892578, | |
| "step": 30525 | |
| }, | |
| { | |
| "epoch": 0.76375, | |
| "grad_norm": 2.834287166595459, | |
| "learning_rate": 1.3803445265460097e-05, | |
| "loss": 1.6804330444335938, | |
| "step": 30550 | |
| }, | |
| { | |
| "epoch": 0.764375, | |
| "grad_norm": 2.0975701808929443, | |
| "learning_rate": 1.373405404100076e-05, | |
| "loss": 1.8407998657226563, | |
| "step": 30575 | |
| }, | |
| { | |
| "epoch": 0.765, | |
| "grad_norm": 1.7582730054855347, | |
| "learning_rate": 1.3664809895115894e-05, | |
| "loss": 1.7505709838867187, | |
| "step": 30600 | |
| }, | |
| { | |
| "epoch": 0.765625, | |
| "grad_norm": 2.2350873947143555, | |
| "learning_rate": 1.3595713108629033e-05, | |
| "loss": 1.663536834716797, | |
| "step": 30625 | |
| }, | |
| { | |
| "epoch": 0.76625, | |
| "grad_norm": 2.039515495300293, | |
| "learning_rate": 1.3526763961766053e-05, | |
| "loss": 1.7569868469238281, | |
| "step": 30650 | |
| }, | |
| { | |
| "epoch": 0.766875, | |
| "grad_norm": 3.4348556995391846, | |
| "learning_rate": 1.3457962734154094e-05, | |
| "loss": 1.9571942138671874, | |
| "step": 30675 | |
| }, | |
| { | |
| "epoch": 0.7675, | |
| "grad_norm": 1.7568317651748657, | |
| "learning_rate": 1.3389309704820375e-05, | |
| "loss": 1.8475888061523438, | |
| "step": 30700 | |
| }, | |
| { | |
| "epoch": 0.768125, | |
| "grad_norm": 2.7721359729766846, | |
| "learning_rate": 1.3320805152191124e-05, | |
| "loss": 1.715853729248047, | |
| "step": 30725 | |
| }, | |
| { | |
| "epoch": 0.76875, | |
| "grad_norm": 1.5988038778305054, | |
| "learning_rate": 1.3252449354090351e-05, | |
| "loss": 1.7608572387695312, | |
| "step": 30750 | |
| }, | |
| { | |
| "epoch": 0.769375, | |
| "grad_norm": 3.4690773487091064, | |
| "learning_rate": 1.3184242587738843e-05, | |
| "loss": 1.776175994873047, | |
| "step": 30775 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 2.4853203296661377, | |
| "learning_rate": 1.3116185129752966e-05, | |
| "loss": 1.8254159545898438, | |
| "step": 30800 | |
| }, | |
| { | |
| "epoch": 0.770625, | |
| "grad_norm": 2.576169729232788, | |
| "learning_rate": 1.3048277256143538e-05, | |
| "loss": 1.8097720336914063, | |
| "step": 30825 | |
| }, | |
| { | |
| "epoch": 0.77125, | |
| "grad_norm": 1.8006389141082764, | |
| "learning_rate": 1.298051924231477e-05, | |
| "loss": 1.7565538024902343, | |
| "step": 30850 | |
| }, | |
| { | |
| "epoch": 0.771875, | |
| "grad_norm": 3.603454828262329, | |
| "learning_rate": 1.2912911363063046e-05, | |
| "loss": 1.942030029296875, | |
| "step": 30875 | |
| }, | |
| { | |
| "epoch": 0.7725, | |
| "grad_norm": 1.529112696647644, | |
| "learning_rate": 1.2845453892575916e-05, | |
| "loss": 1.8287332153320313, | |
| "step": 30900 | |
| }, | |
| { | |
| "epoch": 0.773125, | |
| "grad_norm": 3.150308847427368, | |
| "learning_rate": 1.2778147104430937e-05, | |
| "loss": 1.89943603515625, | |
| "step": 30925 | |
| }, | |
| { | |
| "epoch": 0.77375, | |
| "grad_norm": 2.4806392192840576, | |
| "learning_rate": 1.2710991271594558e-05, | |
| "loss": 1.726670684814453, | |
| "step": 30950 | |
| }, | |
| { | |
| "epoch": 0.774375, | |
| "grad_norm": 3.3716561794281006, | |
| "learning_rate": 1.2643986666421026e-05, | |
| "loss": 1.7697877502441406, | |
| "step": 30975 | |
| }, | |
| { | |
| "epoch": 0.775, | |
| "grad_norm": 2.3253931999206543, | |
| "learning_rate": 1.2577133560651244e-05, | |
| "loss": 1.833367919921875, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.775, | |
| "eval_loss": 0.5539537072181702, | |
| "eval_runtime": 35.0043, | |
| "eval_samples_per_second": 41.595, | |
| "eval_steps_per_second": 2.6, | |
| "eval_wer": 0.2512752761569232, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.775625, | |
| "grad_norm": 2.8593220710754395, | |
| "learning_rate": 1.251043222541174e-05, | |
| "loss": 1.749234161376953, | |
| "step": 31025 | |
| }, | |
| { | |
| "epoch": 0.77625, | |
| "grad_norm": 2.367875099182129, | |
| "learning_rate": 1.2443882931213513e-05, | |
| "loss": 1.7033932495117188, | |
| "step": 31050 | |
| }, | |
| { | |
| "epoch": 0.776875, | |
| "grad_norm": 3.678410530090332, | |
| "learning_rate": 1.2377485947950957e-05, | |
| "loss": 1.8244671630859375, | |
| "step": 31075 | |
| }, | |
| { | |
| "epoch": 0.7775, | |
| "grad_norm": 2.936824083328247, | |
| "learning_rate": 1.231124154490076e-05, | |
| "loss": 1.9685108947753907, | |
| "step": 31100 | |
| }, | |
| { | |
| "epoch": 0.778125, | |
| "grad_norm": 2.7324280738830566, | |
| "learning_rate": 1.2245149990720789e-05, | |
| "loss": 1.9652584838867186, | |
| "step": 31125 | |
| }, | |
| { | |
| "epoch": 0.77875, | |
| "grad_norm": 11.172932624816895, | |
| "learning_rate": 1.217921155344905e-05, | |
| "loss": 1.8724055480957031, | |
| "step": 31150 | |
| }, | |
| { | |
| "epoch": 0.779375, | |
| "grad_norm": 2.0782206058502197, | |
| "learning_rate": 1.2113426500502573e-05, | |
| "loss": 1.9373057556152344, | |
| "step": 31175 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 1.9793192148208618, | |
| "learning_rate": 1.2047795098676318e-05, | |
| "loss": 1.7272933959960937, | |
| "step": 31200 | |
| }, | |
| { | |
| "epoch": 0.780625, | |
| "grad_norm": 3.1125380992889404, | |
| "learning_rate": 1.198231761414213e-05, | |
| "loss": 1.7556739807128907, | |
| "step": 31225 | |
| }, | |
| { | |
| "epoch": 0.78125, | |
| "grad_norm": 1.9790034294128418, | |
| "learning_rate": 1.1916994312447582e-05, | |
| "loss": 1.8389309692382811, | |
| "step": 31250 | |
| }, | |
| { | |
| "epoch": 0.781875, | |
| "grad_norm": 4.696771621704102, | |
| "learning_rate": 1.1851825458514998e-05, | |
| "loss": 1.7909353637695313, | |
| "step": 31275 | |
| }, | |
| { | |
| "epoch": 0.7825, | |
| "grad_norm": 1.9680570363998413, | |
| "learning_rate": 1.1786811316640312e-05, | |
| "loss": 1.8586251831054688, | |
| "step": 31300 | |
| }, | |
| { | |
| "epoch": 0.783125, | |
| "grad_norm": 2.4509968757629395, | |
| "learning_rate": 1.1721952150492016e-05, | |
| "loss": 1.7676254272460938, | |
| "step": 31325 | |
| }, | |
| { | |
| "epoch": 0.78375, | |
| "grad_norm": 2.500845193862915, | |
| "learning_rate": 1.1657248223110107e-05, | |
| "loss": 1.7392453002929686, | |
| "step": 31350 | |
| }, | |
| { | |
| "epoch": 0.784375, | |
| "grad_norm": 3.8576433658599854, | |
| "learning_rate": 1.159269979690496e-05, | |
| "loss": 1.776149139404297, | |
| "step": 31375 | |
| }, | |
| { | |
| "epoch": 0.785, | |
| "grad_norm": 2.191619873046875, | |
| "learning_rate": 1.1528307133656336e-05, | |
| "loss": 1.7632757568359374, | |
| "step": 31400 | |
| }, | |
| { | |
| "epoch": 0.785625, | |
| "grad_norm": 2.6814188957214355, | |
| "learning_rate": 1.146407049451229e-05, | |
| "loss": 1.7561300659179688, | |
| "step": 31425 | |
| }, | |
| { | |
| "epoch": 0.78625, | |
| "grad_norm": 3.2490785121917725, | |
| "learning_rate": 1.139999013998812e-05, | |
| "loss": 1.8816397094726562, | |
| "step": 31450 | |
| }, | |
| { | |
| "epoch": 0.786875, | |
| "grad_norm": 1.9389029741287231, | |
| "learning_rate": 1.133606632996529e-05, | |
| "loss": 1.722158203125, | |
| "step": 31475 | |
| }, | |
| { | |
| "epoch": 0.7875, | |
| "grad_norm": 2.8934407234191895, | |
| "learning_rate": 1.1272299323690372e-05, | |
| "loss": 1.9442939758300781, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.788125, | |
| "grad_norm": 2.762143135070801, | |
| "learning_rate": 1.120868937977404e-05, | |
| "loss": 1.7703840637207031, | |
| "step": 31525 | |
| }, | |
| { | |
| "epoch": 0.78875, | |
| "grad_norm": 2.2039294242858887, | |
| "learning_rate": 1.1145236756189992e-05, | |
| "loss": 1.7439324951171875, | |
| "step": 31550 | |
| }, | |
| { | |
| "epoch": 0.789375, | |
| "grad_norm": 2.3841187953948975, | |
| "learning_rate": 1.108194171027388e-05, | |
| "loss": 1.7626524353027344, | |
| "step": 31575 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 1.8977760076522827, | |
| "learning_rate": 1.1018804498722335e-05, | |
| "loss": 1.7853363037109375, | |
| "step": 31600 | |
| }, | |
| { | |
| "epoch": 0.790625, | |
| "grad_norm": 2.9132497310638428, | |
| "learning_rate": 1.0955825377591822e-05, | |
| "loss": 1.655137481689453, | |
| "step": 31625 | |
| }, | |
| { | |
| "epoch": 0.79125, | |
| "grad_norm": 2.650176525115967, | |
| "learning_rate": 1.089300460229769e-05, | |
| "loss": 1.8052880859375, | |
| "step": 31650 | |
| }, | |
| { | |
| "epoch": 0.791875, | |
| "grad_norm": 2.930634021759033, | |
| "learning_rate": 1.0830342427613144e-05, | |
| "loss": 1.6885800170898437, | |
| "step": 31675 | |
| }, | |
| { | |
| "epoch": 0.7925, | |
| "grad_norm": 2.0222082138061523, | |
| "learning_rate": 1.0767839107668137e-05, | |
| "loss": 1.6808914184570312, | |
| "step": 31700 | |
| }, | |
| { | |
| "epoch": 0.793125, | |
| "grad_norm": 2.67364239692688, | |
| "learning_rate": 1.070549489594836e-05, | |
| "loss": 1.8964996337890625, | |
| "step": 31725 | |
| }, | |
| { | |
| "epoch": 0.79375, | |
| "grad_norm": 2.072345018386841, | |
| "learning_rate": 1.0643310045294258e-05, | |
| "loss": 1.772504119873047, | |
| "step": 31750 | |
| }, | |
| { | |
| "epoch": 0.794375, | |
| "grad_norm": 4.171526908874512, | |
| "learning_rate": 1.0581284807899989e-05, | |
| "loss": 1.7411439514160156, | |
| "step": 31775 | |
| }, | |
| { | |
| "epoch": 0.795, | |
| "grad_norm": 2.6056482791900635, | |
| "learning_rate": 1.0519419435312366e-05, | |
| "loss": 1.8352873229980469, | |
| "step": 31800 | |
| }, | |
| { | |
| "epoch": 0.795625, | |
| "grad_norm": 4.4820404052734375, | |
| "learning_rate": 1.0457714178429889e-05, | |
| "loss": 1.8728425598144531, | |
| "step": 31825 | |
| }, | |
| { | |
| "epoch": 0.79625, | |
| "grad_norm": 2.4168143272399902, | |
| "learning_rate": 1.039616928750165e-05, | |
| "loss": 1.7988645935058594, | |
| "step": 31850 | |
| }, | |
| { | |
| "epoch": 0.796875, | |
| "grad_norm": 2.274263858795166, | |
| "learning_rate": 1.0334785012126424e-05, | |
| "loss": 1.7650752258300781, | |
| "step": 31875 | |
| }, | |
| { | |
| "epoch": 0.7975, | |
| "grad_norm": 2.7945079803466797, | |
| "learning_rate": 1.027356160125157e-05, | |
| "loss": 1.73777587890625, | |
| "step": 31900 | |
| }, | |
| { | |
| "epoch": 0.798125, | |
| "grad_norm": 4.31943416595459, | |
| "learning_rate": 1.0212499303172063e-05, | |
| "loss": 1.7402381896972656, | |
| "step": 31925 | |
| }, | |
| { | |
| "epoch": 0.79875, | |
| "grad_norm": 2.1098852157592773, | |
| "learning_rate": 1.0151598365529463e-05, | |
| "loss": 1.7993563842773437, | |
| "step": 31950 | |
| }, | |
| { | |
| "epoch": 0.799375, | |
| "grad_norm": 2.7025465965270996, | |
| "learning_rate": 1.0090859035310957e-05, | |
| "loss": 1.7128355407714844, | |
| "step": 31975 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 4.48005485534668, | |
| "learning_rate": 1.003028155884827e-05, | |
| "loss": 1.7984645080566406, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_loss": 0.6490027904510498, | |
| "eval_runtime": 31.0771, | |
| "eval_samples_per_second": 46.851, | |
| "eval_steps_per_second": 2.928, | |
| "eval_wer": 0.2780285515064773, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.800625, | |
| "grad_norm": 3.453477621078491, | |
| "learning_rate": 9.969866181816757e-06, | |
| "loss": 1.7686375427246093, | |
| "step": 32025 | |
| }, | |
| { | |
| "epoch": 0.80125, | |
| "grad_norm": 2.8005478382110596, | |
| "learning_rate": 9.90961314923437e-06, | |
| "loss": 1.8163677978515624, | |
| "step": 32050 | |
| }, | |
| { | |
| "epoch": 0.801875, | |
| "grad_norm": 3.175250768661499, | |
| "learning_rate": 9.849522705460651e-06, | |
| "loss": 1.718264617919922, | |
| "step": 32075 | |
| }, | |
| { | |
| "epoch": 0.8025, | |
| "grad_norm": 1.9626777172088623, | |
| "learning_rate": 9.789595094195774e-06, | |
| "loss": 1.7565240478515625, | |
| "step": 32100 | |
| }, | |
| { | |
| "epoch": 0.803125, | |
| "grad_norm": 2.761385917663574, | |
| "learning_rate": 9.729830558479497e-06, | |
| "loss": 1.8164970397949218, | |
| "step": 32125 | |
| }, | |
| { | |
| "epoch": 0.80375, | |
| "grad_norm": 1.9622324705123901, | |
| "learning_rate": 9.670229340690257e-06, | |
| "loss": 1.83629638671875, | |
| "step": 32150 | |
| }, | |
| { | |
| "epoch": 0.804375, | |
| "grad_norm": 2.9744925498962402, | |
| "learning_rate": 9.610791682544124e-06, | |
| "loss": 1.8486016845703126, | |
| "step": 32175 | |
| }, | |
| { | |
| "epoch": 0.805, | |
| "grad_norm": 2.3984286785125732, | |
| "learning_rate": 9.551517825093853e-06, | |
| "loss": 1.6849533081054688, | |
| "step": 32200 | |
| }, | |
| { | |
| "epoch": 0.805625, | |
| "grad_norm": 3.0938682556152344, | |
| "learning_rate": 9.4924080087279e-06, | |
| "loss": 1.721339111328125, | |
| "step": 32225 | |
| }, | |
| { | |
| "epoch": 0.80625, | |
| "grad_norm": 3.4632070064544678, | |
| "learning_rate": 9.433462473169413e-06, | |
| "loss": 1.9128927612304687, | |
| "step": 32250 | |
| }, | |
| { | |
| "epoch": 0.806875, | |
| "grad_norm": 2.368891954421997, | |
| "learning_rate": 9.374681457475325e-06, | |
| "loss": 1.7942294311523437, | |
| "step": 32275 | |
| }, | |
| { | |
| "epoch": 0.8075, | |
| "grad_norm": 1.5907002687454224, | |
| "learning_rate": 9.31606520003534e-06, | |
| "loss": 1.7313438415527345, | |
| "step": 32300 | |
| }, | |
| { | |
| "epoch": 0.808125, | |
| "grad_norm": 2.9634206295013428, | |
| "learning_rate": 9.257613938570964e-06, | |
| "loss": 1.720403289794922, | |
| "step": 32325 | |
| }, | |
| { | |
| "epoch": 0.80875, | |
| "grad_norm": 3.453320026397705, | |
| "learning_rate": 9.199327910134575e-06, | |
| "loss": 1.8403993225097657, | |
| "step": 32350 | |
| }, | |
| { | |
| "epoch": 0.809375, | |
| "grad_norm": 2.8648300170898438, | |
| "learning_rate": 9.141207351108405e-06, | |
| "loss": 1.81146728515625, | |
| "step": 32375 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 1.9086005687713623, | |
| "learning_rate": 9.083252497203643e-06, | |
| "loss": 1.7399058532714844, | |
| "step": 32400 | |
| }, | |
| { | |
| "epoch": 0.810625, | |
| "grad_norm": 2.4632346630096436, | |
| "learning_rate": 9.025463583459448e-06, | |
| "loss": 1.971685333251953, | |
| "step": 32425 | |
| }, | |
| { | |
| "epoch": 0.81125, | |
| "grad_norm": 2.4881186485290527, | |
| "learning_rate": 8.967840844241993e-06, | |
| "loss": 1.8623391723632812, | |
| "step": 32450 | |
| }, | |
| { | |
| "epoch": 0.811875, | |
| "grad_norm": 2.466813325881958, | |
| "learning_rate": 8.910384513243536e-06, | |
| "loss": 1.7597869873046874, | |
| "step": 32475 | |
| }, | |
| { | |
| "epoch": 0.8125, | |
| "grad_norm": 2.094214916229248, | |
| "learning_rate": 8.853094823481423e-06, | |
| "loss": 1.8171624755859375, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.813125, | |
| "grad_norm": 3.203645944595337, | |
| "learning_rate": 8.795972007297204e-06, | |
| "loss": 1.8229298400878906, | |
| "step": 32525 | |
| }, | |
| { | |
| "epoch": 0.81375, | |
| "grad_norm": 2.904171943664551, | |
| "learning_rate": 8.739016296355667e-06, | |
| "loss": 1.7683810424804687, | |
| "step": 32550 | |
| }, | |
| { | |
| "epoch": 0.814375, | |
| "grad_norm": 2.6263437271118164, | |
| "learning_rate": 8.682227921643882e-06, | |
| "loss": 1.9821664428710937, | |
| "step": 32575 | |
| }, | |
| { | |
| "epoch": 0.815, | |
| "grad_norm": 2.7604169845581055, | |
| "learning_rate": 8.62560711347029e-06, | |
| "loss": 1.8277806091308593, | |
| "step": 32600 | |
| }, | |
| { | |
| "epoch": 0.815625, | |
| "grad_norm": 3.64284348487854, | |
| "learning_rate": 8.569154101463739e-06, | |
| "loss": 1.9213003540039062, | |
| "step": 32625 | |
| }, | |
| { | |
| "epoch": 0.81625, | |
| "grad_norm": 3.081942558288574, | |
| "learning_rate": 8.512869114572575e-06, | |
| "loss": 2.1078387451171876, | |
| "step": 32650 | |
| }, | |
| { | |
| "epoch": 0.816875, | |
| "grad_norm": 2.209322690963745, | |
| "learning_rate": 8.456752381063727e-06, | |
| "loss": 2.023459777832031, | |
| "step": 32675 | |
| }, | |
| { | |
| "epoch": 0.8175, | |
| "grad_norm": 2.176339864730835, | |
| "learning_rate": 8.400804128521744e-06, | |
| "loss": 2.013303070068359, | |
| "step": 32700 | |
| }, | |
| { | |
| "epoch": 0.818125, | |
| "grad_norm": 2.2729368209838867, | |
| "learning_rate": 8.345024583847911e-06, | |
| "loss": 1.8791629028320314, | |
| "step": 32725 | |
| }, | |
| { | |
| "epoch": 0.81875, | |
| "grad_norm": 1.9897018671035767, | |
| "learning_rate": 8.289413973259264e-06, | |
| "loss": 1.9229840087890624, | |
| "step": 32750 | |
| }, | |
| { | |
| "epoch": 0.819375, | |
| "grad_norm": 2.0523717403411865, | |
| "learning_rate": 8.233972522287776e-06, | |
| "loss": 1.869620361328125, | |
| "step": 32775 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 2.4422755241394043, | |
| "learning_rate": 8.178700455779337e-06, | |
| "loss": 1.7394786071777344, | |
| "step": 32800 | |
| }, | |
| { | |
| "epoch": 0.820625, | |
| "grad_norm": 3.256755828857422, | |
| "learning_rate": 8.123597997892918e-06, | |
| "loss": 1.9161381530761719, | |
| "step": 32825 | |
| }, | |
| { | |
| "epoch": 0.82125, | |
| "grad_norm": 2.2085988521575928, | |
| "learning_rate": 8.068665372099631e-06, | |
| "loss": 1.797237091064453, | |
| "step": 32850 | |
| }, | |
| { | |
| "epoch": 0.821875, | |
| "grad_norm": 3.7364280223846436, | |
| "learning_rate": 8.013902801181788e-06, | |
| "loss": 1.69505126953125, | |
| "step": 32875 | |
| }, | |
| { | |
| "epoch": 0.8225, | |
| "grad_norm": 3.602630376815796, | |
| "learning_rate": 7.959310507232086e-06, | |
| "loss": 1.78424560546875, | |
| "step": 32900 | |
| }, | |
| { | |
| "epoch": 0.823125, | |
| "grad_norm": 2.4896175861358643, | |
| "learning_rate": 7.904888711652625e-06, | |
| "loss": 1.7590841674804687, | |
| "step": 32925 | |
| }, | |
| { | |
| "epoch": 0.82375, | |
| "grad_norm": 1.9129937887191772, | |
| "learning_rate": 7.85063763515405e-06, | |
| "loss": 1.6826028442382812, | |
| "step": 32950 | |
| }, | |
| { | |
| "epoch": 0.824375, | |
| "grad_norm": 5.0725908279418945, | |
| "learning_rate": 7.79655749775462e-06, | |
| "loss": 1.7976095581054687, | |
| "step": 32975 | |
| }, | |
| { | |
| "epoch": 0.825, | |
| "grad_norm": 2.914917469024658, | |
| "learning_rate": 7.742648518779378e-06, | |
| "loss": 1.72054931640625, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.825, | |
| "eval_loss": 0.5828079581260681, | |
| "eval_runtime": 29.47, | |
| "eval_samples_per_second": 49.406, | |
| "eval_steps_per_second": 3.088, | |
| "eval_wer": 0.2614774854123087, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.825625, | |
| "grad_norm": 3.8014869689941406, | |
| "learning_rate": 7.688910916859193e-06, | |
| "loss": 1.725146026611328, | |
| "step": 33025 | |
| }, | |
| { | |
| "epoch": 0.82625, | |
| "grad_norm": 2.1874876022338867, | |
| "learning_rate": 7.635344909929931e-06, | |
| "loss": 1.8538673400878907, | |
| "step": 33050 | |
| }, | |
| { | |
| "epoch": 0.826875, | |
| "grad_norm": 8.311996459960938, | |
| "learning_rate": 7.5819507152315395e-06, | |
| "loss": 1.743026885986328, | |
| "step": 33075 | |
| }, | |
| { | |
| "epoch": 0.8275, | |
| "grad_norm": 1.8048982620239258, | |
| "learning_rate": 7.528728549307141e-06, | |
| "loss": 1.7382862854003907, | |
| "step": 33100 | |
| }, | |
| { | |
| "epoch": 0.828125, | |
| "grad_norm": 2.1023449897766113, | |
| "learning_rate": 7.4756786280022195e-06, | |
| "loss": 1.6402699279785156, | |
| "step": 33125 | |
| }, | |
| { | |
| "epoch": 0.82875, | |
| "grad_norm": 1.9426621198654175, | |
| "learning_rate": 7.422801166463705e-06, | |
| "loss": 1.7723580932617187, | |
| "step": 33150 | |
| }, | |
| { | |
| "epoch": 0.829375, | |
| "grad_norm": 2.3035759925842285, | |
| "learning_rate": 7.370096379139102e-06, | |
| "loss": 1.8502493286132813, | |
| "step": 33175 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 3.0953152179718018, | |
| "learning_rate": 7.3175644797756235e-06, | |
| "loss": 1.7976002502441406, | |
| "step": 33200 | |
| }, | |
| { | |
| "epoch": 0.830625, | |
| "grad_norm": 4.002242565155029, | |
| "learning_rate": 7.265205681419346e-06, | |
| "loss": 1.7345849609375, | |
| "step": 33225 | |
| }, | |
| { | |
| "epoch": 0.83125, | |
| "grad_norm": 2.259481430053711, | |
| "learning_rate": 7.213020196414283e-06, | |
| "loss": 1.8480720520019531, | |
| "step": 33250 | |
| }, | |
| { | |
| "epoch": 0.831875, | |
| "grad_norm": 3.710392713546753, | |
| "learning_rate": 7.161008236401601e-06, | |
| "loss": 1.7738151550292969, | |
| "step": 33275 | |
| }, | |
| { | |
| "epoch": 0.8325, | |
| "grad_norm": 2.3810150623321533, | |
| "learning_rate": 7.109170012318717e-06, | |
| "loss": 1.7561918640136718, | |
| "step": 33300 | |
| }, | |
| { | |
| "epoch": 0.833125, | |
| "grad_norm": 4.321732997894287, | |
| "learning_rate": 7.057505734398451e-06, | |
| "loss": 1.664906005859375, | |
| "step": 33325 | |
| }, | |
| { | |
| "epoch": 0.83375, | |
| "grad_norm": 2.6167469024658203, | |
| "learning_rate": 7.0060156121681875e-06, | |
| "loss": 1.827561492919922, | |
| "step": 33350 | |
| }, | |
| { | |
| "epoch": 0.834375, | |
| "grad_norm": 3.129591464996338, | |
| "learning_rate": 6.954699854448976e-06, | |
| "loss": 1.7982583618164063, | |
| "step": 33375 | |
| }, | |
| { | |
| "epoch": 0.835, | |
| "grad_norm": 4.002767086029053, | |
| "learning_rate": 6.903558669354754e-06, | |
| "loss": 1.838609161376953, | |
| "step": 33400 | |
| }, | |
| { | |
| "epoch": 0.835625, | |
| "grad_norm": 3.472646474838257, | |
| "learning_rate": 6.852592264291468e-06, | |
| "loss": 1.7804832458496094, | |
| "step": 33425 | |
| }, | |
| { | |
| "epoch": 0.83625, | |
| "grad_norm": 2.0477006435394287, | |
| "learning_rate": 6.801800845956224e-06, | |
| "loss": 1.7463336181640625, | |
| "step": 33450 | |
| }, | |
| { | |
| "epoch": 0.836875, | |
| "grad_norm": 2.4562184810638428, | |
| "learning_rate": 6.751184620336471e-06, | |
| "loss": 1.6017431640625, | |
| "step": 33475 | |
| }, | |
| { | |
| "epoch": 0.8375, | |
| "grad_norm": 2.299734115600586, | |
| "learning_rate": 6.70074379270913e-06, | |
| "loss": 1.8297604370117186, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.838125, | |
| "grad_norm": 1.7309660911560059, | |
| "learning_rate": 6.650478567639811e-06, | |
| "loss": 1.551192626953125, | |
| "step": 33525 | |
| }, | |
| { | |
| "epoch": 0.83875, | |
| "grad_norm": 2.7305476665496826, | |
| "learning_rate": 6.6003891489819445e-06, | |
| "loss": 1.8145831298828126, | |
| "step": 33550 | |
| }, | |
| { | |
| "epoch": 0.839375, | |
| "grad_norm": 5.103797912597656, | |
| "learning_rate": 6.550475739875983e-06, | |
| "loss": 1.8275979614257813, | |
| "step": 33575 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 3.882408857345581, | |
| "learning_rate": 6.5007385427485464e-06, | |
| "loss": 1.8506681823730469, | |
| "step": 33600 | |
| }, | |
| { | |
| "epoch": 0.840625, | |
| "grad_norm": 3.419426918029785, | |
| "learning_rate": 6.451177759311616e-06, | |
| "loss": 1.8308029174804688, | |
| "step": 33625 | |
| }, | |
| { | |
| "epoch": 0.84125, | |
| "grad_norm": 2.466341018676758, | |
| "learning_rate": 6.401793590561728e-06, | |
| "loss": 1.8315766906738282, | |
| "step": 33650 | |
| }, | |
| { | |
| "epoch": 0.841875, | |
| "grad_norm": 6.760977268218994, | |
| "learning_rate": 6.352586236779145e-06, | |
| "loss": 1.7919219970703124, | |
| "step": 33675 | |
| }, | |
| { | |
| "epoch": 0.8425, | |
| "grad_norm": 2.317819595336914, | |
| "learning_rate": 6.303555897527047e-06, | |
| "loss": 1.7891563415527343, | |
| "step": 33700 | |
| }, | |
| { | |
| "epoch": 0.843125, | |
| "grad_norm": 2.2070393562316895, | |
| "learning_rate": 6.2547027716507276e-06, | |
| "loss": 1.6730448913574218, | |
| "step": 33725 | |
| }, | |
| { | |
| "epoch": 0.84375, | |
| "grad_norm": 2.119097948074341, | |
| "learning_rate": 6.206027057276753e-06, | |
| "loss": 1.8259173583984376, | |
| "step": 33750 | |
| }, | |
| { | |
| "epoch": 0.844375, | |
| "grad_norm": 2.139061689376831, | |
| "learning_rate": 6.1575289518122246e-06, | |
| "loss": 1.9159573364257811, | |
| "step": 33775 | |
| }, | |
| { | |
| "epoch": 0.845, | |
| "grad_norm": 3.0915138721466064, | |
| "learning_rate": 6.109208651943921e-06, | |
| "loss": 1.8343881225585938, | |
| "step": 33800 | |
| }, | |
| { | |
| "epoch": 0.845625, | |
| "grad_norm": 2.0934252738952637, | |
| "learning_rate": 6.061066353637518e-06, | |
| "loss": 1.7832978820800782, | |
| "step": 33825 | |
| }, | |
| { | |
| "epoch": 0.84625, | |
| "grad_norm": 3.439866542816162, | |
| "learning_rate": 6.0131022521368265e-06, | |
| "loss": 1.7515235900878907, | |
| "step": 33850 | |
| }, | |
| { | |
| "epoch": 0.846875, | |
| "grad_norm": 3.6658577919006348, | |
| "learning_rate": 5.965316541962912e-06, | |
| "loss": 1.7941871643066407, | |
| "step": 33875 | |
| }, | |
| { | |
| "epoch": 0.8475, | |
| "grad_norm": 1.667855143547058, | |
| "learning_rate": 5.917709416913414e-06, | |
| "loss": 1.7450450134277344, | |
| "step": 33900 | |
| }, | |
| { | |
| "epoch": 0.848125, | |
| "grad_norm": 2.7896885871887207, | |
| "learning_rate": 5.870281070061695e-06, | |
| "loss": 1.8268212890625, | |
| "step": 33925 | |
| }, | |
| { | |
| "epoch": 0.84875, | |
| "grad_norm": 2.9469149112701416, | |
| "learning_rate": 5.823031693756065e-06, | |
| "loss": 1.774947052001953, | |
| "step": 33950 | |
| }, | |
| { | |
| "epoch": 0.849375, | |
| "grad_norm": 6.280209064483643, | |
| "learning_rate": 5.7759614796190294e-06, | |
| "loss": 1.7362049865722655, | |
| "step": 33975 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 2.2316510677337646, | |
| "learning_rate": 5.7290706185464615e-06, | |
| "loss": 2.169898681640625, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "eval_loss": 0.6066582798957825, | |
| "eval_runtime": 15.7699, | |
| "eval_samples_per_second": 92.328, | |
| "eval_steps_per_second": 5.77, | |
| "eval_wer": 0.27318433703989137, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.850625, | |
| "grad_norm": 2.755112886428833, | |
| "learning_rate": 5.68235930070688e-06, | |
| "loss": 2.3385797119140626, | |
| "step": 34025 | |
| }, | |
| { | |
| "epoch": 0.85125, | |
| "grad_norm": 2.185634136199951, | |
| "learning_rate": 5.635827715540643e-06, | |
| "loss": 2.19286376953125, | |
| "step": 34050 | |
| }, | |
| { | |
| "epoch": 0.851875, | |
| "grad_norm": 2.4951114654541016, | |
| "learning_rate": 5.589476051759213e-06, | |
| "loss": 1.7330796813964844, | |
| "step": 34075 | |
| }, | |
| { | |
| "epoch": 0.8525, | |
| "grad_norm": 2.175602912902832, | |
| "learning_rate": 5.543304497344348e-06, | |
| "loss": 1.6058052062988282, | |
| "step": 34100 | |
| }, | |
| { | |
| "epoch": 0.853125, | |
| "grad_norm": 2.480367660522461, | |
| "learning_rate": 5.497313239547375e-06, | |
| "loss": 1.6626614379882811, | |
| "step": 34125 | |
| }, | |
| { | |
| "epoch": 0.85375, | |
| "grad_norm": 2.0979690551757812, | |
| "learning_rate": 5.45150246488842e-06, | |
| "loss": 1.7422903442382813, | |
| "step": 34150 | |
| }, | |
| { | |
| "epoch": 0.854375, | |
| "grad_norm": 2.525118350982666, | |
| "learning_rate": 5.405872359155633e-06, | |
| "loss": 2.0258154296875, | |
| "step": 34175 | |
| }, | |
| { | |
| "epoch": 0.855, | |
| "grad_norm": 2.7268192768096924, | |
| "learning_rate": 5.360423107404488e-06, | |
| "loss": 1.9753034973144532, | |
| "step": 34200 | |
| }, | |
| { | |
| "epoch": 0.855625, | |
| "grad_norm": 3.076606273651123, | |
| "learning_rate": 5.3151548939569364e-06, | |
| "loss": 1.7373371887207032, | |
| "step": 34225 | |
| }, | |
| { | |
| "epoch": 0.85625, | |
| "grad_norm": 1.924033522605896, | |
| "learning_rate": 5.2700679024007715e-06, | |
| "loss": 1.6625048828125, | |
| "step": 34250 | |
| }, | |
| { | |
| "epoch": 0.856875, | |
| "grad_norm": 2.83964467048645, | |
| "learning_rate": 5.225162315588811e-06, | |
| "loss": 1.6937838745117189, | |
| "step": 34275 | |
| }, | |
| { | |
| "epoch": 0.8575, | |
| "grad_norm": 1.8477532863616943, | |
| "learning_rate": 5.18043831563817e-06, | |
| "loss": 1.8366981506347657, | |
| "step": 34300 | |
| }, | |
| { | |
| "epoch": 0.858125, | |
| "grad_norm": 4.50243616104126, | |
| "learning_rate": 5.135896083929542e-06, | |
| "loss": 1.787650604248047, | |
| "step": 34325 | |
| }, | |
| { | |
| "epoch": 0.85875, | |
| "grad_norm": 2.1316020488739014, | |
| "learning_rate": 5.091535801106445e-06, | |
| "loss": 1.8170756530761718, | |
| "step": 34350 | |
| }, | |
| { | |
| "epoch": 0.859375, | |
| "grad_norm": 4.047319412231445, | |
| "learning_rate": 5.04735764707448e-06, | |
| "loss": 1.7935894775390624, | |
| "step": 34375 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 2.057403564453125, | |
| "learning_rate": 5.003361801000628e-06, | |
| "loss": 1.769434356689453, | |
| "step": 34400 | |
| }, | |
| { | |
| "epoch": 0.860625, | |
| "grad_norm": 4.655451774597168, | |
| "learning_rate": 4.9595484413125005e-06, | |
| "loss": 1.7775167846679687, | |
| "step": 34425 | |
| }, | |
| { | |
| "epoch": 0.86125, | |
| "grad_norm": 3.3887124061584473, | |
| "learning_rate": 4.915917745697646e-06, | |
| "loss": 1.6967140197753907, | |
| "step": 34450 | |
| }, | |
| { | |
| "epoch": 0.861875, | |
| "grad_norm": 3.3376071453094482, | |
| "learning_rate": 4.872469891102788e-06, | |
| "loss": 1.8201528930664062, | |
| "step": 34475 | |
| }, | |
| { | |
| "epoch": 0.8625, | |
| "grad_norm": 2.2252564430236816, | |
| "learning_rate": 4.829205053733121e-06, | |
| "loss": 1.8272137451171875, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.863125, | |
| "grad_norm": 3.0571916103363037, | |
| "learning_rate": 4.7861234090516295e-06, | |
| "loss": 1.83791259765625, | |
| "step": 34525 | |
| }, | |
| { | |
| "epoch": 0.86375, | |
| "grad_norm": 1.7420860528945923, | |
| "learning_rate": 4.743225131778323e-06, | |
| "loss": 1.848790283203125, | |
| "step": 34550 | |
| }, | |
| { | |
| "epoch": 0.864375, | |
| "grad_norm": 3.0190038681030273, | |
| "learning_rate": 4.700510395889584e-06, | |
| "loss": 1.8141970825195313, | |
| "step": 34575 | |
| }, | |
| { | |
| "epoch": 0.865, | |
| "grad_norm": 3.511894941329956, | |
| "learning_rate": 4.657979374617411e-06, | |
| "loss": 1.781380615234375, | |
| "step": 34600 | |
| }, | |
| { | |
| "epoch": 0.865625, | |
| "grad_norm": 3.0194568634033203, | |
| "learning_rate": 4.6156322404487305e-06, | |
| "loss": 1.7678952026367187, | |
| "step": 34625 | |
| }, | |
| { | |
| "epoch": 0.86625, | |
| "grad_norm": 4.2917094230651855, | |
| "learning_rate": 4.5734691651247216e-06, | |
| "loss": 1.7763279724121093, | |
| "step": 34650 | |
| }, | |
| { | |
| "epoch": 0.866875, | |
| "grad_norm": 2.2971198558807373, | |
| "learning_rate": 4.531490319640097e-06, | |
| "loss": 1.8926039123535157, | |
| "step": 34675 | |
| }, | |
| { | |
| "epoch": 0.8675, | |
| "grad_norm": 3.393141031265259, | |
| "learning_rate": 4.489695874242422e-06, | |
| "loss": 1.8522438049316405, | |
| "step": 34700 | |
| }, | |
| { | |
| "epoch": 0.868125, | |
| "grad_norm": 2.4287848472595215, | |
| "learning_rate": 4.448085998431412e-06, | |
| "loss": 1.9589414978027344, | |
| "step": 34725 | |
| }, | |
| { | |
| "epoch": 0.86875, | |
| "grad_norm": 2.3030974864959717, | |
| "learning_rate": 4.4066608609582335e-06, | |
| "loss": 1.8348329162597656, | |
| "step": 34750 | |
| }, | |
| { | |
| "epoch": 0.869375, | |
| "grad_norm": 2.124169111251831, | |
| "learning_rate": 4.365420629824863e-06, | |
| "loss": 1.675376434326172, | |
| "step": 34775 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 2.064378261566162, | |
| "learning_rate": 4.324365472283359e-06, | |
| "loss": 1.8321926879882813, | |
| "step": 34800 | |
| }, | |
| { | |
| "epoch": 0.870625, | |
| "grad_norm": 3.2759928703308105, | |
| "learning_rate": 4.2834955548352155e-06, | |
| "loss": 1.6870762634277343, | |
| "step": 34825 | |
| }, | |
| { | |
| "epoch": 0.87125, | |
| "grad_norm": 2.5168912410736084, | |
| "learning_rate": 4.2428110432306824e-06, | |
| "loss": 1.9046495056152344, | |
| "step": 34850 | |
| }, | |
| { | |
| "epoch": 0.871875, | |
| "grad_norm": 2.5389134883880615, | |
| "learning_rate": 4.202312102468053e-06, | |
| "loss": 1.8369245910644532, | |
| "step": 34875 | |
| }, | |
| { | |
| "epoch": 0.8725, | |
| "grad_norm": 2.956695556640625, | |
| "learning_rate": 4.161998896793062e-06, | |
| "loss": 1.8323869323730468, | |
| "step": 34900 | |
| }, | |
| { | |
| "epoch": 0.873125, | |
| "grad_norm": 3.2679851055145264, | |
| "learning_rate": 4.121871589698173e-06, | |
| "loss": 1.9071043395996095, | |
| "step": 34925 | |
| }, | |
| { | |
| "epoch": 0.87375, | |
| "grad_norm": 1.711238980293274, | |
| "learning_rate": 4.0819303439219194e-06, | |
| "loss": 1.7988421630859375, | |
| "step": 34950 | |
| }, | |
| { | |
| "epoch": 0.874375, | |
| "grad_norm": 3.8262767791748047, | |
| "learning_rate": 4.0421753214482775e-06, | |
| "loss": 1.7684358215332032, | |
| "step": 34975 | |
| }, | |
| { | |
| "epoch": 0.875, | |
| "grad_norm": 2.993664026260376, | |
| "learning_rate": 4.002606683505944e-06, | |
| "loss": 1.7843077087402344, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.875, | |
| "eval_loss": 0.5375370383262634, | |
| "eval_runtime": 15.6366, | |
| "eval_samples_per_second": 93.115, | |
| "eval_steps_per_second": 5.82, | |
| "eval_wer": 0.2539909721457668, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.875625, | |
| "grad_norm": 3.2466351985931396, | |
| "learning_rate": 3.963224590567754e-06, | |
| "loss": 1.9520001220703125, | |
| "step": 35025 | |
| }, | |
| { | |
| "epoch": 0.87625, | |
| "grad_norm": 2.1774566173553467, | |
| "learning_rate": 3.924029202349994e-06, | |
| "loss": 2.2507861328125, | |
| "step": 35050 | |
| }, | |
| { | |
| "epoch": 0.876875, | |
| "grad_norm": 2.8994381427764893, | |
| "learning_rate": 3.885020677811752e-06, | |
| "loss": 2.0046292114257813, | |
| "step": 35075 | |
| }, | |
| { | |
| "epoch": 0.8775, | |
| "grad_norm": 1.9268085956573486, | |
| "learning_rate": 3.846199175154297e-06, | |
| "loss": 1.8727629089355469, | |
| "step": 35100 | |
| }, | |
| { | |
| "epoch": 0.878125, | |
| "grad_norm": 5.094046592712402, | |
| "learning_rate": 3.8075648518203834e-06, | |
| "loss": 1.901218719482422, | |
| "step": 35125 | |
| }, | |
| { | |
| "epoch": 0.87875, | |
| "grad_norm": 2.2553818225860596, | |
| "learning_rate": 3.7691178644936877e-06, | |
| "loss": 1.8261793518066407, | |
| "step": 35150 | |
| }, | |
| { | |
| "epoch": 0.879375, | |
| "grad_norm": 2.6605892181396484, | |
| "learning_rate": 3.730858369098111e-06, | |
| "loss": 1.8622970581054688, | |
| "step": 35175 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 2.277482032775879, | |
| "learning_rate": 3.6927865207971925e-06, | |
| "loss": 1.7623490905761718, | |
| "step": 35200 | |
| }, | |
| { | |
| "epoch": 0.880625, | |
| "grad_norm": 5.970882415771484, | |
| "learning_rate": 3.654902473993438e-06, | |
| "loss": 1.8307070922851563, | |
| "step": 35225 | |
| }, | |
| { | |
| "epoch": 0.88125, | |
| "grad_norm": 2.443519115447998, | |
| "learning_rate": 3.617206382327715e-06, | |
| "loss": 1.8263980102539064, | |
| "step": 35250 | |
| }, | |
| { | |
| "epoch": 0.881875, | |
| "grad_norm": 4.645899295806885, | |
| "learning_rate": 3.5796983986786383e-06, | |
| "loss": 1.752760467529297, | |
| "step": 35275 | |
| }, | |
| { | |
| "epoch": 0.8825, | |
| "grad_norm": 2.621461868286133, | |
| "learning_rate": 3.542378675161928e-06, | |
| "loss": 1.6981916809082032, | |
| "step": 35300 | |
| }, | |
| { | |
| "epoch": 0.883125, | |
| "grad_norm": 4.516400337219238, | |
| "learning_rate": 3.5052473631298378e-06, | |
| "loss": 1.7078933715820312, | |
| "step": 35325 | |
| }, | |
| { | |
| "epoch": 0.88375, | |
| "grad_norm": 2.835362672805786, | |
| "learning_rate": 3.468304613170459e-06, | |
| "loss": 1.7613107299804687, | |
| "step": 35350 | |
| }, | |
| { | |
| "epoch": 0.884375, | |
| "grad_norm": 2.9467527866363525, | |
| "learning_rate": 3.4315505751071843e-06, | |
| "loss": 1.6842971801757813, | |
| "step": 35375 | |
| }, | |
| { | |
| "epoch": 0.885, | |
| "grad_norm": 4.200562953948975, | |
| "learning_rate": 3.3949853979980796e-06, | |
| "loss": 1.7706721496582032, | |
| "step": 35400 | |
| }, | |
| { | |
| "epoch": 0.885625, | |
| "grad_norm": 2.709312915802002, | |
| "learning_rate": 3.358609230135268e-06, | |
| "loss": 1.8560667419433594, | |
| "step": 35425 | |
| }, | |
| { | |
| "epoch": 0.88625, | |
| "grad_norm": 2.4234044551849365, | |
| "learning_rate": 3.322422219044341e-06, | |
| "loss": 1.8872952270507812, | |
| "step": 35450 | |
| }, | |
| { | |
| "epoch": 0.886875, | |
| "grad_norm": 3.2601585388183594, | |
| "learning_rate": 3.286424511483732e-06, | |
| "loss": 1.9887364196777344, | |
| "step": 35475 | |
| }, | |
| { | |
| "epoch": 0.8875, | |
| "grad_norm": 2.9329090118408203, | |
| "learning_rate": 3.250616253444172e-06, | |
| "loss": 1.7786875915527345, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.888125, | |
| "grad_norm": 5.336656093597412, | |
| "learning_rate": 3.2149975901480445e-06, | |
| "loss": 1.5961141967773438, | |
| "step": 35525 | |
| }, | |
| { | |
| "epoch": 0.88875, | |
| "grad_norm": 2.2060787677764893, | |
| "learning_rate": 3.179568666048849e-06, | |
| "loss": 1.656315460205078, | |
| "step": 35550 | |
| }, | |
| { | |
| "epoch": 0.889375, | |
| "grad_norm": 2.2672340869903564, | |
| "learning_rate": 3.144329624830561e-06, | |
| "loss": 1.7324395751953126, | |
| "step": 35575 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 3.183228015899658, | |
| "learning_rate": 3.109280609407095e-06, | |
| "loss": 1.8609237670898438, | |
| "step": 35600 | |
| }, | |
| { | |
| "epoch": 0.890625, | |
| "grad_norm": 3.837341785430908, | |
| "learning_rate": 3.074421761921681e-06, | |
| "loss": 1.8584852600097657, | |
| "step": 35625 | |
| }, | |
| { | |
| "epoch": 0.89125, | |
| "grad_norm": 1.8350588083267212, | |
| "learning_rate": 3.0397532237463334e-06, | |
| "loss": 1.838435821533203, | |
| "step": 35650 | |
| }, | |
| { | |
| "epoch": 0.891875, | |
| "grad_norm": 2.089864730834961, | |
| "learning_rate": 3.00527513548125e-06, | |
| "loss": 1.6750315856933593, | |
| "step": 35675 | |
| }, | |
| { | |
| "epoch": 0.8925, | |
| "grad_norm": 2.768101930618286, | |
| "learning_rate": 2.970987636954248e-06, | |
| "loss": 1.7642198181152344, | |
| "step": 35700 | |
| }, | |
| { | |
| "epoch": 0.893125, | |
| "grad_norm": 4.682522773742676, | |
| "learning_rate": 2.93689086722021e-06, | |
| "loss": 1.8345341491699219, | |
| "step": 35725 | |
| }, | |
| { | |
| "epoch": 0.89375, | |
| "grad_norm": 2.2468557357788086, | |
| "learning_rate": 2.9029849645604733e-06, | |
| "loss": 1.727734375, | |
| "step": 35750 | |
| }, | |
| { | |
| "epoch": 0.894375, | |
| "grad_norm": 2.3150441646575928, | |
| "learning_rate": 2.8692700664823326e-06, | |
| "loss": 1.8260313415527343, | |
| "step": 35775 | |
| }, | |
| { | |
| "epoch": 0.895, | |
| "grad_norm": 1.7598942518234253, | |
| "learning_rate": 2.8357463097184457e-06, | |
| "loss": 1.871545867919922, | |
| "step": 35800 | |
| }, | |
| { | |
| "epoch": 0.895625, | |
| "grad_norm": 2.230487585067749, | |
| "learning_rate": 2.8024138302262912e-06, | |
| "loss": 1.8273837280273437, | |
| "step": 35825 | |
| }, | |
| { | |
| "epoch": 0.89625, | |
| "grad_norm": 1.7393227815628052, | |
| "learning_rate": 2.769272763187608e-06, | |
| "loss": 1.7831106567382813, | |
| "step": 35850 | |
| }, | |
| { | |
| "epoch": 0.896875, | |
| "grad_norm": 6.020104885101318, | |
| "learning_rate": 2.736323243007832e-06, | |
| "loss": 1.7169265747070312, | |
| "step": 35875 | |
| }, | |
| { | |
| "epoch": 0.8975, | |
| "grad_norm": 2.3624324798583984, | |
| "learning_rate": 2.7035654033156e-06, | |
| "loss": 1.8291258239746093, | |
| "step": 35900 | |
| }, | |
| { | |
| "epoch": 0.898125, | |
| "grad_norm": 3.4619598388671875, | |
| "learning_rate": 2.670999376962163e-06, | |
| "loss": 1.8627418518066405, | |
| "step": 35925 | |
| }, | |
| { | |
| "epoch": 0.89875, | |
| "grad_norm": 2.849405288696289, | |
| "learning_rate": 2.6386252960208636e-06, | |
| "loss": 1.8383404541015624, | |
| "step": 35950 | |
| }, | |
| { | |
| "epoch": 0.899375, | |
| "grad_norm": 2.2616469860076904, | |
| "learning_rate": 2.6064432917866088e-06, | |
| "loss": 1.7829039001464844, | |
| "step": 35975 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 3.064157247543335, | |
| "learning_rate": 2.574453494775303e-06, | |
| "loss": 1.8201005554199219, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "eval_loss": 0.554100513458252, | |
| "eval_runtime": 16.1019, | |
| "eval_samples_per_second": 90.424, | |
| "eval_steps_per_second": 5.651, | |
| "eval_wer": 0.25641307937905977, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.900625, | |
| "grad_norm": 2.1680119037628174, | |
| "learning_rate": 2.5426560347233676e-06, | |
| "loss": 1.727500457763672, | |
| "step": 36025 | |
| }, | |
| { | |
| "epoch": 0.90125, | |
| "grad_norm": 1.8727469444274902, | |
| "learning_rate": 2.5110510405871758e-06, | |
| "loss": 1.7958604431152343, | |
| "step": 36050 | |
| }, | |
| { | |
| "epoch": 0.901875, | |
| "grad_norm": 2.405038356781006, | |
| "learning_rate": 2.4796386405425643e-06, | |
| "loss": 1.8284333801269532, | |
| "step": 36075 | |
| }, | |
| { | |
| "epoch": 0.9025, | |
| "grad_norm": 1.9716167449951172, | |
| "learning_rate": 2.4484189619842724e-06, | |
| "loss": 1.804036865234375, | |
| "step": 36100 | |
| }, | |
| { | |
| "epoch": 0.903125, | |
| "grad_norm": 3.0352747440338135, | |
| "learning_rate": 2.4173921315254534e-06, | |
| "loss": 1.8188287353515624, | |
| "step": 36125 | |
| }, | |
| { | |
| "epoch": 0.90375, | |
| "grad_norm": 2.3595378398895264, | |
| "learning_rate": 2.386558274997153e-06, | |
| "loss": 1.6624153137207032, | |
| "step": 36150 | |
| }, | |
| { | |
| "epoch": 0.904375, | |
| "grad_norm": 2.5205132961273193, | |
| "learning_rate": 2.355917517447803e-06, | |
| "loss": 1.5833612060546876, | |
| "step": 36175 | |
| }, | |
| { | |
| "epoch": 0.905, | |
| "grad_norm": 2.584782838821411, | |
| "learning_rate": 2.3254699831427196e-06, | |
| "loss": 1.732833251953125, | |
| "step": 36200 | |
| }, | |
| { | |
| "epoch": 0.905625, | |
| "grad_norm": 3.403421640396118, | |
| "learning_rate": 2.2952157955635876e-06, | |
| "loss": 1.84, | |
| "step": 36225 | |
| }, | |
| { | |
| "epoch": 0.90625, | |
| "grad_norm": 2.878230094909668, | |
| "learning_rate": 2.2651550774079378e-06, | |
| "loss": 1.7251376342773437, | |
| "step": 36250 | |
| }, | |
| { | |
| "epoch": 0.906875, | |
| "grad_norm": 4.145801544189453, | |
| "learning_rate": 2.2352879505887105e-06, | |
| "loss": 1.8131666564941407, | |
| "step": 36275 | |
| }, | |
| { | |
| "epoch": 0.9075, | |
| "grad_norm": 2.8788821697235107, | |
| "learning_rate": 2.2056145362337077e-06, | |
| "loss": 1.6951284790039063, | |
| "step": 36300 | |
| }, | |
| { | |
| "epoch": 0.908125, | |
| "grad_norm": 3.0356996059417725, | |
| "learning_rate": 2.176134954685122e-06, | |
| "loss": 1.80813720703125, | |
| "step": 36325 | |
| }, | |
| { | |
| "epoch": 0.90875, | |
| "grad_norm": 2.2123701572418213, | |
| "learning_rate": 2.146849325499056e-06, | |
| "loss": 1.8175534057617186, | |
| "step": 36350 | |
| }, | |
| { | |
| "epoch": 0.909375, | |
| "grad_norm": 14.553244590759277, | |
| "learning_rate": 2.1177577674450054e-06, | |
| "loss": 1.7521932983398438, | |
| "step": 36375 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 2.0285041332244873, | |
| "learning_rate": 2.0888603985054154e-06, | |
| "loss": 1.7324940490722656, | |
| "step": 36400 | |
| }, | |
| { | |
| "epoch": 0.910625, | |
| "grad_norm": 7.003868579864502, | |
| "learning_rate": 2.06015733587519e-06, | |
| "loss": 1.7898599243164062, | |
| "step": 36425 | |
| }, | |
| { | |
| "epoch": 0.91125, | |
| "grad_norm": 3.5562081336975098, | |
| "learning_rate": 2.0316486959612023e-06, | |
| "loss": 1.7405706787109374, | |
| "step": 36450 | |
| }, | |
| { | |
| "epoch": 0.911875, | |
| "grad_norm": 3.8729641437530518, | |
| "learning_rate": 2.003334594381845e-06, | |
| "loss": 1.8011514282226562, | |
| "step": 36475 | |
| }, | |
| { | |
| "epoch": 0.9125, | |
| "grad_norm": 2.2035903930664062, | |
| "learning_rate": 1.97521514596653e-06, | |
| "loss": 1.7877006530761719, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.913125, | |
| "grad_norm": 2.501753091812134, | |
| "learning_rate": 1.9472904647552524e-06, | |
| "loss": 1.6645137023925782, | |
| "step": 36525 | |
| }, | |
| { | |
| "epoch": 0.91375, | |
| "grad_norm": 2.543362855911255, | |
| "learning_rate": 1.9195606639981376e-06, | |
| "loss": 1.6369126892089845, | |
| "step": 36550 | |
| }, | |
| { | |
| "epoch": 0.914375, | |
| "grad_norm": 2.655812978744507, | |
| "learning_rate": 1.8920258561549442e-06, | |
| "loss": 1.7173043823242187, | |
| "step": 36575 | |
| }, | |
| { | |
| "epoch": 0.915, | |
| "grad_norm": 2.323878049850464, | |
| "learning_rate": 1.8646861528946126e-06, | |
| "loss": 1.6694361877441406, | |
| "step": 36600 | |
| }, | |
| { | |
| "epoch": 0.915625, | |
| "grad_norm": 1.8715022802352905, | |
| "learning_rate": 1.8375416650948495e-06, | |
| "loss": 1.5087576293945313, | |
| "step": 36625 | |
| }, | |
| { | |
| "epoch": 0.91625, | |
| "grad_norm": 1.7889714241027832, | |
| "learning_rate": 1.8105925028416403e-06, | |
| "loss": 1.5942111206054688, | |
| "step": 36650 | |
| }, | |
| { | |
| "epoch": 0.916875, | |
| "grad_norm": 2.7169477939605713, | |
| "learning_rate": 1.783838775428831e-06, | |
| "loss": 1.7970245361328125, | |
| "step": 36675 | |
| }, | |
| { | |
| "epoch": 0.9175, | |
| "grad_norm": 2.767868757247925, | |
| "learning_rate": 1.7572805913576517e-06, | |
| "loss": 1.7720469665527343, | |
| "step": 36700 | |
| }, | |
| { | |
| "epoch": 0.918125, | |
| "grad_norm": 1.6290792226791382, | |
| "learning_rate": 1.730918058336306e-06, | |
| "loss": 1.68819580078125, | |
| "step": 36725 | |
| }, | |
| { | |
| "epoch": 0.91875, | |
| "grad_norm": 1.8957173824310303, | |
| "learning_rate": 1.7047512832795265e-06, | |
| "loss": 1.8046199035644532, | |
| "step": 36750 | |
| }, | |
| { | |
| "epoch": 0.919375, | |
| "grad_norm": 2.383280038833618, | |
| "learning_rate": 1.6787803723081308e-06, | |
| "loss": 1.651231231689453, | |
| "step": 36775 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 2.514039993286133, | |
| "learning_rate": 1.6530054307486053e-06, | |
| "loss": 1.7484793090820312, | |
| "step": 36800 | |
| }, | |
| { | |
| "epoch": 0.920625, | |
| "grad_norm": 2.1577301025390625, | |
| "learning_rate": 1.627426563132667e-06, | |
| "loss": 1.7132496643066406, | |
| "step": 36825 | |
| }, | |
| { | |
| "epoch": 0.92125, | |
| "grad_norm": 2.5882413387298584, | |
| "learning_rate": 1.6020438731968512e-06, | |
| "loss": 1.8347972106933594, | |
| "step": 36850 | |
| }, | |
| { | |
| "epoch": 0.921875, | |
| "grad_norm": 2.74226713180542, | |
| "learning_rate": 1.5768574638820644e-06, | |
| "loss": 1.7782229614257812, | |
| "step": 36875 | |
| }, | |
| { | |
| "epoch": 0.9225, | |
| "grad_norm": 2.338207244873047, | |
| "learning_rate": 1.5518674373332154e-06, | |
| "loss": 1.7182350158691406, | |
| "step": 36900 | |
| }, | |
| { | |
| "epoch": 0.923125, | |
| "grad_norm": 2.256319761276245, | |
| "learning_rate": 1.527073894898745e-06, | |
| "loss": 1.6511669921875, | |
| "step": 36925 | |
| }, | |
| { | |
| "epoch": 0.92375, | |
| "grad_norm": 2.0828137397766113, | |
| "learning_rate": 1.5024769371302649e-06, | |
| "loss": 1.6159465026855468, | |
| "step": 36950 | |
| }, | |
| { | |
| "epoch": 0.924375, | |
| "grad_norm": 2.1062991619110107, | |
| "learning_rate": 1.4780766637821075e-06, | |
| "loss": 1.72166015625, | |
| "step": 36975 | |
| }, | |
| { | |
| "epoch": 0.925, | |
| "grad_norm": 3.453977108001709, | |
| "learning_rate": 1.4538731738109379e-06, | |
| "loss": 1.6862596130371095, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.925, | |
| "eval_loss": 0.536156177520752, | |
| "eval_runtime": 22.6748, | |
| "eval_samples_per_second": 64.212, | |
| "eval_steps_per_second": 4.013, | |
| "eval_wer": 0.2511651803735917, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.925625, | |
| "grad_norm": 1.8163586854934692, | |
| "learning_rate": 1.4298665653753817e-06, | |
| "loss": 1.5453179931640626, | |
| "step": 37025 | |
| }, | |
| { | |
| "epoch": 0.92625, | |
| "grad_norm": 2.8806331157684326, | |
| "learning_rate": 1.4060569358355702e-06, | |
| "loss": 1.6244830322265624, | |
| "step": 37050 | |
| }, | |
| { | |
| "epoch": 0.926875, | |
| "grad_norm": 1.2571202516555786, | |
| "learning_rate": 1.382444381752801e-06, | |
| "loss": 1.5596502685546876, | |
| "step": 37075 | |
| }, | |
| { | |
| "epoch": 0.9275, | |
| "grad_norm": 2.857180118560791, | |
| "learning_rate": 1.359028998889106e-06, | |
| "loss": 1.649947052001953, | |
| "step": 37100 | |
| }, | |
| { | |
| "epoch": 0.928125, | |
| "grad_norm": 1.8312004804611206, | |
| "learning_rate": 1.3358108822068726e-06, | |
| "loss": 1.6752377319335938, | |
| "step": 37125 | |
| }, | |
| { | |
| "epoch": 0.92875, | |
| "grad_norm": 1.8094779253005981, | |
| "learning_rate": 1.312790125868485e-06, | |
| "loss": 1.7051475524902344, | |
| "step": 37150 | |
| }, | |
| { | |
| "epoch": 0.929375, | |
| "grad_norm": 2.4011728763580322, | |
| "learning_rate": 1.2899668232358997e-06, | |
| "loss": 1.6410353088378906, | |
| "step": 37175 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 3.042506456375122, | |
| "learning_rate": 1.2673410668703035e-06, | |
| "loss": 1.60734375, | |
| "step": 37200 | |
| }, | |
| { | |
| "epoch": 0.930625, | |
| "grad_norm": 2.4088544845581055, | |
| "learning_rate": 1.2449129485317236e-06, | |
| "loss": 1.7188037109375, | |
| "step": 37225 | |
| }, | |
| { | |
| "epoch": 0.93125, | |
| "grad_norm": 2.2511439323425293, | |
| "learning_rate": 1.2226825591786396e-06, | |
| "loss": 1.7060801696777343, | |
| "step": 37250 | |
| }, | |
| { | |
| "epoch": 0.931875, | |
| "grad_norm": 1.8890022039413452, | |
| "learning_rate": 1.2006499889676393e-06, | |
| "loss": 1.7287527465820312, | |
| "step": 37275 | |
| }, | |
| { | |
| "epoch": 0.9325, | |
| "grad_norm": 2.0284175872802734, | |
| "learning_rate": 1.1788153272530522e-06, | |
| "loss": 1.6813333129882813, | |
| "step": 37300 | |
| }, | |
| { | |
| "epoch": 0.933125, | |
| "grad_norm": 2.921224355697632, | |
| "learning_rate": 1.1571786625865722e-06, | |
| "loss": 1.7179515075683593, | |
| "step": 37325 | |
| }, | |
| { | |
| "epoch": 0.93375, | |
| "grad_norm": 2.481013774871826, | |
| "learning_rate": 1.1357400827169074e-06, | |
| "loss": 1.6767776489257813, | |
| "step": 37350 | |
| }, | |
| { | |
| "epoch": 0.934375, | |
| "grad_norm": 3.9224562644958496, | |
| "learning_rate": 1.1144996745894032e-06, | |
| "loss": 1.6462677001953125, | |
| "step": 37375 | |
| }, | |
| { | |
| "epoch": 0.935, | |
| "grad_norm": 2.3437349796295166, | |
| "learning_rate": 1.0934575243457367e-06, | |
| "loss": 1.7150559997558594, | |
| "step": 37400 | |
| }, | |
| { | |
| "epoch": 0.935625, | |
| "grad_norm": 2.7706611156463623, | |
| "learning_rate": 1.072613717323523e-06, | |
| "loss": 1.6845948791503906, | |
| "step": 37425 | |
| }, | |
| { | |
| "epoch": 0.93625, | |
| "grad_norm": 1.9120298624038696, | |
| "learning_rate": 1.051968338055992e-06, | |
| "loss": 1.5821540832519532, | |
| "step": 37450 | |
| }, | |
| { | |
| "epoch": 0.936875, | |
| "grad_norm": 1.6960283517837524, | |
| "learning_rate": 1.0315214702716347e-06, | |
| "loss": 1.74364013671875, | |
| "step": 37475 | |
| }, | |
| { | |
| "epoch": 0.9375, | |
| "grad_norm": 2.3166093826293945, | |
| "learning_rate": 1.0112731968938583e-06, | |
| "loss": 1.8236286926269532, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.938125, | |
| "grad_norm": 2.3062477111816406, | |
| "learning_rate": 9.912236000406804e-07, | |
| "loss": 1.7095428466796876, | |
| "step": 37525 | |
| }, | |
| { | |
| "epoch": 0.93875, | |
| "grad_norm": 2.3025269508361816, | |
| "learning_rate": 9.71372761024364e-07, | |
| "loss": 1.589400634765625, | |
| "step": 37550 | |
| }, | |
| { | |
| "epoch": 0.939375, | |
| "grad_norm": 1.6075117588043213, | |
| "learning_rate": 9.51720760351088e-07, | |
| "loss": 1.6802767944335937, | |
| "step": 37575 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 2.460344076156616, | |
| "learning_rate": 9.322676777206551e-07, | |
| "loss": 1.7251095581054687, | |
| "step": 37600 | |
| }, | |
| { | |
| "epoch": 0.940625, | |
| "grad_norm": 2.1842353343963623, | |
| "learning_rate": 9.130135920261184e-07, | |
| "loss": 1.6881637573242188, | |
| "step": 37625 | |
| }, | |
| { | |
| "epoch": 0.94125, | |
| "grad_norm": 2.6601858139038086, | |
| "learning_rate": 8.939585813535101e-07, | |
| "loss": 1.5861434936523438, | |
| "step": 37650 | |
| }, | |
| { | |
| "epoch": 0.941875, | |
| "grad_norm": 2.3520233631134033, | |
| "learning_rate": 8.751027229814857e-07, | |
| "loss": 1.66027099609375, | |
| "step": 37675 | |
| }, | |
| { | |
| "epoch": 0.9425, | |
| "grad_norm": 2.950151205062866, | |
| "learning_rate": 8.564460933810415e-07, | |
| "loss": 1.6581344604492188, | |
| "step": 37700 | |
| }, | |
| { | |
| "epoch": 0.943125, | |
| "grad_norm": 2.2846779823303223, | |
| "learning_rate": 8.37988768215181e-07, | |
| "loss": 1.6575910949707031, | |
| "step": 37725 | |
| }, | |
| { | |
| "epoch": 0.94375, | |
| "grad_norm": 1.790955662727356, | |
| "learning_rate": 8.197308223386102e-07, | |
| "loss": 1.6929818725585937, | |
| "step": 37750 | |
| }, | |
| { | |
| "epoch": 0.944375, | |
| "grad_norm": 2.891195297241211, | |
| "learning_rate": 8.016723297974649e-07, | |
| "loss": 1.7916607666015625, | |
| "step": 37775 | |
| }, | |
| { | |
| "epoch": 0.945, | |
| "grad_norm": 2.523653268814087, | |
| "learning_rate": 7.838133638289724e-07, | |
| "loss": 1.8589387512207032, | |
| "step": 37800 | |
| }, | |
| { | |
| "epoch": 0.945625, | |
| "grad_norm": 1.7026145458221436, | |
| "learning_rate": 7.661539968611631e-07, | |
| "loss": 1.7661341857910156, | |
| "step": 37825 | |
| }, | |
| { | |
| "epoch": 0.94625, | |
| "grad_norm": 4.2935099601745605, | |
| "learning_rate": 7.486943005125869e-07, | |
| "loss": 1.6951966857910157, | |
| "step": 37850 | |
| }, | |
| { | |
| "epoch": 0.946875, | |
| "grad_norm": 2.9389050006866455, | |
| "learning_rate": 7.314343455920191e-07, | |
| "loss": 1.77512939453125, | |
| "step": 37875 | |
| }, | |
| { | |
| "epoch": 0.9475, | |
| "grad_norm": 3.3536126613616943, | |
| "learning_rate": 7.143742020981836e-07, | |
| "loss": 1.6644300842285156, | |
| "step": 37900 | |
| }, | |
| { | |
| "epoch": 0.948125, | |
| "grad_norm": 1.9991897344589233, | |
| "learning_rate": 6.975139392194352e-07, | |
| "loss": 1.638291015625, | |
| "step": 37925 | |
| }, | |
| { | |
| "epoch": 0.94875, | |
| "grad_norm": 2.235459089279175, | |
| "learning_rate": 6.808536253335274e-07, | |
| "loss": 1.7456976318359374, | |
| "step": 37950 | |
| }, | |
| { | |
| "epoch": 0.949375, | |
| "grad_norm": 1.9106206893920898, | |
| "learning_rate": 6.643933280072845e-07, | |
| "loss": 1.6784976196289063, | |
| "step": 37975 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 2.1578049659729004, | |
| "learning_rate": 6.481331139963631e-07, | |
| "loss": 1.7110411071777343, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "eval_loss": 0.5339155197143555, | |
| "eval_runtime": 15.9119, | |
| "eval_samples_per_second": 91.504, | |
| "eval_steps_per_second": 5.719, | |
| "eval_wer": 0.24951374362361922, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.950625, | |
| "grad_norm": 1.551908016204834, | |
| "learning_rate": 6.3207304924498e-07, | |
| "loss": 1.8510499572753907, | |
| "step": 38025 | |
| }, | |
| { | |
| "epoch": 0.95125, | |
| "grad_norm": 3.8508167266845703, | |
| "learning_rate": 6.16213198885629e-07, | |
| "loss": 1.7350880432128906, | |
| "step": 38050 | |
| }, | |
| { | |
| "epoch": 0.951875, | |
| "grad_norm": 1.9229135513305664, | |
| "learning_rate": 6.00553627238809e-07, | |
| "loss": 1.8570075988769532, | |
| "step": 38075 | |
| }, | |
| { | |
| "epoch": 0.9525, | |
| "grad_norm": 1.83743417263031, | |
| "learning_rate": 5.85094397812802e-07, | |
| "loss": 1.7778363037109375, | |
| "step": 38100 | |
| }, | |
| { | |
| "epoch": 0.953125, | |
| "grad_norm": 2.328683376312256, | |
| "learning_rate": 5.698355733033678e-07, | |
| "loss": 1.69666259765625, | |
| "step": 38125 | |
| }, | |
| { | |
| "epoch": 0.95375, | |
| "grad_norm": 2.3831284046173096, | |
| "learning_rate": 5.547772155935327e-07, | |
| "loss": 1.56741943359375, | |
| "step": 38150 | |
| }, | |
| { | |
| "epoch": 0.954375, | |
| "grad_norm": 1.9232983589172363, | |
| "learning_rate": 5.399193857533014e-07, | |
| "loss": 1.678133544921875, | |
| "step": 38175 | |
| }, | |
| { | |
| "epoch": 0.955, | |
| "grad_norm": 2.525306463241577, | |
| "learning_rate": 5.2526214403944e-07, | |
| "loss": 1.7175205993652343, | |
| "step": 38200 | |
| }, | |
| { | |
| "epoch": 0.955625, | |
| "grad_norm": 2.119749069213867, | |
| "learning_rate": 5.108055498952213e-07, | |
| "loss": 1.5824749755859375, | |
| "step": 38225 | |
| }, | |
| { | |
| "epoch": 0.95625, | |
| "grad_norm": 2.245394229888916, | |
| "learning_rate": 4.96549661950152e-07, | |
| "loss": 1.6923651123046874, | |
| "step": 38250 | |
| }, | |
| { | |
| "epoch": 0.956875, | |
| "grad_norm": 2.622114419937134, | |
| "learning_rate": 4.824945380198009e-07, | |
| "loss": 1.6658657836914061, | |
| "step": 38275 | |
| }, | |
| { | |
| "epoch": 0.9575, | |
| "grad_norm": 1.8155590295791626, | |
| "learning_rate": 4.686402351054997e-07, | |
| "loss": 1.6661819458007812, | |
| "step": 38300 | |
| }, | |
| { | |
| "epoch": 0.958125, | |
| "grad_norm": 1.6947946548461914, | |
| "learning_rate": 4.549868093941534e-07, | |
| "loss": 1.7582823181152343, | |
| "step": 38325 | |
| }, | |
| { | |
| "epoch": 0.95875, | |
| "grad_norm": 3.318251609802246, | |
| "learning_rate": 4.415343162580021e-07, | |
| "loss": 1.7042401123046875, | |
| "step": 38350 | |
| }, | |
| { | |
| "epoch": 0.959375, | |
| "grad_norm": 3.134300947189331, | |
| "learning_rate": 4.2828281025437124e-07, | |
| "loss": 1.8077430725097656, | |
| "step": 38375 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 2.304856061935425, | |
| "learning_rate": 4.1523234512549357e-07, | |
| "loss": 1.6531776428222655, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.960625, | |
| "grad_norm": 2.284179210662842, | |
| "learning_rate": 4.0238297379825986e-07, | |
| "loss": 1.7920649719238282, | |
| "step": 38425 | |
| }, | |
| { | |
| "epoch": 0.96125, | |
| "grad_norm": 2.297605514526367, | |
| "learning_rate": 3.897347483840186e-07, | |
| "loss": 1.6981309509277345, | |
| "step": 38450 | |
| }, | |
| { | |
| "epoch": 0.961875, | |
| "grad_norm": 3.2381296157836914, | |
| "learning_rate": 3.7728772017834866e-07, | |
| "loss": 1.7605181884765626, | |
| "step": 38475 | |
| }, | |
| { | |
| "epoch": 0.9625, | |
| "grad_norm": 3.2656912803649902, | |
| "learning_rate": 3.650419396608762e-07, | |
| "loss": 1.748956756591797, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.963125, | |
| "grad_norm": 2.112867593765259, | |
| "learning_rate": 3.529974564950467e-07, | |
| "loss": 1.8367524719238282, | |
| "step": 38525 | |
| }, | |
| { | |
| "epoch": 0.96375, | |
| "grad_norm": 2.0899088382720947, | |
| "learning_rate": 3.4115431952793097e-07, | |
| "loss": 1.818851776123047, | |
| "step": 38550 | |
| }, | |
| { | |
| "epoch": 0.964375, | |
| "grad_norm": 1.844185471534729, | |
| "learning_rate": 3.2951257679004755e-07, | |
| "loss": 1.6990737915039062, | |
| "step": 38575 | |
| }, | |
| { | |
| "epoch": 0.965, | |
| "grad_norm": 2.458848237991333, | |
| "learning_rate": 3.180722754951238e-07, | |
| "loss": 1.8073634338378906, | |
| "step": 38600 | |
| }, | |
| { | |
| "epoch": 0.965625, | |
| "grad_norm": 1.6090002059936523, | |
| "learning_rate": 3.0683346203995154e-07, | |
| "loss": 1.5878190612792968, | |
| "step": 38625 | |
| }, | |
| { | |
| "epoch": 0.96625, | |
| "grad_norm": 2.879241704940796, | |
| "learning_rate": 2.9579618200415995e-07, | |
| "loss": 1.6844888305664063, | |
| "step": 38650 | |
| }, | |
| { | |
| "epoch": 0.966875, | |
| "grad_norm": 1.7200433015823364, | |
| "learning_rate": 2.849604801500538e-07, | |
| "loss": 1.6891416931152343, | |
| "step": 38675 | |
| }, | |
| { | |
| "epoch": 0.9675, | |
| "grad_norm": 1.7613885402679443, | |
| "learning_rate": 2.743264004224422e-07, | |
| "loss": 1.7006707763671876, | |
| "step": 38700 | |
| }, | |
| { | |
| "epoch": 0.968125, | |
| "grad_norm": 2.2189667224884033, | |
| "learning_rate": 2.638939859484213e-07, | |
| "loss": 1.7751622009277344, | |
| "step": 38725 | |
| }, | |
| { | |
| "epoch": 0.96875, | |
| "grad_norm": 4.008902549743652, | |
| "learning_rate": 2.5366327903723064e-07, | |
| "loss": 1.7802212524414063, | |
| "step": 38750 | |
| }, | |
| { | |
| "epoch": 0.969375, | |
| "grad_norm": 4.5463948249816895, | |
| "learning_rate": 2.4363432118008066e-07, | |
| "loss": 1.6750816345214843, | |
| "step": 38775 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 2.473290205001831, | |
| "learning_rate": 2.338071530499697e-07, | |
| "loss": 1.7390989685058593, | |
| "step": 38800 | |
| }, | |
| { | |
| "epoch": 0.970625, | |
| "grad_norm": 2.0372326374053955, | |
| "learning_rate": 2.2418181450152842e-07, | |
| "loss": 1.6478369140625, | |
| "step": 38825 | |
| }, | |
| { | |
| "epoch": 0.97125, | |
| "grad_norm": 3.5534214973449707, | |
| "learning_rate": 2.147583445708534e-07, | |
| "loss": 1.731019287109375, | |
| "step": 38850 | |
| }, | |
| { | |
| "epoch": 0.971875, | |
| "grad_norm": 2.763908863067627, | |
| "learning_rate": 2.0553678147535727e-07, | |
| "loss": 1.7417121887207032, | |
| "step": 38875 | |
| }, | |
| { | |
| "epoch": 0.9725, | |
| "grad_norm": 4.6495490074157715, | |
| "learning_rate": 1.96517162613602e-07, | |
| "loss": 1.741954345703125, | |
| "step": 38900 | |
| }, | |
| { | |
| "epoch": 0.973125, | |
| "grad_norm": 1.9675019979476929, | |
| "learning_rate": 1.876995245651658e-07, | |
| "loss": 1.8249575805664062, | |
| "step": 38925 | |
| }, | |
| { | |
| "epoch": 0.97375, | |
| "grad_norm": 2.9809000492095947, | |
| "learning_rate": 1.7908390309046008e-07, | |
| "loss": 1.8326338195800782, | |
| "step": 38950 | |
| }, | |
| { | |
| "epoch": 0.974375, | |
| "grad_norm": 2.2002809047698975, | |
| "learning_rate": 1.7067033313063474e-07, | |
| "loss": 1.766726531982422, | |
| "step": 38975 | |
| }, | |
| { | |
| "epoch": 0.975, | |
| "grad_norm": 2.0205509662628174, | |
| "learning_rate": 1.6245884880738415e-07, | |
| "loss": 1.7379258728027345, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.975, | |
| "eval_loss": 0.5394518375396729, | |
| "eval_runtime": 26.5289, | |
| "eval_samples_per_second": 54.884, | |
| "eval_steps_per_second": 3.43, | |
| "eval_wer": 0.25054130426804655, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.975625, | |
| "grad_norm": 2.9340131282806396, | |
| "learning_rate": 1.5444948342284161e-07, | |
| "loss": 1.723140411376953, | |
| "step": 39025 | |
| }, | |
| { | |
| "epoch": 0.97625, | |
| "grad_norm": 2.8210349082946777, | |
| "learning_rate": 1.4664226945944604e-07, | |
| "loss": 1.734684600830078, | |
| "step": 39050 | |
| }, | |
| { | |
| "epoch": 0.976875, | |
| "grad_norm": 1.819295883178711, | |
| "learning_rate": 1.3903723857978669e-07, | |
| "loss": 1.676427001953125, | |
| "step": 39075 | |
| }, | |
| { | |
| "epoch": 0.9775, | |
| "grad_norm": 1.742743730545044, | |
| "learning_rate": 1.3163442162648642e-07, | |
| "loss": 1.78322021484375, | |
| "step": 39100 | |
| }, | |
| { | |
| "epoch": 0.978125, | |
| "grad_norm": 2.4667458534240723, | |
| "learning_rate": 1.2443384862209085e-07, | |
| "loss": 1.7316343688964844, | |
| "step": 39125 | |
| }, | |
| { | |
| "epoch": 0.97875, | |
| "grad_norm": 2.3788604736328125, | |
| "learning_rate": 1.1743554876891827e-07, | |
| "loss": 1.6234529113769531, | |
| "step": 39150 | |
| }, | |
| { | |
| "epoch": 0.979375, | |
| "grad_norm": 2.1760969161987305, | |
| "learning_rate": 1.1063955044897656e-07, | |
| "loss": 1.8981874084472656, | |
| "step": 39175 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 2.968036413192749, | |
| "learning_rate": 1.040458812238132e-07, | |
| "loss": 1.7655755615234374, | |
| "step": 39200 | |
| }, | |
| { | |
| "epoch": 0.980625, | |
| "grad_norm": 1.812878131866455, | |
| "learning_rate": 9.76545678344265e-08, | |
| "loss": 1.8474006652832031, | |
| "step": 39225 | |
| }, | |
| { | |
| "epoch": 0.98125, | |
| "grad_norm": 2.5654356479644775, | |
| "learning_rate": 9.146563620113791e-08, | |
| "loss": 1.745243682861328, | |
| "step": 39250 | |
| }, | |
| { | |
| "epoch": 0.981875, | |
| "grad_norm": 2.2209763526916504, | |
| "learning_rate": 8.547911142350873e-08, | |
| "loss": 1.8014009094238281, | |
| "step": 39275 | |
| }, | |
| { | |
| "epoch": 0.9825, | |
| "grad_norm": 2.118145227432251, | |
| "learning_rate": 7.969501778022915e-08, | |
| "loss": 1.6867680358886719, | |
| "step": 39300 | |
| }, | |
| { | |
| "epoch": 0.983125, | |
| "grad_norm": 2.5861423015594482, | |
| "learning_rate": 7.411337872900714e-08, | |
| "loss": 1.8427943420410156, | |
| "step": 39325 | |
| }, | |
| { | |
| "epoch": 0.98375, | |
| "grad_norm": 1.8700131177902222, | |
| "learning_rate": 6.873421690650194e-08, | |
| "loss": 1.7121469116210937, | |
| "step": 39350 | |
| }, | |
| { | |
| "epoch": 0.984375, | |
| "grad_norm": 3.14858341217041, | |
| "learning_rate": 6.355755412820185e-08, | |
| "loss": 1.6600949096679687, | |
| "step": 39375 | |
| }, | |
| { | |
| "epoch": 0.985, | |
| "grad_norm": 3.3519256114959717, | |
| "learning_rate": 5.8583411388352104e-08, | |
| "loss": 1.7041355895996093, | |
| "step": 39400 | |
| }, | |
| { | |
| "epoch": 0.985625, | |
| "grad_norm": 2.1169350147247314, | |
| "learning_rate": 5.381180885986603e-08, | |
| "loss": 1.6906387329101562, | |
| "step": 39425 | |
| }, | |
| { | |
| "epoch": 0.98625, | |
| "grad_norm": 2.5736663341522217, | |
| "learning_rate": 4.924276589424181e-08, | |
| "loss": 1.6989271545410156, | |
| "step": 39450 | |
| }, | |
| { | |
| "epoch": 0.986875, | |
| "grad_norm": 1.5114718675613403, | |
| "learning_rate": 4.487630102150142e-08, | |
| "loss": 1.6499102783203126, | |
| "step": 39475 | |
| }, | |
| { | |
| "epoch": 0.9875, | |
| "grad_norm": 4.95428466796875, | |
| "learning_rate": 4.071243195007401e-08, | |
| "loss": 1.6292936706542969, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.988125, | |
| "grad_norm": 2.8042359352111816, | |
| "learning_rate": 3.6751175566768216e-08, | |
| "loss": 1.7181492614746094, | |
| "step": 39525 | |
| }, | |
| { | |
| "epoch": 0.98875, | |
| "grad_norm": 4.529180526733398, | |
| "learning_rate": 3.299254793668327e-08, | |
| "loss": 1.7275039672851562, | |
| "step": 39550 | |
| }, | |
| { | |
| "epoch": 0.989375, | |
| "grad_norm": 1.9917725324630737, | |
| "learning_rate": 2.9436564303142455e-08, | |
| "loss": 1.7353329467773437, | |
| "step": 39575 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 5.400338649749756, | |
| "learning_rate": 2.6083239087637545e-08, | |
| "loss": 1.6605230712890624, | |
| "step": 39600 | |
| }, | |
| { | |
| "epoch": 0.990625, | |
| "grad_norm": 1.7397327423095703, | |
| "learning_rate": 2.2932585889762213e-08, | |
| "loss": 1.9418272399902343, | |
| "step": 39625 | |
| }, | |
| { | |
| "epoch": 0.99125, | |
| "grad_norm": 3.377164602279663, | |
| "learning_rate": 1.9984617487173174e-08, | |
| "loss": 1.9733157348632813, | |
| "step": 39650 | |
| }, | |
| { | |
| "epoch": 0.991875, | |
| "grad_norm": 1.9735697507858276, | |
| "learning_rate": 1.7239345835518005e-08, | |
| "loss": 1.783287353515625, | |
| "step": 39675 | |
| }, | |
| { | |
| "epoch": 0.9925, | |
| "grad_norm": 2.605458974838257, | |
| "learning_rate": 1.469678206840186e-08, | |
| "loss": 1.6987759399414062, | |
| "step": 39700 | |
| }, | |
| { | |
| "epoch": 0.993125, | |
| "grad_norm": 2.2757301330566406, | |
| "learning_rate": 1.235693649733194e-08, | |
| "loss": 1.7240098571777345, | |
| "step": 39725 | |
| }, | |
| { | |
| "epoch": 0.99375, | |
| "grad_norm": 2.2107725143432617, | |
| "learning_rate": 1.0219818611684196e-08, | |
| "loss": 1.6976016235351563, | |
| "step": 39750 | |
| }, | |
| { | |
| "epoch": 0.994375, | |
| "grad_norm": 2.1432361602783203, | |
| "learning_rate": 8.28543707865892e-09, | |
| "loss": 1.7875912475585938, | |
| "step": 39775 | |
| }, | |
| { | |
| "epoch": 0.995, | |
| "grad_norm": 2.3716068267822266, | |
| "learning_rate": 6.553799743247435e-09, | |
| "loss": 1.8155738830566406, | |
| "step": 39800 | |
| }, | |
| { | |
| "epoch": 0.995625, | |
| "grad_norm": 1.833296775817871, | |
| "learning_rate": 5.024913628209893e-09, | |
| "loss": 1.7683047485351562, | |
| "step": 39825 | |
| }, | |
| { | |
| "epoch": 0.99625, | |
| "grad_norm": 2.624851942062378, | |
| "learning_rate": 3.6987849340197613e-09, | |
| "loss": 2.011045379638672, | |
| "step": 39850 | |
| }, | |
| { | |
| "epoch": 0.996875, | |
| "grad_norm": 2.636291742324829, | |
| "learning_rate": 2.5754190388693757e-09, | |
| "loss": 1.768120574951172, | |
| "step": 39875 | |
| }, | |
| { | |
| "epoch": 0.9975, | |
| "grad_norm": 2.8680660724639893, | |
| "learning_rate": 1.6548204986310823e-09, | |
| "loss": 1.7996044921875, | |
| "step": 39900 | |
| }, | |
| { | |
| "epoch": 0.998125, | |
| "grad_norm": 3.2169315814971924, | |
| "learning_rate": 9.369930468516863e-10, | |
| "loss": 1.7843731689453124, | |
| "step": 39925 | |
| }, | |
| { | |
| "epoch": 0.99875, | |
| "grad_norm": 3.211052417755127, | |
| "learning_rate": 4.219395947080429e-10, | |
| "loss": 1.6871730041503907, | |
| "step": 39950 | |
| }, | |
| { | |
| "epoch": 0.999375, | |
| "grad_norm": 2.170994758605957, | |
| "learning_rate": 1.0966223103481276e-10, | |
| "loss": 1.7490692138671875, | |
| "step": 39975 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 2.818118095397949, | |
| "learning_rate": 1.6222229670681544e-13, | |
| "loss": 1.897050018310547, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.5416486263275146, | |
| "eval_runtime": 24.3051, | |
| "eval_samples_per_second": 59.905, | |
| "eval_steps_per_second": 3.744, | |
| "eval_wer": 0.25090829021248484, | |
| "step": 40000 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 40000, | |
| "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": 6.173402256087083e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |