learning_rate: 3e-05
train_batch_size: 16
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
num_epochs: 30
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Per Hf Metric	Per Avg
0.4745	1.0	102	0.4702	0.1373	0.1659
0.4131	2.0	204	0.4603	0.1239	0.1532
0.3714	3.0	306	0.3487	0.1317	0.1491
0.3915	4.0	408	0.4590	0.1170	0.1528
0.3434	5.0	510	0.5181	0.1248	0.1531
0.334	6.0	612	0.4217	0.1199	0.1496
0.3233	7.0	714	0.4525	0.1229	0.1499
0.309	8.0	816	0.3458	0.1160	0.1368
0.2979	9.0	918	0.4236	0.1221	0.1521
0.2817	10.0	1020	0.4991	0.1136	0.1406
0.2952	11.0	1122	0.4337	0.1136	0.1407
0.2672	12.0	1224	0.4654	0.1122	0.1451
0.2729	13.0	1326	0.3516	0.1133	0.1341
0.2681	14.0	1428	0.3788	0.1171	0.1462
0.2894	15.0	1530	0.4290	0.1086	0.1361
0.2435	16.0	1632	0.4427	0.1056	0.1369
0.2429	17.0	1734	0.4372	0.1154	0.1413
0.2466	18.0	1836	0.4003	0.1166	0.1421

alinerodrigues/wav2vec2-large-xlsr-facebook-300m-aug-words-phoneme-exp-1-v6

wav2vec2-large-xlsr-facebook-300m-aug-words-phoneme-exp-1-v6