learning_rate: 1e-05
train_batch_size: 16
eval_batch_size: 16
seed: 42
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
lr_scheduler_warmup_steps: 25
num_epochs: 36
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
8.8732	1.1905	50	3.1684	1.0	1.0
3.0216	2.3810	100	2.7731	1.0	1.0
1.9809	3.5714	150	2.1986	1.0	0.9151
1.314	4.7619	200	1.8100	1.0	0.7256
0.9895	5.9524	250	1.5858	1.0	0.5766
0.7938	7.1429	300	1.5591	1.0	0.5396
0.7051	8.3333	350	1.4336	0.9968	0.4525
0.5897	9.5238	400	1.4223	0.9905	0.4138
0.4997	10.7143	450	1.3381	0.9889	0.3753
0.4195	11.9048	500	1.3785	0.9762	0.3763
0.4047	13.0952	550	1.4765	0.9841	0.4044
0.3613	14.2857	600	1.3639	0.9810	0.3607
0.3281	15.4762	650	1.4331	0.9794	0.3823
0.2896	16.6667	700	1.4105	0.9810	0.3730
0.2903	17.8571	750	1.4395	0.9873	0.3755
0.3066	19.0476	800	1.4222	0.9889	0.3708
0.2572	20.2381	850	1.3705	0.9825	0.3413
0.2427	21.4286	900	1.3780	0.9683	0.3421
0.2359	22.6190	950	1.3623	0.9698	0.3338
0.2554	23.8095	1000	1.3816	0.9762	0.3395
0.2371	25.0	1050	1.4188	0.9778	0.3528
0.1956	26.1905	1100	1.3944	0.9651	0.3433
0.2049	27.3810	1150	1.3707	0.9635	0.3307
0.1992	28.5714	1200	1.4121	0.9714	0.3415
0.2122	29.7619	1250	1.4312	0.9667	0.3477
0.1948	30.9524	1300	1.4533	0.9778	0.3505
0.1828	32.1429	1350	1.4262	0.9698	0.3437
0.1889	33.3333	1400	1.4192	0.9683	0.3439
0.1873	34.5238	1450	1.4188	0.9683	0.3375

tadgeis/xlsr-1b-ckt-chuklang-only

xlsr-1b-ckt-chuklang-only