learning_rate: 0.0003
train_batch_size: 8
eval_batch_size: 16
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
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: 500
num_epochs: 20
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer	Wer
9.9869	0.4202	200	2.9912	0.9997	1.0
2.9223	0.8403	400	2.2054	0.8978	1.0
1.2808	1.2605	600	0.5408	0.1381	0.6373
0.8764	1.6807	800	0.4139	0.1072	0.5153
0.786	2.1008	1000	0.3735	0.1013	0.4644
0.6716	2.5210	1200	0.3725	0.1016	0.4618
0.6509	2.9412	1400	0.3220	0.0846	0.4069
0.5638	3.3613	1600	0.3009	0.0806	0.3910
0.5591	3.7815	1800	0.2967	0.0806	0.3868
0.4944	4.2017	2000	0.2892	0.0719	0.3428
0.4604	4.6218	2200	0.2852	0.0738	0.3467
0.4848	5.0420	2400	0.2853	0.0735	0.3476
0.3953	5.4622	2600	0.2706	0.0692	0.3299
0.3959	5.8824	2800	0.2809	0.0687	0.3303
0.3432	6.3025	3000	0.3067	0.0698	0.3314
0.3388	6.7227	3200	0.2769	0.0722	0.3333
0.3269	7.1429	3400	0.2881	0.0685	0.3212
0.2849	7.5630	3600	0.2809	0.0691	0.3235
0.2888	7.9832	3800	0.3104	0.0668	0.3187
0.2383	8.4034	4000	0.3056	0.0648	0.3098
0.2493	8.8235	4200	0.3020	0.0658	0.3133
0.2261	9.2437	4400	0.3117	0.0663	0.3088
0.1979	9.6639	4600	0.3058	0.0690	0.3217
0.2018	10.0840	4800	0.3211	0.0649	0.3025
0.1717	10.5042	5000	0.3270	0.0657	0.3095
0.1809	10.9244	5200	0.3282	0.0649	0.3049
0.1654	11.3445	5400	0.3343	0.0655	0.3074
0.1436	11.7647	5600	0.3235	0.0672	0.3090
0.1374	12.1849	5800	0.3651	0.0660	0.3022
0.1311	12.6050	6000	0.3747	0.0650	0.3002
0.1308	13.0252	6200	0.3683	0.0650	0.3041
0.1099	13.4454	6400	0.3608	0.0653	0.3013
0.1134	13.8655	6600	0.3590	0.0651	0.2993
0.0963	14.2857	6800	0.3882	0.0642	0.2986
0.0992	14.7059	7000	0.3805	0.0623	0.2941
0.0931	15.1261	7200	0.4033	0.0638	0.2981
0.0909	15.5462	7400	0.3998	0.0635	0.2918
0.0763	15.9664	7600	0.4060	0.0628	0.2940
0.0746	16.3866	7800	0.4058	0.0639	0.2935
0.0752	16.8067	8000	0.4147	0.0629	0.2882
0.077	17.2269	8200	0.4181	0.0638	0.2882
0.0717	17.6471	8400	0.4201	0.0636	0.2925
0.0612	18.0672	8600	0.4324	0.0626	0.2870
0.066	18.4874	8800	0.4316	0.0618	0.2864
0.0637	18.9076	9000	0.4286	0.0623	0.2876
0.0541	19.3277	9200	0.4401	0.0623	0.2874
0.061	19.7479	9400	0.4378	0.0617	0.2853

Framework versions

Transformers 4.57.1
Pytorch 2.9.1+cu128
Datasets 4.4.1
Tokenizers 0.22.1

sulaimank/wav2vec2-large-xlsr-300m-ttj

wav2vec2-large-xlsr-300m-ttj

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

sulaimank/
wav2vec2-large-xlsr-300m-ttj