Fine_Tuned_XLSR_English

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the timit_asr dataset. It achieves the following results on the evaluation set:

• Loss: 0.4033
• Wer: 0.3163

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.3757	1.0	500	3.1570	1.0
2.4891	2.01	1000	0.9252	0.8430
0.8725	3.01	1500	0.4581	0.4931
0.544	4.02	2000	0.3757	0.4328
0.4043	5.02	2500	0.3621	0.4087
0.3376	6.02	3000	0.3682	0.3931
0.2937	7.03	3500	0.3541	0.3743
0.2573	8.03	4000	0.3565	0.3593
0.2257	9.04	4500	0.3634	0.3654
0.215	10.04	5000	0.3695	0.3537
0.1879	11.04	5500	0.3690	0.3486
0.1599	12.05	6000	0.3743	0.3490
0.1499	13.05	6500	0.4108	0.3424
0.147	14.06	7000	0.4048	0.3400
0.1355	15.06	7500	0.3988	0.3357
0.1278	16.06	8000	0.3672	0.3384
0.1189	17.07	8500	0.4011	0.3340
0.1089	18.07	9000	0.3948	0.3300
0.1039	19.08	9500	0.4062	0.3317
0.0971	20.08	10000	0.4041	0.3252
0.0902	21.08	10500	0.4112	0.3301
0.0883	22.09	11000	0.4154	0.3292
0.0864	23.09	11500	0.3746	0.3189
0.0746	24.1	12000	0.3991	0.3230
0.0711	25.1	12500	0.3916	0.3200
0.0712	26.1	13000	0.4024	0.3193
0.0663	27.11	13500	0.3976	0.3184
0.0626	28.11	14000	0.4046	0.3168
0.0641	29.12	14500	0.4033	0.3163

Framework versions

• Transformers 4.17.0
• Pytorch 1.12.1+cu113
• Datasets 1.18.3
• Tokenizers 0.12.1