wav2vec2-large-xlsr-53_toy_train_data_masked_audio_10ms

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5945
• Wer: 0.4929

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
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.4049	1.05	250	3.3497	1.0
3.0851	2.1	500	3.4440	1.0
2.3512	3.15	750	1.5938	0.9317
1.1762	4.2	1000	0.8481	0.7333
0.903	5.25	1250	0.7180	0.6484
0.6754	6.3	1500	0.6603	0.6044
0.5961	7.35	1750	0.6410	0.5778
0.5325	8.4	2000	0.6245	0.5545
0.4685	9.45	2250	0.5925	0.5359
0.4526	10.5	2500	0.5991	0.5345
0.3975	11.55	2750	0.5916	0.5228
0.3672	12.6	3000	0.5882	0.5037
0.3774	13.65	3250	0.5693	0.5028
0.3489	14.7	3500	0.5645	0.5018
0.3593	15.75	3750	0.5977	0.5043
0.3167	16.81	4000	0.6049	0.5018
0.3225	17.86	4250	0.6172	0.4921
0.2807	18.91	4500	0.5937	0.4923
0.2889	19.96	4750	0.5945	0.4929

Framework versions

• Transformers 4.17.0
• Pytorch 1.11.0+cu102
• Datasets 2.0.0
• Tokenizers 0.11.6