wav2vec2-large-xls-r-300m-en-libri-more-steps

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

• Loss: 1.7624
• Wer: 0.8772
• Cer: 0.3762

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: 0.001
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
No log	1.94	33	2.9987	1.0	1.0
No log	3.88	66	2.8951	1.0	1.0
No log	5.82	99	2.8732	1.0	1.0
3.781	7.76	132	2.6057	1.0	1.0
3.781	9.71	165	1.9015	1.0154	0.5616
3.781	11.65	198	1.5226	0.9263	0.4462
2.2258	13.59	231	1.5116	0.8913	0.3967
2.2258	15.53	264	1.5634	0.8922	0.3842
2.2258	17.47	297	1.7016	0.8876	0.3796
0.7946	19.41	330	1.7624	0.8772	0.3762

Framework versions

• Transformers 4.20.1
• Pytorch 1.11.0+cpu
• Datasets 1.18.3
• Tokenizers 0.12.1