wav2vec2-base_toy_train_data_random_noise_0.1

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

• Loss: 0.9263
• Wer: 0.7213

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: 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: 1000
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.1296	2.1	250	3.5088	1.0
3.0728	4.2	500	3.1694	1.0
1.8686	6.3	750	1.3414	0.9321
1.1241	8.4	1000	1.0196	0.8321
0.8704	10.5	1250	0.9387	0.7962
0.6734	12.6	1500	0.9309	0.7640
0.5832	14.7	1750	0.9329	0.7346
0.5207	16.8	2000	0.9060	0.7247
0.4857	18.9	2250	0.9263	0.7213

Framework versions

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