wav2vec2-base_toy_train_data_masked_audio

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: 1.1950
• Wer: 0.7340

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
3.1287	2.1	250	3.4581	1.0
3.0259	4.2	500	2.8099	0.9999
1.4881	6.3	750	1.2929	0.8950
0.9665	8.4	1000	1.1675	0.8346
0.7614	10.5	1250	1.1388	0.8003
0.5858	12.6	1500	1.1510	0.7672
0.5005	14.7	1750	1.1606	0.7532
0.4486	16.8	2000	1.1571	0.7427
0.4224	18.9	2250	1.1950	0.7340

Framework versions

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