wav2vec2-base-timit-fine-tuned

This model is a fine-tuned version of facebook/wav2vec2-base on the TIMIT_ASR - NA dataset. It achieves the following results on the evaluation set:

• Loss: 0.4233
• Wer: 0.4329

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: 64
• eval_batch_size: 32
• 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: 20.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.158	1.7241	100	3.6803	1.0
2.9744	3.4483	200	3.1165	1.0
2.9266	5.1724	300	3.0175	1.0
2.1336	6.8966	400	2.2135	1.0117
1.0119	8.6207	500	1.0227	0.8251
0.4995	10.3448	600	0.7700	0.6574
0.3233	12.0690	700	0.4970	0.5241
0.2452	13.7931	800	0.4585	0.4908
0.181	15.5172	900	0.4626	0.4814
0.1419	17.2414	1000	0.4917	0.4775
0.1175	18.9655	1100	0.4279	0.4359

Framework versions

• Transformers 4.42.0.dev0
• Pytorch 2.3.0a0+gitcd033a1
• Datasets 2.19.1
• Tokenizers 0.19.1