wav2vec2-xls-r-300m-cv7-istech

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice_7_0 dataset. It achieves the following results on the evaluation set:

• Loss: 4.3463
• Wer: 1.0023

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: 0.0002
• 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: 20
• num_epochs: 0.5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
15.5865	0.0445	20	18.4600	1.0
17.7134	0.0890	40	15.6651	1.0
15.0397	0.1335	60	13.0063	1.0
10.5751	0.1780	80	10.4916	1.0
14.7612	0.2225	100	8.3979	1.0001
6.4409	0.2670	120	6.8707	1.0004
9.0843	0.3115	140	5.7886	1.0010
8.1412	0.3560	160	5.1050	1.0007
5.6313	0.4004	180	4.6788	1.0024
9.6107	0.4449	200	4.4458	1.0025
4.4122	0.4894	220	4.3463	1.0023

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1