wav2vec2-xlsr-53-ft-btb-cv-other-cy

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: inf
• Wer: 0.3727

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: 0.0003
• train_batch_size: 8
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 800
• training_steps: 8000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.9036	0.0643	500	inf	0.9912
1.3889	0.1286	1000	inf	0.7934
1.0207	0.1929	1500	inf	0.6829
0.8901	0.2572	2000	inf	0.6450
0.8467	0.3215	2500	inf	0.5458
0.7813	0.3859	3000	inf	0.5367
0.7456	0.4502	3500	inf	0.5156
0.7312	0.5145	4000	inf	0.4801
0.6929	0.5788	4500	inf	0.4659
0.6567	0.6431	5000	inf	0.4405
0.6413	0.7074	5500	inf	0.4199
0.6187	0.7717	6000	inf	0.4102
0.6086	0.8360	6500	inf	0.4037
0.585	0.9003	7000	inf	0.3872
0.5509	0.9646	7500	inf	0.3832
0.5263	1.0289	8000	inf	0.3727

Framework versions

• Transformers 4.44.0
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1