wav2vec2-large-xlsr-53-third-train

This model is a fine-tuned version of fathyy/wav2vec2-large-xlsr-53-second-train on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.9881
• Wer: 0.9367

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: 0.001
• train_batch_size: 6
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 12
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1200
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
2.3482	2.9940	500	2.2909	0.9669
2.3547	5.9880	1000	2.4888	0.9759
2.3457	8.9820	1500	2.3550	0.9880
2.1551	11.9760	2000	2.2335	0.9458
1.9369	14.9701	2500	2.1364	0.9458
1.6783	17.9641	3000	1.9881	0.9367

Framework versions

• Transformers 4.45.2
• Pytorch 2.4.0
• Datasets 3.0.1
• Tokenizers 0.20.0