wav2vec-finedtuned-thai

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

• Loss: 2298.1921
• Wer: 0.9944

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: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 20
• training_steps: 200
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	2.0	20	15211.8594	1.2318
17347.0725	4.0	40	12302.1621	1.0916
14383.5075	6.0	60	4903.1660	0.9925
5520.3212	8.0	80	3211.9661	0.9944
3232.9334	10.0	100	2760.6001	0.9944
3232.9334	12.0	120	2523.9800	0.9944
2877.9325	14.0	140	2545.0522	0.9944
2709.8559	16.0	160	2380.6616	0.9944
2578.7244	18.0	180	2229.2996	0.9944
2568.8528	20.0	200	2298.1921	0.9944

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu126
• Datasets 3.3.2
• Tokenizers 0.21.0