wav2vec2-ASR-google-colab

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: 2.8244
• Wer: 0.9468

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 1000
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
8.2357	2.9940	500	3.2836	0.9993
2.8848	5.9880	1000	2.3607	1.0030
1.8264	8.9820	1500	2.2085	0.9963
1.2993	11.9760	2000	2.1122	0.9700
0.9819	14.9701	2500	2.4227	0.9700
0.8020	17.9641	3000	2.4763	0.9446
0.6464	20.9581	3500	2.6294	0.9618
0.5507	23.9521	4000	2.7564	0.9483
0.4691	26.9461	4500	2.8887	0.9461
0.4202	29.9401	5000	2.8239	0.9468
0.4202	30.0	5010	2.8244	0.9468

Framework versions

• Transformers 5.12.1
• Pytorch 2.11.0+cu128
• Datasets 5.0.0
• Tokenizers 0.22.2