Wav2Vec2_MediumDataset

This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5124
• Wer: 0.3416

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.4758	1.54	500	1.6004	0.9884
0.81	3.09	1000	0.5195	0.5002
0.3974	4.63	1500	0.4514	0.4470
0.2548	6.17	2000	0.4596	0.4104
0.1895	7.72	2500	0.4736	0.4019
0.1547	9.26	3000	0.4821	0.3944
0.1253	10.8	3500	0.5069	0.3735
0.1018	12.35	4000	0.4846	0.3793
0.0955	13.89	4500	0.5465	0.3667
0.0818	15.43	5000	0.4956	0.3768
0.0761	16.98	5500	0.4702	0.3754
0.062	18.52	6000	0.4915	0.3563
0.0567	20.06	6500	0.8820	0.5120
0.0502	21.6	7000	0.5224	0.3735
0.0449	23.15	7500	0.5546	0.3561
0.0332	24.69	8000	0.5240	0.3455
0.0306	26.23	8500	0.5289	0.3429
0.0292	27.78	9000	0.5211	0.3406
0.0275	29.32	9500	0.5124	0.3416

Framework versions

• Transformers 4.17.0
• Pytorch 2.5.1+cu121
• Datasets 1.18.3
• Tokenizers 0.20.3