wav2vec2-burak-new-300-v2-8

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.2841
• Wer: 0.2120

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: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 151
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
6.0739	9.43	500	3.1506	1.0
1.6652	18.87	1000	0.3396	0.4136
0.4505	28.3	1500	0.2632	0.3138
0.3115	37.74	2000	0.2536	0.2849
0.2421	47.17	2500	0.2674	0.2588
0.203	56.6	3000	0.2552	0.2471
0.181	66.04	3500	0.2636	0.2595
0.1581	75.47	4000	0.2527	0.2416
0.1453	84.91	4500	0.2773	0.2257
0.1305	94.34	5000	0.2825	0.2257
0.1244	103.77	5500	0.2754	0.2312
0.1127	113.21	6000	0.2772	0.2223
0.1094	122.64	6500	0.2720	0.2223
0.1033	132.08	7000	0.2863	0.2202
0.099	141.51	7500	0.2853	0.2140
0.0972	150.94	8000	0.2841	0.2120

Framework versions

• Transformers 4.25.1
• Pytorch 1.12.1+cu113
• Datasets 2.7.1
• Tokenizers 0.13.2