xlsr-nm-nomi

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

• Loss: 0.3324
• Wer: 0.3245

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: 0.0004
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Use 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: 132
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.9773	6.0606	200	3.0522	1.0
2.8875	12.1212	400	2.3569	0.9959
1.5131	18.1818	600	0.5795	0.6024
0.4675	24.2424	800	0.4022	0.4523
0.2474	30.3030	1000	0.3396	0.4422
0.1573	36.3636	1200	0.3188	0.3611
0.1162	42.4242	1400	0.3450	0.3570
0.0858	48.4848	1600	0.3162	0.3469
0.0675	54.5455	1800	0.2832	0.3327
0.058	60.6061	2000	0.2904	0.3266
0.0415	66.6667	2200	0.3555	0.3306
0.0348	72.7273	2400	0.3116	0.3327
0.0234	78.7879	2600	0.2944	0.3245
0.0215	84.8485	2800	0.3259	0.3266
0.0208	90.9091	3000	0.3312	0.3185
0.0168	96.9697	3200	0.3324	0.3245

Framework versions

• Transformers 4.47.0.dev0
• Pytorch 2.4.0
• Datasets 3.0.1
• Tokenizers 0.20.0