xlsr-nm-nomo

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: 1.0423
• Wer: 0.3916

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
4.7334	6.4590	200	3.1082	1.0
2.9544	12.9180	400	2.8429	0.9956
2.0771	19.3607	600	1.2204	0.8341
0.7271	25.8197	800	1.0868	0.5531
0.3103	32.2623	1000	1.0536	0.4912
0.1852	38.7213	1200	0.9030	0.4469
0.1399	45.1639	1400	0.8980	0.4491
0.0864	51.6230	1600	0.8315	0.4292
0.0643	58.0656	1800	0.9488	0.4004
0.0525	64.5246	2000	0.9354	0.4137
0.0455	70.9836	2200	0.9717	0.4093
0.0383	77.4262	2400	0.9781	0.4004
0.0261	83.8852	2600	1.1244	0.3938
0.0265	90.3279	2800	1.0439	0.4004
0.0197	96.7869	3000	1.0423	0.3916

Framework versions

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