wav2vec2-large-mms-1b-aft-mmc

This model is a fine-tuned version of facebook/mms-1b-all on the audiofolder dataset. It achieves the following results on the evaluation set:

• Loss: 1.3464
• Wer: 0.6981

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: 0.001
• train_batch_size: 2
• 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: 100
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
13.1415	0.5263	100	9.0823	1.2933
3.968	1.0526	200	3.8510	0.9984
3.4887	1.5789	300	3.2492	0.9984
1.9099	2.1053	400	1.8789	0.7918
1.5364	2.6316	500	1.5877	0.7513
1.3236	3.1579	600	1.5096	0.7653
1.2527	3.6842	700	1.3780	0.7131
1.1843	4.2105	800	1.3713	0.7081
1.1542	4.7368	900	1.3464	0.6981

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.1.1
• Tokenizers 0.22.1