xls-r-3e-4-300m-mocho-120

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: 0.0184
• Cer: 0.0127

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: 0.0003
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.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: 500
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
3.9611	3.6364	400	1.0612	0.3579
1.0495	7.2727	800	0.4822	0.1765
0.6497	10.9091	1200	0.2971	0.1197
0.4654	14.5455	1600	0.2116	0.0833
0.3305	18.1818	2000	0.1511	0.0655
0.2453	21.8182	2400	0.0776	0.0446
0.1603	25.4545	2800	0.0426	0.0273
0.1116	29.0909	3200	0.0184	0.0127

Framework versions

• Transformers 4.49.0.dev0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0