results

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice_17_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.7289
• Wer: 0.5866

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 200
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.4919	2.9851	100	1.2236	0.9106
1.0086	5.9701	200	0.9436	0.8291
0.6072	8.9552	300	0.8277	0.7117
0.55	11.9403	400	0.7774	0.6726
0.3398	14.9254	500	0.7344	0.6212
0.2543	17.9104	600	0.7368	0.6212
0.3558	20.8955	700	0.7313	0.5788
0.1751	23.8806	800	0.7060	0.5855
0.1502	26.8657	900	0.7289	0.5866

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1