wav2vec2-large-xls-r-300m-kyrgyz

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the MOZILLA-FOUNDATION/COMMON_VOICE_7_0 - KY dataset. It achieves the following results on the evaluation set:

• Loss: 0.5817
• Wer: 0.4096

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: 32
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2000
• num_epochs: 100.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.5412	18.69	2000	0.6161	0.5747
1.311	37.38	4000	0.5707	0.5070
1.1367	56.07	6000	0.5372	0.4664
0.9696	74.77	8000	0.5443	0.4328
0.8163	93.46	10000	0.5916	0.4124

Framework versions

• Transformers 4.16.0.dev0
• Pytorch 1.10.1+cu102
• Datasets 1.17.1.dev0
• Tokenizers 0.11.0