masapasa/xls-r-300m-it-cv8-ds13

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the MOZILLA-FOUNDATION/COMMON_VOICE_8_0 - SV-SE dataset. It achieves the following results on the evaluation set:

• Loss: 0.3549
• Wer: 0.3827

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: 7.5e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2000
• num_epochs: 50.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.4129	5.49	500	3.3224	1.0
2.9323	10.98	1000	2.9128	1.0000
1.6839	16.48	1500	0.7740	0.6854
1.485	21.97	2000	0.5830	0.5976
1.362	27.47	2500	0.4866	0.4905
1.2752	32.96	3000	0.4240	0.4967
1.1957	38.46	3500	0.3899	0.4258
1.1646	43.95	4000	0.3597	0.4014
1.1265	49.45	4500	0.3559	0.3829

Framework versions

• Transformers 4.17.0.dev0
• Pytorch 1.10.2+cu102
• Datasets 1.18.3
• Tokenizers 0.11.0