wav2vec2-large-xls-r-300m-luo-googlefluers-1hr-v1

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

• Loss: 1.0253
• Wer: 0.5023
• Cer: 0.1370

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: 0.0005
• 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: 500
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
9.2323	13.3333	100	3.2827	1.0	1.0
2.9657	26.6667	200	2.8565	1.0	1.0
1.8583	40.0	300	0.7909	0.6233	0.1719
0.2287	53.3333	400	0.9148	0.5632	0.1543
0.1116	66.6667	500	0.9245	0.571	0.1542
0.07	80.0	600	1.0463	0.5342	0.1447
0.0414	93.3333	700	1.0253	0.5023	0.1370

Framework versions

• Transformers 4.42.3
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1