wav2vec2-large-xlsr-53-Hindi-Version2

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the fleurs dataset. It achieves the following results on the evaluation set:

• Loss: 0.5247
• Wer: 0.3987

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: 3e-05
• 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
3.5791	6.7568	500	3.5172	1.0
3.3826	13.5135	1000	3.3714	1.0
3.338	20.2703	1500	3.3281	1.0
2.3266	27.0270	2000	1.9820	0.9158
1.2023	33.7838	2500	0.9234	0.6469
0.902	40.5405	3000	0.7439	0.5546
0.7956	47.2973	3500	0.6575	0.4950
0.7554	54.0541	4000	0.6092	0.4620
0.6757	60.8108	4500	0.5787	0.4428
0.6819	67.5676	5000	0.5561	0.4259
0.5737	74.3243	5500	0.5473	0.4176
0.5747	81.0811	6000	0.5285	0.4043
0.5419	87.8378	6500	0.5263	0.4011
0.4713	94.5946	7000	0.5247	0.3987

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.4.1+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1