xlsr_hindi_LMless_300m

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice_16_1 dataset. It achieves the following results on the evaluation set:

• Loss: 0.9207
• Wer : 0.4042

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: 10
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 3
• total_train_batch_size: 30
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
11.7707	3.6145	200	3.6870	1.0
3.047	7.2289	400	1.6327	0.8998
0.8469	10.8434	600	0.9648	0.5696
0.3949	14.4578	800	0.8641	0.4718
0.2465	18.0723	1000	0.8983	0.4442
0.1692	21.6867	1200	0.9318	0.4260
0.1307	25.3012	1400	0.9252	0.4097
0.1041	28.9157	1600	0.9207	0.4042

Framework versions

• Transformers 4.40.1
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1