w2v-bert-2.0-marathi-practice-CV16.0

This model is a fine-tuned version of facebook/w2v-bert-2.0 on the common_voice_16_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.6493
• Wer: 0.8590

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: 5e-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: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0394	5.7554	400	0.4974	0.8968
0.0492	11.5108	800	0.5152	0.8860
0.0134	17.2662	1200	0.5789	0.8739
0.0018	23.0216	1600	0.6334	0.8613
0.0002	28.7770	2000	0.6493	0.8590

Framework versions

• Transformers 4.41.0
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1