w2v-bert-2.0-nepali

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

• Loss: 0.1758
• Wer: 0.2372

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: 6
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 300
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.0353	1.3636	300	0.2813	0.3767
0.2048	2.7273	600	0.2015	0.2829
0.0982	4.0909	900	0.1758	0.2372

Framework versions

• Transformers 4.47.1
• Pytorch 2.4.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0