w2v-bert2-pashto-augmented

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

• Loss: 0.5954
• Wer: 0.3431

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: 4
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• 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
• training_steps: 700
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.0422	1.1713	100	3.0380	0.9640
2.3141	2.3426	200	2.0336	0.9464
0.7365	3.5139	300	0.6768	0.4520
0.557	4.6852	400	0.6051	0.3913
0.5101	5.8565	500	0.6571	0.3853
0.3803	7.0278	600	0.5946	0.3497
0.2452	8.1991	700	0.5954	0.3431

Framework versions

• Transformers 4.40.2
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1