testing_tensorboard_w_new_access_token

This model is a fine-tuned version of facebook/w2v-bert-2.0 on the common_voice_17_0 dataset. It achieves the following results on the evaluation set:

• Loss: 3.1867
• Wer: 1.0
• Cer: 0.9653

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: 0.0001
• 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: constant
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
3.3114	0.6154	200	3.2030	1.0	1.0
3.1797	1.2308	400	3.1973	1.0	1.0
3.1791	1.8462	600	3.1899	1.0	1.0
3.1767	2.4615	800	3.1787	1.0	1.0
3.1681	3.0769	1000	3.1870	1.0	0.9987
3.1783	3.6923	1200	3.1996	0.9998	0.9822
3.167	4.3077	1400	3.1726	1.0	1.0
3.171	4.9231	1600	3.1743	1.0	0.9653
3.1654	5.5385	1800	3.1926	1.0000	0.9487
3.1714	6.1538	2000	3.1700	1.0	0.9653
3.1638	6.7692	2200	3.1688	1.0	0.9653
3.164	7.3846	2400	3.1934	1.0000	0.9487
3.1729	8.0	2600	3.1689	1.0	0.9653
3.1652	8.6154	2800	3.1660	1.0	0.9653
3.1569	9.2308	3000	3.1890	1.0000	0.9487
3.1639	9.8462	3200	3.1867	1.0	0.9653

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1