w2v-bert-2.0-Vietnamese-colab-CV17.0

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: 1.0607
• Wer: 0.2729

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: Use OptimizerNames.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: 500
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
2.8799	3.2609	300	0.7434	0.3899
0.1626	6.5217	600	0.8157	0.3578
0.0823	9.7826	900	0.8759	0.3704
0.04	13.0435	1200	0.9129	0.3195
0.0169	16.3043	1500	0.9113	0.2904
0.0056	19.5652	1800	0.9906	0.2809
0.0016	22.8261	2100	1.0506	0.2848
0.0005	26.0870	2400	1.0502	0.2730
0.0002	29.3478	2700	1.0607	0.2729

Framework versions

• Transformers 4.50.0
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1