w2v-bert-2.0-Chinese-colab-CV16.0-aishell-ark-gs-vtb-new_tokenizer

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: 1.2200
• Wer: 1.7271

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: 4
• seed: 42
• gradient_accumulation_steps: 64
• total_train_batch_size: 256
• 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_ratio: 0.1
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
5.059	0.9976	156	1.4359	1.7492
2.9898	1.9912	312	1.3586	1.8124
2.17	2.9848	468	1.2909	1.7153
1.9683	3.9784	624	1.2440	1.7769
1.6888	4.9720	780	1.2200	1.7271

Framework versions

• Transformers 4.49.0
• Pytorch 2.5.1+cu121
• Datasets 2.17.1
• Tokenizers 0.21.0