Edit model card

bert-base-cantonese

This model is a continue pre-train version of bert-base-chinese on Cantonese Common Crawl dataset with 198m tokens.

Model description

This model has extended 500 more Chinese characters which very common in Cantonese, such as 冧, 噉, 麪, 笪, 冚, 乸 etc.

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: 24
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 8
  • total_train_batch_size: 192
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 1.0

Training results

Framework versions

  • Transformers 4.35.0.dev0
  • Pytorch 2.1.1+cu121
  • Datasets 2.14.6
  • Tokenizers 0.14.1
Downloads last month
76
Safetensors
Model size
103M params
Tensor type
F32
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for indiejoseph/bert-base-cantonese

Finetuned
(147)
this model
Finetunes
1 model