erfanzar / EasyDeL

Accelerate your training with this open-source library. Optimize performance with streamlined training and serving options with JAX. 🚀
https://easydel.readthedocs.io/en/latest/
Apache License 2.0
168 stars 19 forks source link

Use custom tokenizer #39

Closed JinSeoungwoo closed 8 months ago

JinSeoungwoo commented 8 months ago

I want to add a token to the tokenizer of my Mistral model and train it. what settings should I change? Only embedding size?

erfanzar commented 8 months ago

yes in order to train you model with custom embedding layer and custom tokenizer you should only change the config and replace the wte data with old wte