Adapt custom llama2_chat tokenize to work with llama3

[LeDuySon]

Hi guys, i have customized the https://github.com/OpenAccess-AI-Collective/axolotl/blob/main/src/axolotl/prompt_strategies/llama2_chat.py to work with my own chat template. Now i want to adapt it to llama3, what should i change to work with it? Right now, if i'm using the llama2_chat directly, it will have problem in the image below. Do you guys have any suggestions?

[NanoCode012]

Hey, we just integrated better chat_template support that can allow you to simply swap the template when working with different models.

Looking at your issue however, did you perhaps reuse llama2 tokenizer for llama3?