How to fine tune a base model?

[aw632]

I am trying to fine tune the google/gemma-3n-E2B · Hugging Face model, which is a base model (NOT instruct). How do I fine tune this model on my dataset which I prepared? All the guides I see on HF for finetuning seem to use chat template models, I can’t find any tutorials for fine tuning VLM that don’t have chat templates.

[John6666]

Whether or not it is necessary to apply a Chat Template to chat-style text data in advance to convert it into a predefined format is the only difference in the tasks required for fine-tuning, whether it is applied to the base model or the Instruct model. However, best practices in terms of accuracy may vary…
Google appears to have provided an explanation regarding fine-tuning as a VLM.

Other resources:

Fine-Tuning a Vision Language Model (Qwen2-VL-7B) with the Hugging Face...

Fine-Tuning Your First Large Language Model (LLM) with PyTorch and Hugging Face