Gemma3-DPO

[Kshitiz-Khandel]

Does Axolotl support DPO training for Gemma 3? I didn’t see any DPO examples or config files for Gemma 3 in the Axolotl examples/ directory.

[NanoCode012]

As a text only training, you can use regular config + rl: dpo.

However, I don’t think we support for multimodal yet