Update the weights to match the reflect update from google
hi @pythiccoder !
the common change across models (the chat template updating) was done on this model too via this PR: https://huggingface.co/google/gemma-4-E2B-it-qat-mobile-transformers/discussions/1
are you missing anything in particular?
Luciano Martins.
Thanks for the clarification when I saw the original update I assumed there was additional training to adapt the template. I understand now that there was a compatibility bug in the chat template that affected tool use correct?
Hey @pythiccoder ,
It was not necessarily a compatibility bug - we detected (and learned from users) that some tool usage scenarios are not being handled by the chat template in the best possible way... so we decided to make it better while also applying some feedbacks.
The only effort on your side, as a dev using Gemma4, is to validate if your dev setup has the latest chat_template available.
cheers,
Luciano Martins.