Instructions to use google/gemma-4-12B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use google/gemma-4-12B with Transformers:
# Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("google/gemma-4-12B") model = AutoModelForMultimodalLM.from_pretrained("google/gemma-4-12B") - Notebooks
- Google Colab
- Kaggle
Model doesn't support images in tool call outputs
#10
by dpe1 - opened
It hangs for a while, I suspect it gets fed base64 string instead of an image. And then it starts spamming same line something like think start think end. Compared to qwen3.5 which works well with tools that return images
Hi @dpe1 ,
Thanks for reporting this issue. Could you please share us the exact code snippet and full traces of output for generating this issue which we be helpful for us investigate further.