How to use coolthor/gemma-4-12B-it-FP8-dynamic with Transformers:
# Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("coolthor/gemma-4-12B-it-FP8-dynamic") model = AutoModelForMultimodalLM.from_pretrained("coolthor/gemma-4-12B-it-FP8-dynamic", device_map="auto")
https://huggingface.co/google/gemma-4-12B-it/commit/711c1368e39f1712f48ff0eb7bcdbbb760d52db0
Done — synced chat_template.jinja to the upstream fix (711c1368) in commit 425ac4c8. Thanks for the heads-up! 🙏
chat_template.jinja
711c1368
425ac4c8
· Sign up or log in to comment
