Module gemma3 

Gemma LLM architecture (Google) inference implementation.

See “Introducing Gemma 3: The most capable model you can run on a single GPU or TPU”

Based on implementations from HuggingFace transformers.

Structs

Config

Model