fix: use vision_size for Gemma4 specializations to support chunked embedding in vLLM v1 (#1135) - #1146

fix: use vision_size for Gemma4 specializations to support chunked embedding in vLLM v1 (#1135)#1146

Rishin Raj (quic-rishinr) merged 1 commit into

qualcomm:release/v1.22.0qualcomm/efficient-transformers:release/v1.22.0from

quic-gthiruko:cherry-pick/vision-size-gemma4-v1.22.0quic-gthiruko/efficient-transformers:cherry-pick/vision-size-gemma4-v1.22.0Copy head branch name to clipboard