KaiDong-coder/hotset

gpt-oss-120B inference on one 48 GB GPU: model-exact placement, MXFP4 mul_mat_id fast paths, and a routing-traced byte model for what comes next.

1 good first / help-wanted issue · C++ · last activity Oct 22, 2026

cuda gpt-oss inference-engine llama-cpp llama-cpp-fork llm-inference local-llm mixture-of-experts moe mxfp4

1 Open Issue Need Help Last updated: Oct 22, 2026

Open Issues Need Help

good first issue

gpt-oss-120B inference on one 48 GB GPU: model-exact placement, MXFP4 mul_mat_id fast paths, and a routing-traced byte model for what comes next.

C++

#cuda#gpt-oss#inference-engine#llama-cpp#llama-cpp-fork#llm-inference#local-llm#mixture-of-experts#moe#mxfp4