feat(mamba2): integrate model loading and recurrent inference - #575

Open

big-hip wants to merge 5 commits into

InfiniTensor:InfiniLM-v0.2.9cfrom

big-hip:feat/mamba2-model-inference

Commits on Sep 17, 2026

fix(linear): isolate packed layouts and finalize tied weights
big-hip
committed
254de97
feat(mamba2): integrate model loading and recurrent inference
big-hip
committed
d337a05
fix(mamba2): reject pipeline service before worker setup
big-hip
committed
5f89ae3

Commits on Sep 20, 2026

fix(graph): preserve attention cache during decode recapture
big-hip
committed
8a591cb

Commits on Sep 21, 2026

test(layers): reuse prefill and decode regression setup
big-hip
committed
9a19d14