feat(engine): integrate prefix eviction and chunked parallel execution - #573

Open

big-hip wants to merge 9 commits into

InfiniTensor:InfiniLM-v0.2.9cfrom

big-hip:feat/cache-chunk-graphs

Commits on Sep 16, 2026

feat(engine): integrate prefix eviction and chunked parallel execution
big-hip
committed
29d915d
style: align Python formatting with CI Ruff version
big-hip
committed
32d00aa
docs: quantify cache chunking and graph tradeoffs
big-hip
committed
c7599cd
refactor(scheduler): reuse decode scheduling and test setup
big-hip
committed
a7345fa
chore: move experiment artifacts out of the source diff
big-hip
committed
f6eaf6c
refactor(engine): reuse native input conversion
big-hip
committed
b509415
refactor(engine): defer experimental prefill graphs
big-hip
committed
04022fa

Commits on Sep 17, 2026

test(engine): consolidate cache and chunk configuration coverage
big-hip
committed
cb243d9

Commits on Sep 21, 2026

test(engine): trim chunk validation scaffolding
big-hip
committed
6ab1130