fix(mq4v2): one MQ-V2 prefill admit rule for llama and qwen35; discriminating GEMM parity; spec §9 - #690

Merged

Kaden-Schutt merged 5 commits into

fix/mq4v2-followups

Sep 13, 2026

Commits on Sep 4, 2026

fix(prefill): share MQ-V2 admit rule between llama and qwen35 gates
Kaden-Schutt
committed
8bd5e34
fix(mq4v2): discriminating disjoint-halves GEMM parity arm + doc corrections
Kaden-Schutt
committed
db9c1b2
style: rustfmt changed files
Kaden-Schutt
committed
077aac5
chore: refresh crate maps (check-crate-maps --write)
Kaden-Schutt
committed
8d4450b
fix(runtime): llama never admits MQ-V2 to batched prefill; forward_prefill_chunk has no V2 arms
Kaden-Schutt
committed
4e360fb