fix(batch): deepseek-v4 serves concurrent requests (one-row prompt batches, CacheList promotion) - #38

Merged

asher merged 3 commits into

mainfrom

dsv4-batch-decode

Aug 10, 2026

Commits on Aug 9, 2026

feat(deepseek4): batch decode by keeping prompt batches at one row
asher
committed
c870b6a
fix(batch): stop re-merging an already-batched CacheList on admission
asher
committed
d280e87

Commits on Aug 10, 2026

changelog: deepseek4 concurrent serving fix
asher
committed
7892c15