[None][perf] Allow prefix-tokenization cache for MiniMax-M3 text-only prompts - #19613

Open

zheyuf wants to merge 6 commits into

NVIDIA:mainfrom

zheyuf:zheyu/perf/minimaxm3-prefix-token-cache-main

Commits on Oct 5, 2026

[None][perf] Use the prefix-tokenization cache for MiniMax-M3 text-only prompts
zheyuf
committed
517b1f3
[None][perf] Drop the MiniMax-M3 prefix-token-cache unit test and docs change
zheyuf
committed
b321c71
[None][perf] Serve MiniMax-M3 prompts from the cache under DefaultInputProcessor's rules
zheyuf
committed
39a7179

Commits on Oct 6, 2026

[None][perf] Share the tokenization-cache hooks across multimodal input processors

zheyuf
and
claude
committed
0801b9d
[None][perf] Trim comments and drop getattr in the tokenization-cache hooks

zheyuf
and
claude
committed
f39f7bf
[None][perf] Document the tokenization-cache opt-in contract

zheyuf
and
claude
committed
4eec68b