vLLM(@vllm_project):MTP3 was checked alongside the baseline on B300 DEP8: AIME25: 0.9500 → 0.9667 GPQA: 0.9268 → 0.9242 OCRBench: 0.896 → 0.895 Two validation rounds completed with zero request errors. These are results from the tested build, not a claim that final-release validation is complete. 4/6

vLLM

@vllm_project

A high-throughput and memory-efficient inference and serving engine for LLMs. Join to discuss together with the community!

가입 March 2024

37 팔로잉 중 54.1K 팬

vLLM@vllm_project

2026.10.15 03:26

MTP3 was checked alongside the baseline on B300 DEP8: AIME25: 0.9500 → 0.9667 GPQA: 0.9268 → 0.9242 OCRBench: 0.896 → 0.895 Two validation rounds completed with zero request errors. These are results from the tested build, not a claim that final-release validation is complete. 4/6