config(minimaxm3): refresh full GB200 aggregate AgentX curve / 刷新完整 GB200 聚合 AgentX 曲线 - #2806
Conversation
|
Thanks for the contribution! Please reach out to respective companies' CODEOWNER to fill in the latest PR_REVIEW_CHECKLIST.md before pinging core maintainer on Slack for review. In order for the signoff PR check bot to trigger, you must follow the PR_REVIEW_CHECKLIST.md template correctly, including the phrase For PR verification, add the PR authors are responsible for ensuring that after merging, all GitHub Action jobs fully pass. A lot of the time, failures are just flakes and simply re-running the failed jobs will fix it. See GitHub's docs on re-running failed jobs 感谢你的贡献！请联系相应公司的 CODEOWNER 填写最新的 PR_REVIEW_CHECKLIST.md，然后再在 Slack 上联系核心维护者进行审阅。为了触发 signoff PR 检查机器人，你必须正确遵循 PR_REVIEW_CHECKLIST.md 模板，包括保留英文语句 如需进行 PR 验证，请为此 PR 添加 PR 作者有责任确保合并后所有 GitHub Action 任务完全通过。 很多时候失败只是偶发抖动（flake），重新运行失败的任务即可解决。参见 GitHub 关于重新运行失败任务的文档 |
1 similar comment
|
Thanks for the contribution! Please reach out to respective companies' CODEOWNER to fill in the latest PR_REVIEW_CHECKLIST.md before pinging core maintainer on Slack for review. In order for the signoff PR check bot to trigger, you must follow the PR_REVIEW_CHECKLIST.md template correctly, including the phrase For PR verification, add the PR authors are responsible for ensuring that after merging, all GitHub Action jobs fully pass. A lot of the time, failures are just flakes and simply re-running the failed jobs will fix it. See GitHub's docs on re-running failed jobs 感谢你的贡献！请联系相应公司的 CODEOWNER 填写最新的 PR_REVIEW_CHECKLIST.md，然后再在 Slack 上联系核心维护者进行审阅。为了触发 signoff PR 检查机器人，你必须正确遵循 PR_REVIEW_CHECKLIST.md 模板，包括保留英文语句 如需进行 PR 验证，请为此 PR 添加 PR 作者有责任确保合并后所有 GitHub Action 任务完全通过。 很多时候失败只是偶发抖动（flake），重新运行失败的任务即可解决。参见 GitHub 关于重新运行失败任务的文档 |
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.
❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
Reviewed by Cursor Bugbot for commit d2bcf37. Configure here.
更新 MiniMax-M3 GB200 聚合配置，仅保留 SimpleCPU 卸载配方，并使用已提交的 thinking-on EAGLE3-GQA 接受长度。
d2bcf37 to
e0c832a
Compare
将 MiniMax-M3 GB200 聚合 AgentX 配置与仓库中已提交的 EAGLE3-GQA thinking-on 接受长度保持一致。
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=33660908146 |
为 MiniMax-M3 GB200 聚合式 AgentX 后端启用已挂载的 Hugging Face 缓存，避免投机解码草稿模型重复下载。
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=33663455779 |
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=33666532482 |
恢复 MiniMax-M3 GB200 聚合 AgentX 的完整配置矩阵，使完整扫描覆盖所有已配置的并发点。
RohitNagraj
changed the title
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=33685870198 |
1 participant
Description
Refresh the complete MiniMax-M3 NVFP4 GB200 aggregate Dynamo-vLLM AgentX configuration so the full sweep covers every configured concurrency point. Use the mounted Hugging Face cache for the TP4 SimpleCPU-offload backend and the committed thinking-on MiniMax-M3 EAGLE3-GQA acceptance target.
中文说明
刷新完整的 MiniMax-M3 NVFP4 GB200 聚合 Dynamo-vLLM AgentX 配置，使完整扫描覆盖所有已配置的并发点。TP4 SimpleCPU 卸载后端使用已挂载的 Hugging Face 缓存，并采用仓库中已提交的 MiniMax-M3 thinking-on EAGLE3-GQA 接受长度目标。
Related Issue
N/A
Type of Change
Checklist
perf-changelog.yamland have not edited historical entriesOWNER/MEMBER/COLLABORATOR) has commented/reuse-sweep-runon this PR. Do this only once there is a final full sweep that is all green with evals passing, since after this comment the sweep label will no longer automatically kick off new sweeps. Remove and re-add the label to force one.Note
Low Risk
Benchmark recipe and changelog-only changes with no application logic or security impact.
Overview
Updates the GB200 TP4 aggregate Dynamo-vLLM SimpleCPU-offload agentic recipe (
agg-tp4-vllm-simple-agentic.yaml) soaggregated_environmentsetsHF_HUB_CACHE,HUGGINGFACE_HUB_CACHE, andTRANSFORMERS_CACHEto/hf_hub_cache, aligning model download/cache behavior with other MiniMax-M3 agentic recipes that use the mounted hub cache.Adds a
perf-changelog.yamlentry forminimaxm3-fp4-gb200-dynamo-vllm-agentic-agg-mtp(agentic-coding) documenting a refresh of that aggregate AgentX configuration, including the HF cache change and the committed thinking-on EAGLE3-GQA acceptance target.Reviewed by Cursor Bugbot for commit d8d4221. Bugbot is set up for automated code reviews on this repo. Configure here.