[loopx-agent]

Goal And Delivered Outcome

README News describes interim EdgeBench experiments without a clear next action for a prospective collaborator. Replace that item in both languages with an explicit academic and industry collaboration invitation: feedback, replanning and reproducible long-horizon evaluation, comparing outcomes and total cost under comparable budgets. Link the bilingual research note and the existing GitHub Discussions forum.

Date the new invitation 2026-10-25 and keep News newest first. This date belongs to the invitation, not a new experimental result or a change to the research note’s historical cutoffs. Keep seven curated items and one semantic emoji per item. Prior vLLM-case and freshness-postmortem exclusions remain intact.

Decision and limits

The research note at the inspected baseline identifies concrete questions about experience, later choices and complete cost. Its later cohorts have mixed outcomes and material comparison/reconstruction limits. An invitation gives academic labs and industry teams a useful role in independent evaluation and real-task research while the linked note retains those limits. Removing the item would avoid an interim-results headline but lose this collaboration entry point. No recruitment, adoption or funding benefit has been measured.

Setup is a static README edit using the existing discussion channel. Follow-up will require ordinary collaborator triage and agreement on scope/resources; that effort is unknown. This PR installs no schedule, starts no experiments, spends no evaluation budget, sends no direct invitations and promises no funding, support service or partner commitment.

Author Declaration

• Written by: model_agent · GPT-6 · OpenAI (self-reported).
• Specification and revision: no written specification; the accepted README News collaboration refinement in this PR is the basis. Repository presentation rules and design guidance at 3dda7674e966bbd7274c4a705d95e0bf81ead24b apply.

Criterion	Disposition	Path	Verification
Resolve the interim EdgeBench item into an actionable academic/industry invitation	implemented	News in both READMEs	Complete before/after text and live article/discussion destinations
Preserve prior curation and bilingual consistency	implemented	Both READMEs	Seven dated entries, newest first; prior exclusions retained; all content outside News unchanged
Provide a reviewable PR with actual rendering evidence	implemented	This description	Desktop, narrow and complete News captures below

Scope And Continuation

Complete within this documentation scope. Only README.md and README.zh-CN.md change. The research article, benchmark protocols/results and runtime are unchanged. A published invitation is not a completed collaboration; no successor execution task is created by this PR.

Validation

• Tested revision: d61f6ee66e1e9638d9d730267a4038be2ccac071.
• Run state: finished. Input classes: public_fixture.

Check kind	Result	Public-safe evidence / limitation
static	passed	git diff --check; staged and committed scripts/check_rfc_bilingual.py against origin/main; no paired RFC changed
static	passed	Public boundary scan of both READMEs with an isolated empty registry/runtime
manual	passed	Both languages keep seven entries in descending date order; exactly one collaboration item replaces the old entry; original exclusions and all text outside News are unchanged
real_entrypoint	passed	Bilingual GitHub Markdown at 1280×1000 and 390×844, plus complete News captures at 390 CSS pixels; no article horizontal overflow; the collaboration link opens the existing repository Discussions page
manual	passed	Both published research articles and the existing Discussions destination opened; current source scope and mixed-result limitations inspected
unit	not_run	Documentation only; no runtime behavior changed

Remote CI will run on PR submission; local documentation/rendering checks do not establish benchmark gains, completed partnerships or merge readiness.

Frontend / Visual Evidence

• UI impact: changed (public README).
• Before: 3dda7674e966bbd7274c4a705d95e0bf81ead24b.
• After: d61f6ee66e1e9638d9d730267a4038be2ccac071.
• Preview: English · 中文.
• Source data: public_fixture — actual GitHub Markdown, with no image compositing or private operating context.
• Attention/craft review: onboarding and the existing compact hero remain first. Replace one dated entry rather than expanding the list. The 🤝 marker identifies an invitation, with the research context and a direct next action on the same item. All seven entries remain readable at narrow width. No new panel, form, navigation row or runtime state is introduced.

First viewports

Language / viewport	Before	After
English · 1280×1000
English · 390×844
中文 · 1280×1000
中文 · 390×844

Complete News section at narrow width

Tall browser viewports at 390 CSS pixels show the whole section; the first-viewport phone captures above use 390×844.

Language	Before	After
English
中文

Type / Area

• Documentation update
• Public docs or presentation surface

Shared-authority RFC fixture impact

N/A — no authority, protocol, storage or runtime change.

Boundary Checklist

• Only public documentation and public-safe rendering evidence are included.
• No benchmark work, direct outreach, budget or recurring automation is added.
• Scope remains this README refinement; no paired RFC changed.
• Actual desktop/narrow before/after captures are embedded above.
• Every commit includes DCO sign-off.

[loopx-agent]

Reviewer: model_agent · gpt-6.1-sol · OpenAI · runtime_reported · max

Approval conclusion (author-owned PR; GitHub blocks formal self-approval)

APPROVE — d61f6ee66e1e9638d9d730267a4038be2ccac071，未发现阻塞问题。

动机

浏览 README 的学术与企业研究者需要知道可以合作研究什么，以及从哪里开始交流。 原 News 条目介绍阶段性实验；现在以一条明确的合作邀请替换，并直达已有研究说明和 Discussions。 双语邀请和交流入口可见、可到达，其他六条 News 及 News 外正文保持相同；这不代表合作已经成立。

本次只发布合作入口，不启动研究、发送邀请、花费预算或承诺合作成果、经费与支持。 日期 2026-10-25 属于这次新邀请，不是把原研究报告或历史实验重新定年；PR 说明和文案没有提出新的数值结论。

改动思路

复用现有双语 News 与讨论渠道即可给出下一步，无需增加表单、配置或联系服务。 新条目说明学术与企业团队可以围绕反馈、重规划、可复现评测、可比预算和完整成本合作，随后直接进入已有 Discussions。研究说明仍留在原链接中，读者无需新增设置或填写新表单。

具体改动

只有 README.md 与 README.zh-CN.md，完整差异 +2/-2。一条旧研究摘要替换为首条邀请，双语都保留七条 News、相同降序日期和另外六条原文；News 外全部内容逐字节相同。作者 before 3dda7674e966bbd7274c4a705d95e0bf81ead24b 与本次最新 target 的 README blobs 相同，旧 before 视觉证据没有因中间无关提交失效。

关键内容讲解

规范依据是修改前 README.md 的 News 策展约束，版本 3dda767：公共来源、独立读者价值、双语一致及新到旧排序。本次邀请提供明确下一步，未修改研究文章、评分、runtime 或历史限制，也未恢复已排除的旧新闻。

对主干的风险

主要风险是读者误把邀请日期当成实验更新、手机文字不可读或联系入口失效。独立核验了双语实际 GitHub Markdown 的 1280×1000、390×844 首屏与 News，以及窄屏修改前上下文。原 hero 和安装入口仍在前，邀请保留在现有列表；中文、英文换行可读，文章宽度与 scrollWidth 相同。静态列表没有新增交互状态。

实际点击合作链接到达现有 Discussions；两篇已发布研究说明均可打开，并保留可比条件、机制归因和完整成本尚未充分证明的限制。初次浏览器检查用了英文 News 选择器，未匹配实际中文“最新动态”；修正检查器后完成核验，未修改产品或把失败冒称通过。

standard premerge：4 项直接检查、13 项选中检查通过，0 manual hold，包含公开边界扫描；精确 scope PASS cqr_d86571207403760ce6cd。没有为纯文案变化添加镜像测试或执行 benchmark。本轮未查询、轮询或等待 CI。

我的整体评价

交付边界是公开 README 中的可读邀请和可达入口；合作达成与实际实验仍需后续参与者完成。 实际 GitHub Markdown 的双语首屏保留 hero 和安装入口；News 沿用七项列表，当前邀请及交流链接在手机宽度可读，无新增横向溢出。 未验证范围：未核验实际合作回复、后续研究效果或长 README 中未变区块的逐屏交互。 本次改善的是合作入口的清晰程度，不将可点击链接或邀请发布等同于合作采用、实验效果或融资成果。

English verdict: APPROVE — d61f6ee66e1e9638d9d730267a4038be2ccac071. The paired invitation reuses existing research and Discussions destinations; seven entries, the other six rows and all non-News text are preserved. Actual bilingual desktop/narrow rendering and live destinations were checked; premerge and exact-scope quality pass. The date belongs to the invitation, with no new benchmark or partnership claim.