Skip to content

knowledge: remove the CannBot stage-evidence distillation from #114 - #489

Draft
jiangkuaixue123 wants to merge 2 commits into
InferMatrixLab:mainfrom
jiangkuaixue123:codex/remove-cannbot-distillation
Draft

jiangkuaixue123 wants to merge 2 commits into
InferMatrixLab:mainfrom
jiangkuaixue123:codex/remove-cannbot-distillation

Conversation

@jiangkuaixue123

Copy link
Copy Markdown
Collaborator

Refs #91 and #476. Follow-up to #484.

The maintainer requested removal of the direct CannBot distillation introduced by #114. This removes its stage-report procedure, readset, accompanying license copy, review cases, and current coverage entries. The Ascend profiling procedure matches its pre-#114 version. The older mirrored skills, external lookup adapter, and historical campaign snapshots are preserved.

The packaged skill index and coverage checks succeeded. The skill catalog, index, and coverage tests passed locally.

{"schema_version":1,"head_sha":"91a40ba22dbf0c67209d0ec2ec6354e45b184f7f","human_only":false,"material_ai_authors":[{"tool":"Codex","model_id":"unknown","effort":"unknown","role":"author"}]}

…ixLab#114

Signed-off-by: jiangkuaixue123 <jiangxiaozhou111@163.com>
Signed-off-by: jiangkuaixue123 <jiangxiaozhou111@163.com>
@github-actions

github-actions Bot commented Nov 18, 2026 •

Copy link
Copy Markdown

Skill evaluation

PLANNED: not executed; there are no task outcomes yet.
No approval, model efficacy or current-head test result is implied.

  • Next: run the pinned source/index --check below at head 91a40ba22dbf0c67209d0ec2ec6354e45b184f7f.
  • Task selection: no supported automatic comparison; choose a relevant source/content check and disposition uncovered claims.
  • Selected comparisons: 0 · Estimated task calls: 0. Uncovered/unsupported changes: 12; Canonical library paths: 6; Generated projection paths: 2; Other paths: 4. These are paths, not independent claims.
  • Return: command, input/source pin, exit status, outputs/failures and untested claims for independent verified/refuted/unverified disposition.
Author verification steps (nonpaid; not yet run)
  1. Bind the checkout. Run git rev-parse HEAD; it must equal 91a40ba22dbf0c67209d0ec2ec6354e45b184f7f. If it differs, obtain a current plan; do not reuse old results.
  2. Check source/index consistency. At that PR checkout, using the reviewed repository tool and its documented development dependencies, run:
python tools/build_skill_index.py --source-revision 4eef72ee00e8d1f7792c79d79519da60705608bd --check

The declared source pin matches this head's canonical library bytes. The command has not run: retain its exit code and full drift/success output. It checks the generated projection, not claim truth or skill efficacy.
No supported automatic comparison is selected. Review the canonical changes and explain which claims need evidence, are unaffected, or remain unverified; do not treat generated records as separate behavioral claims.
Uncovered scope by location: Canonical library paths: 6; Generated projection paths: 2; Other paths: 4. Path counts are not claim counts; no metadata-only conclusion is inferred.
3. Return evidence for independent review: head, check/task/source pin, command, environment, exit status, outputs/failures and untested claims. Label scripted controls. Request each claim's disposition: verified, refuted or unverified; this plan proves neither model provenance nor benefit.
Check recipes and evidence limits

Evidence and logs · Retention: 30 days.
Retained artifacts are verifiable only before Unix time 1797575033 while the linked artifact is available. Expired evidence is unverified.

Plan bindings, budget, frozen exam and advisory details

Advisory: obtain missing evidence

Retained evidence availability: planned / unexecuted. No approval or merge authorization is issued.
Applicable only while these revisions and inputs match the current PR.
Paid execution requires trusted label authorization and cumulative PR reservation. Unknown accounting blocks new paid calls; planning is not execution.
Pipeline evidence only; offline fixed outputs do not establish skill efficacy.
Unchanged-correct is valid neutral evidence, not incremental benefit.
Content/source review and authorship admission remain separate.
Trusted regrading checks retained answers, not model/hardware provenance or a hostile executor; this is not an efficacy certification.

  • Retained evidence is planned / unexecuted.
  • No supported comparison cases were selected.
  • Uncovered/unsupported scope needs a case or explicit reviewer disposition.
    Plan SHA-256: b73ffeea1909293525b82d1bda317ac0255960e035a81b39a7bd7ed4bf23afab
    Base: 8aba657da79e8ed9f6e6fbb33437e9d9fdd8f1c7 · Head: 91a40ba22dbf0c67209d0ec2ec6354e45b184f7f
    Trusted implementation: 8aba657da79e8ed9f6e6fbb33437e9d9fdd8f1c7

Model/settings: {"effort":"not_applicable","name":"claude-opus-5-5"}
Budget: {"comparisons":0,"estimated_calls":0,"limits_per_comparison":{"campaign_seconds":600,"max_task_calls":2,"repeats":1,"timeout_seconds":180,"token_stop_new_trials":200000},"provider_config_sha256":"b28d80f8194fbe0f5a9f01515efb24b218979fd45f702776560da276f4b30d04","reservation":null}

Skill Comparison Cases Estimated calls

Automatic execution / cumulative PR budget:

{"reason":"Trusted configuration or pricing missing, invalid or expired","remaining_usd":"unknown","reserved_or_unresolved_usd":"unknown","spent_or_legacy_usd":"unknown","state":"blocked"}

Projection check selection (not an executed result):

{"declared_in":"src/inferrsi/knowledge/data/repo_packs/inferrsi/knowledge/_provenance.json","source_revision":"4eef72ee00e8d1f7792c79d79519da60705608bd","status":"ready"}

Provider/settings: {"client":{"api_version":"2023-06-01","endpoint":"https://yibuapi.com/v1/messages","kind":"anthropic-messages-full-context-v1","max_output_tokens":4096,"max_read_bytes":32768,"max_request_bytes":262144,"max_requests":1,"max_tool_calls":12,"provider":"yibuapi"},"limits":{"token_stop_new_trials":200000},"model":{"effort":"not_applicable","name":"claude-opus-5-5"},"pricing":{"currency":"USD","expires_at":1793271228,"group":"default","max_input_tokens":1000000,"model":"claude-opus-5-5","rate_evidence":"https://github.com/InferMatrixLab/InferRSI/issues/215#issuecomment-6467531341","rates":{"cache_read":"0.4","cache_write":"10","input":"8","output":"40"},"request_fee":"0","unit":"per_million_tokens"},"spend_limit":{"amount":"200","currency":"USD"}}

Uncovered / unsupported paths (12)
  • uncovered: "docs/reference-matrix.md"
  • uncovered: "docs/reviews/issue91-cannbot.md"
  • uncovered: "docs/searchspace-matrix.md"
  • uncovered: "docs/skill-coverage.md"
  • uncovered: "src/inferrsi/knowledge/data/repo_packs/inferrsi/knowledge/_provenance.json"
  • uncovered: "src/inferrsi/knowledge/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-performance-analysis.json"
  • uncovered: "src/inferrsi/knowledge/data/skill_library/hardware/README.md"
  • uncovered: "src/inferrsi/knowledge/data/skill_library/hardware/ascend/ascend-performance-analysis/SKILL.md"
  • uncovered: "src/inferrsi/knowledge/data/skill_library/hardware/ascend/ascend-performance-analysis/metadata.json"
  • uncovered: "src/inferrsi/knowledge/data/skill_library/hardware/ascend/ascend-performance-analysis/references/cannbot-CANN-2.0-LICENSE.md"
  • uncovered: "src/inferrsi/knowledge/data/skill_library/hardware/ascend/ascend-performance-analysis/references/cannbot-stage-evidence.md"
  • uncovered: "src/inferrsi/knowledge/data/skill_library/hardware/source-inventory.json"

@zengchuang-hw zengchuang-hw left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

PR review verdict — DO NOT PASS

Independent first review of base 8aba657da79e8ed9f6e6fbb33437e9d9fdd8f1c7 → head 91a40ba22dbf0c67209d0ec2ec6354e45b184f7f (12 changed files). Previous review: none; resolved prior findings: not applicable.

Reviewer: dot, acting under zengchuang-hw's standing delegation to independently review other contributors' PRs and publish findings. This is a COMMENT, not REQUEST_CHANGES; I am not approving or merging this revision.

Required before approval

  1. Update the installed-resource acceptance test for the intentional deletions. tests/check_installed.py:544 still asserts 930 resources, but this PR removes two packaged reference files. Independently enumerating the exact head's Git tree gives 928 files under the Knowledge/Harness data roots. The Python 3.11 installed-package job fails at that assertion with “installed resource count is 928, expected 930”; Python 3.10 is also failed. Adjust the intended inventory expectation while retaining the complete checkout/wheel byte comparison and registered-resource checks, and rerun both installed-package lanes. Do not restore the unwanted references merely to satisfy the old count.
  2. Resolve the current-head knowledge declaration failure truthfully. The PR declares material author model/effort as unknown. The declarations job evaluated this exact head and reports Material AI author 1: model is unapproved or suspended. Provide verifiable actual author/model/effort provenance meeting the repository policy, or route the unresolved provenance to its policy owner; do not replace unknown values by guesses.

Independent scope checks

  • Reviewed the removal from the skill procedure, sidecar, generated abstract, source inventory, README and both coverage matrices. The removed direct CannBot readset/license and review-case file correspond to the stated #114 distillation slice; the diff does not edit the external lookup adapter or historical campaign evidence.
  • Fetched the pre-#114 skill at 61cdfebfd871fcb007d289d4d1f10c55e65813cd under its old src/inferrsi/data/ path and compared its complete text with the current skill: identical. The existing MindIE-SD five-layer analysis and shared-machine measurement procedure are preserved.
  • Read AGENTS.md, the repository review skill, contribution/collaboration rules and linked #91/#476 scope context. The proposed external-query work and this removal do not establish licensing approval or close the broader governance tasks.
  • Checked all 12 file statuses and the complete recursive tree (not truncated), current head/base, own-review history and GitHub check metadata. No other reviewers' opinions were used.

Verification limits

I performed connector-based static inspection and read the actual CI failure logs; I did not run pytest, Ruff, index/coverage regeneration, wheel installation or hardware experiments locally. Author-reported local tests are not independent reruns. At the latest check, full-check 3.10/3.11 and Windows compatibility were still running, macOS was queued, and browser/skill-eval execution were skipped; plan, publish and PR references passed. Therefore this is not an all-green or complete runtime-validation claim.

Non-blocking suggestion: add a focused regression asserting the two retired references are absent while the preserved skill remains readable, alongside the corrected full inventory test.

@Dong1017 Dong1017 added area:knowledge Knowledge line: packs, skill library, knowledge admission kind:task Bounded delivery task with a delegated slice labels Nov 18, 2026

@zengchuang-hw zengchuang-hw left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

PR review verdict — DO NOT PASS

Evidence-driven incremental review of unchanged head 91a40ba22dbf0c67209d0ec2ec6354e45b184f7f (base 8aba657da79e8ed9f6e6fbb33437e9d9fdd8f1c7), following my initial review. No new commits; this update records a newly completed full-check failure, not a duplicate review of unchanged evidence.

Additional required correction: src/inferrsi/knowledge/data/skill_library/hardware/ascend/ascend-performance-analysis/metadata.json:16 replaces the removed source with "sources": []. The existing contract in tests/test_knowledge_contract.py:37–48 deliberately requires omission of empty reader-default fields. Full-check Python 3.11 now confirms this exact assertion fails for ascend-performance-analysis (1 failed, 5075 passed, 36 skipped, 671 subtests passed). Remove the empty sources field rather than weakening the contract, then regenerate the generated index/provenance and coverage as required and rerun the relevant/full checks. Preserve the remaining MindIE-SD attribution in the procedure/inventory; do not invent a normalized source record merely to silence the test.

The installed-resource-count and declaration blockers from the prior review remain unresolved. Windows compatibility also fails on the same 928 versus 930 installed-resource assertion, independently confirmed in its raw job log. No prior blocker is closed by this update.

Actual verification: independently read the sidecar and contract test at this exact head and the newly finished raw CI logs, then rechecked current head and own posted review. No other review opinions used; no local tests or repository code execution, no branch edit, approval or merge. Reviewer remains dot acting for zengchuang-hw under the standing review/publication delegation. No additional non-blocking suggestion.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

area:knowledge Knowledge line: packs, skill library, knowledge admission kind:task Bounded delivery task with a delegated slice

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants