feat(rocmdoc): migrate to ROCmDoc v2 standard (central ccd466e) - #66
Draft
AIwork4me wants to merge 8 commits into
Draft
feat(rocmdoc): migrate to ROCmDoc v2 standard (central ccd466e)#66AIwork4me wants to merge 8 commits into
AIwork4me wants to merge 8 commits into
Conversation
- adapter/run_adapter.py: platform adapter (smoke + SGLang backends) - eval/configs/omnidocbench_v16.yaml: scoring config - model_card.json: platform model card - README.md: 5 conformance sections (Install, Demo, Evaluation, Reproducibility, Known Gaps) - README.zh-CN.md: Chinese translation - pyproject.toml: add omnidocbench-rocm platform extra Gates: CONFORMANT
Add v2 conformance surface: rocmdoc.yaml, model_card_v2.json, .rocmdoc/spec-lock.json, standard CLI (version/capabilities/doctor --json/parse, central exit codes 0/1/2/3/4/5), NOTICE/REPRO/PATCHES, migrations + benchmark-context, deterministic README results block, contract tests. model_card.json stays v1 (central check_repo requires schema_version:const 1). Locked to central commit ccd466ef317fd6a710131db3a19ec9d55a65ce2e. No GPU run; no assurance raised; no historical result altered. Co-Authored-By: Claude <noreply@anthropic.com>
- producer_assurance + run_spec/run_spec_hash + comparison_track_id per result; pytorch-fast (full) and vllm (canary) are distinct tracks (full vs canary) - licenses split: code MIT / model unknown (Baidu Unlimited-OCR weights are NOT MIT — corrected from a false open-source-ai; weights license unverified) - spec-lock -> central 4556ee1. Tests: 229 pass. No public valid score without canonical evidence; platform_review stays not-reviewed. Co-Authored-By: Claude <noreply@anthropic.com>
Re-point .rocmdoc/spec-lock.json to frozen ROCmDoc 1.0 central contract (17b5d65652ddab342a1abedad487d90ec2c4393f, PR AIwork4me/OmniDocBench-ROCm#31). Fix schema typos table_tides_percent -> table_teds_percent and table_tides_structure_only_percent -> table_teds_structure_only_percent (canonical per artifact-schema.json:98; values unchanged; result_ids unchanged). PyTorch-fast 92.451 remains primary (submitted); vLLM 22.3 remains non-primary/non-supported failing measurement (honestly flagged). Verified: CONFORMANT; MANIFEST VALID; README drift OK; 229 passed/3 skipped. Phase-3 subagent review: CLEAN. Stage ONLY spec-lock + card (4 pre-existing modified scripts/*.sh are user env-work, explicitly excluded). Co-Authored-By: Claude <noreply@anthropic.com>
Co-Authored-By: Claude <noreply@anthropic.com>
… ruff commit The prior ruff-format commit used 'git add -A' and included 18 unrelated pre-existing working-tree edits (scripts/*.sh, docs/superpowers/plans|specs/*.md, docs/upstream, examples/README.md). This reverts non-.py paths to 38104be so the PR carries only the 3 legitimate ruff-format .py changes. The WIP is preserved as uncommitted working-tree changes (restored separately). Co-Authored-By: Claude <noreply@anthropic.com>
# Conflicts: # adapter/run_adapter.py
…lock CI) Co-Authored-By: Claude <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Migrate Unlimited-OCR-ROCm to central ROCmDoc v2, locked to
ccd466e.torch/torchvision/torchaudio/transformersmoved core →[backend]extra (version/capabilities/doctor now run without torch); standard subcommands added with legacy flat-CLI (--pdf/--image-dir) +--versionretained; NOTICE + REPRO.yaml + PATCHES.md created (vLLM patch governance);mainnow returns int.Conformance
structural CONFORMANT · base/runtime-core PASS · benchmark-v16 PASS(fixture) · reproducible-score FAIL (no platform evidence — honest) · pytest 229 passed.
Primary: pytorch-fast/linux-rocm 92.451 (
submitted, 4-GPU). v1overall=nullexplained.Status: DRAFT (WIP)
No GPU run; no score altered. Note: 4 user-modified
scripts/*.shleft untouched (not in this PR).🤖 Generated with Claude Code