[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M793H9JSJABGFQJTJFMB17QB":3,"skill-same:01M793H9JSJABGFQJTJFMB17QB":71},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":27,"commitSha":28,"skillDirectory":29,"forksCount":30,"repositoryHtmlUrl":31,"repositoryHomepageUrl":6,"ownerHtmlUrl":32,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":33,"install":34,"resources":46},"01M793H9JSJABGFQJTJFMB17QB","evaluation-and-fusion",null,"Evaluate Pyserini run files, convert run formats, fuse retrieval runs, and use reproducibility matrices safely.","VectorSpaceLab\u002FAREX-Skill","VectorSpaceLab","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F180637102?v=4","01M7538C5WPSJGWF7KHR4EFZ9F","01M7538C6R2GFP53NPXTCZNBRT",351,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fpyserini\u002Fsub-skills\u002Fevaluation-and-fusion\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: evaluation-and-fusion\ndescription: \"Evaluate Pyserini run files, convert run formats, fuse retrieval\n  runs, and use reproducibility matrices safely.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\nlicense: Apache 2.0\n---\n\n# Evaluation And Fusion\n\nUse this sub-skill when the task involves `trec_eval`, qrels, TREC\u002FMS MARCO\u002FKILT\u002FDPR\u002FQA retrieval evaluation, run-file validation, run conversion, reciprocal-rank fusion, interpolation, score normalization, or Pyserini two-click reproducibility matrices.\n\nDo not use this sub-skill to generate search runs from an index. Route Lucene indexing\u002Fsearch\u002Ffetch workflows to `..\u002Findex-search-fetch\u002FSKILL.md`, dense encoding\u002FFaiss\u002Fhybrid retrieval generation to `..\u002Fdense-encoding\u002FSKILL.md`, and Java\u002Feval-jar\u002Fsource-checkout build failures to `..\u002Finstall-and-runtime\u002FSKILL.md` or `..\u002Frepo-development\u002FSKILL.md`.\n\n## Quick Routing\n\n- **TREC evaluation:** Validate a six-column TREC run and qrels, then run `python -m pyserini.eval.trec_eval` with explicit metrics.\n- **MS MARCO evaluation:** Keep three-column MS MARCO runs for `msmarco_passage_eval`\u002F`msmarco_doc_eval`, or convert to six-column TREC before `trec_eval`.\n- **KILT, DPR, and QA retrieval:** Use the dedicated JSON\u002FJSONL evaluators only after checking the expected schema and avoiding implicit annotation downloads.\n- **Run filtering:** Use `pyserini.trectools.TrecRun` and `Qrels` to inspect topics, retain or discard judged documents, normalize scores, or write a clean TREC run.\n- **Fusion:** Use `python -m pyserini.fusion` for `rrf`, `interpolation`, `average`, and `normalize` on valid TREC runs with compatible query\u002Fdocument identifiers.\n- **Reproducibility matrices:** Prefer `--display-commands --dry-run` first; treat full two-click commands as opt-in because they may download indexes, models, topics, qrels, or runs.\n\n## Default Safe Workflow\n\n1. Validate the candidate run before evaluation or fusion:\n\n   ```bash\n   python scripts\u002Fvalidate_trec_run.py run.trec --qrels qrels.txt --strict-ranks --summary\n   ```\n\n2. If the run is MS MARCO format, convert it before using `trec_eval`:\n\n   ```bash\n   python -m pyserini.eval.convert_msmarco_run_to_trec_run --input run.msmarco.tsv --output run.trec\n   ```\n\n3. Evaluate with explicit metrics and a qrels key or qrels path:\n\n   ```bash\n   python -m pyserini.eval.trec_eval -c -m ndcg_cut.10 -m recall.100 qrels.txt run.trec\n   ```\n\n4. Build a fusion command without executing it:\n\n   ```bash\n   python scripts\u002Ffusion_recipe_builder.py --method rrf --runs bm25.trec dense.trec --output fused.trec --runtag demo --k 100 --depth 1000\n   ```\n\n5. Validate the fused run and evaluate it with the same qrels and metrics as the input runs.\n\n## References\n\n- `references\u002Frun-evaluation.md` covers TREC, MS MARCO, KILT, DPR, QA, `TrecRun`, `Qrels`, qrels aliases, and common metrics.\n- `references\u002Freproducibility.md` explains two-click reproduction modules, dry-run\u002Fdisplay-command safety, nondeterminism, and matrix triage.\n- `references\u002Ftroubleshooting.md` maps malformed runs, run\u002Fqrels mismatches, missing eval jars, unjudged documents, and expensive reproduction failures to fixes.\n- `scripts\u002Fvalidate_trec_run.py --help` validates TREC runs and qrels without importing Pyserini or starting Java.\n- `scripts\u002Ffusion_recipe_builder.py --help` generates reproducible Pyserini fusion commands with method-specific checks.\n\n## Acceptance Checklist\n\n- Confirm whether the run is six-column TREC or three-column MS MARCO before choosing an evaluator.\n- Keep query IDs and document IDs as strings; do not coerce them to integers unless a specific evaluator requires it.\n- Use the same qrels\u002Ftopic split and metric definitions for baseline, candidate, and fused runs.\n- Validate every input run before fusion and validate the fused output before evaluation.\n- Treat prebuilt-index reproduction and neural retrieval commands as network\u002Fmodel\u002Fcache-sensitive unless the user has explicitly approved them.\n","# Evaluation And Fusion\n\nUse this sub-skill when the task involves `trec_eval`, qrels, TREC\u002FMS MARCO\u002FKILT\u002FDPR\u002FQA retrieval evaluation, run-file validation, run conversion, reciprocal-rank fusion, interpolation, score normalization, or Pyserini two-click reproducibility matrices.\n\nDo not use this sub-skill to generate search runs from an index. Route Lucene indexing\u002Fsearch\u002Ffetch workflows to `..\u002Findex-search-fetch\u002FSKILL.md`, dense encoding\u002FFaiss\u002Fhybrid retrieval generation to `..\u002Fdense-encoding\u002FSKILL.md`, and Java\u002Feval-jar\u002Fsource-checkout build failures to `..\u002Finstall-and-runtime\u002FSKILL.md` or `..\u002Frepo-development\u002FSKILL.md`.\n\n## Quick Routing\n\n- **TREC evaluation:** Validate a six-column TREC run and qrels, then run `python -m pyserini.eval.trec_eval` with explicit metrics.\n- **MS MARCO evaluation:** Keep three-column MS MARCO runs for `msmarco_passage_eval`\u002F`msmarco_doc_eval`, or convert to six-column TREC before `trec_eval`.\n- **KILT, DPR, and QA retrieval:** Use the dedicated JSON\u002FJSONL evaluators only after checking the expected schema and avoiding implicit annotation downloads.\n- **Run filtering:** Use `pyserini.trectools.TrecRun` and `Qrels` to inspect topics, retain or discard judged documents, normalize scores, or write a clean TREC run.\n- **Fusion:** Use `python -m pyserini.fusion` for `rrf`, `interpolation`, `average`, and `normalize` on valid TREC runs with compatible query\u002Fdocument identifiers.\n- **Reproducibility matrices:** Prefer `--display-commands --dry-run` first; treat full two-click commands as opt-in because they may download indexes, models, topics, qrels, or runs.\n\n## Default Safe Workflow\n\n1. Validate the candidate run before evaluation or fusion:\n\n   ```bash\n   python scripts\u002Fvalidate_trec_run.py run.trec --qrels qrels.txt --strict-ranks --summary\n   ```\n\n2. If the run is MS MARCO format, convert it before using `trec_eval`:\n\n   ```bash\n   python -m pyserini.eval.convert_msmarco_run_to_trec_run --input run.msmarco.tsv --output run.trec\n   ```\n\n3. Evaluate with explicit metrics and a qrels key or qrels path:\n\n   ```bash\n   python -m pyserini.eval.trec_eval -c -m ndcg_cut.10 -m recall.100 qrels.txt run.trec\n   ```\n\n4. Build a fusion command without executing it:\n\n   ```bash\n   python scripts\u002Ffusion_recipe_builder.py --method rrf --runs bm25.trec dense.trec --output fused.trec --runtag demo --k 100 --depth 1000\n   ```\n\n5. Validate the fused run and evaluate it with the same qrels and metrics as the input runs.\n\n## References\n\n- `references\u002Frun-evaluation.md` covers TREC, MS MARCO, KILT, DPR, QA, `TrecRun`, `Qrels`, qrels aliases, and common metrics.\n- `references\u002Freproducibility.md` explains two-click reproduction modules, dry-run\u002Fdisplay-command safety, nondeterminism, and matrix triage.\n- `references\u002Ftroubleshooting.md` maps malformed runs, run\u002Fqrels mismatches, missing eval jars, unjudged documents, and expensive reproduction failures to fixes.\n- `scripts\u002Fvalidate_trec_run.py --help` validates TREC runs and qrels without importing Pyserini or starting Java.\n- `scripts\u002Ffusion_recipe_builder.py --help` generates reproducible Pyserini fusion commands with method-specific checks.\n\n## Acceptance Checklist\n\n- Confirm whether the run is six-column TREC or three-column MS MARCO before choosing an evaluator.\n- Keep query IDs and document IDs as strings; do not coerce them to integers unless a specific evaluator requires it.\n- Use the same qrels\u002Ftopic split and metric definitions for baseline, candidate, and fused runs.\n- Validate every input run before fusion and validate the fused output before evaluation.\n- Treat prebuilt-index reproduction and neural retrieval commands as network\u002Fmodel\u002Fcache-sensitive unless the user has explicitly approved them.\n",{"name":5,"license":23,"metadata":24,"description":7,"disable-model-invocation":26},"Apache 2.0",{"disco-role":25},"operating",true,"main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","skills\u002Frepositories\u002Frepo-skills\u002Fpyserini\u002Fsub-skills\u002Fevaluation-and-fusion",27,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab","2026-09-03 10:25:21",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":26},[36,39,42],{"tool":37,"command":38},"npx","npx skills add VectorSpaceLab\u002FAREX-Skill --skill evaluation-and-fusion",{"tool":40,"command":41},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill evaluation-and-fusion",{"tool":43,"command":44},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill evaluation-and-fusion","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fpyserini\u002Fsub-skills\u002Fevaluation-and-fusion\u002FSKILL.md",[47,53,57,61,67],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"references\u002Freproducibility.md","REFERENCE","参考文档",5223,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fpyserini\u002Fsub-skills\u002Fevaluation-and-fusion\u002Freferences\u002Freproducibility.md",{"path":54,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":55,"sourceHtmlUrl":56},"references\u002Frun-evaluation.md",5841,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fpyserini\u002Fsub-skills\u002Fevaluation-and-fusion\u002Freferences\u002Frun-evaluation.md",{"path":58,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":59,"sourceHtmlUrl":60},"references\u002Ftroubleshooting.md",5198,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fpyserini\u002Fsub-skills\u002Fevaluation-and-fusion\u002Freferences\u002Ftroubleshooting.md",{"path":62,"resourceType":63,"resourceTypeDesc":64,"sizeBytes":65,"sourceHtmlUrl":66},"scripts\u002Ffusion_recipe_builder.py","SCRIPT","脚本",3750,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fpyserini\u002Fsub-skills\u002Fevaluation-and-fusion\u002Fscripts\u002Ffusion_recipe_builder.py",{"path":68,"resourceType":63,"resourceTypeDesc":64,"sizeBytes":69,"sourceHtmlUrl":70},"scripts\u002Fvalidate_trec_run.py",8309,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fpyserini\u002Fsub-skills\u002Fevaluation-and-fusion\u002Fscripts\u002Fvalidate_trec_run.py",[72,79,86,93,100,107],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":18},"01M799NYSH1R3RVAHSM2D88Y84","passnet-triton-opt","Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block\u002Fgrid\u002Fwarp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-triton-opt\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":18},"01M799NS5APEC0N1FZ7XZVW6BN","passnet-solve","END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-solve\u002FSKILL.md",[],[],{"skillNo":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":18},"01M799NNY9BKSM21V5R7QVTBJX","passnet-skill","PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under .\u002Fpass_dir\u002F, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-skill\u002FSKILL.md",[],[],{"skillNo":94,"name":95,"nameZh":6,"description":96,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":97,"categories":98,"keywords":99,"publishedAt":18},"01M799NHH6NP7Z1RK77KGXAN6S","passnet-pattern-fusion","Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()\u002Freplacement_args()\u002Freplacement_func() files, when a pass \"failed to match\", or when deciding how to fuse multiple kernels\u002Fops.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-pattern-fusion\u002FSKILL.md",[],[],{"skillNo":101,"name":102,"nameZh":6,"description":103,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":104,"categories":105,"keywords":106,"publishedAt":18},"01M799ND2J4WNC2PGAZ0A06NDF","passnet-orchestrate","Round\u002Fbatch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep\u002Frevert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-orchestrate\u002FSKILL.md",[],[],{"skillNo":108,"name":109,"nameZh":6,"description":110,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":111,"categories":112,"keywords":113,"publishedAt":18},"01M799N8SBTMY02ZDG2AZF5ZWE","passnet-feedback","Fast, deterministic feedback for PassNet work: pre-flight pattern\u002Fmatch verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-feedback\u002FSKILL.md",[],[]]