[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78S3WM2AC3Z0P3FMKN28ZPY":3,"skill-same:01M78S3WM2AC3Z0P3FMKN28ZPY":70},{"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":26,"commitSha":27,"skillDirectory":28,"forksCount":29,"repositoryHtmlUrl":30,"repositoryHomepageUrl":6,"ownerHtmlUrl":31,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":32,"install":33,"resources":45},"01M78S3WM2AC3Z0P3FMKN28ZPY","evaluation-and-optimization",null,"Guide AdalFlow evaluation metrics, datasets, parameters, gradients, prompt\u002Ffew-shot\u002Ftext-grad optimization, AdalComponent, Trainer, and optimize_anything workflows.","VectorSpaceLab\u002FAREX-Skill","VectorSpaceLab","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F180637102?v=4","01M7538C5WPSJGWF7KHR4EFZ9F","01M7538C6R2GFP53NPXTCZNBRT",351,"MIT","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fadalflow\u002Fsub-skills\u002Fevaluation-and-optimization\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: evaluation-and-optimization\ndescription: \"Guide AdalFlow evaluation metrics, datasets, parameters,\n  gradients, prompt\u002Ffew-shot\u002Ftext-grad optimization, AdalComponent, Trainer, and\n  optimize_anything workflows.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\nlicense: MIT\n---\n\n# AdalFlow evaluation and optimization\n\nUse this sub-skill when the user wants to score task outputs, select or load AdalFlow benchmark datasets, make `Parameter` objects trainable, wrap a task in `AdalComponent`, run `Trainer.diagnose` or `Trainer.fit`, configure few-shot\u002Ftext-gradient optimizers, or use `optimize_anything` for bounded text-artifact search.\n\n## Route boundaries\n\nStay here for:\n\n- `AnswerMatchAcc`, `RetrieverEvaluator`, and cautious `LLMasJudge` evaluation.\n- Dataset loaders for GSM8K, TREC, HotpotQA, and BBH-style tasks, including small-size and skip-expensive guidance.\n- `Parameter`, `ParameterType`, `GradComponent`, `EvalFnToTextLoss`, `LLMAsTextLoss`, `BootstrapFewShot`, `TGDOptimizer`, `AdalComponent`, `Trainer`, and `optimize_anything` workflows.\n\nRoute away instead of duplicating guidance:\n\n- RAG construction, document splitting, local\u002Fvector retrievers, and index pipelines: use `retrieval-rag-and-data-pipelines`.\n- Provider model-client configuration, credentials, `Generator`, and fake provider clients: use `model-client-and-generator-workflows`.\n- Tracing\u002Flogging artifacts and callback storage: use `tracing-observability-and-configuration`.\n\n## Reference map\n\n- [Evaluation metrics](references\u002Fevaluation-metrics.md): metric APIs, input shapes, output objects, and no-network smoke examples.\n- [Optimization and training](references\u002Foptimization-and-training.md): `Parameter`\u002Fgradient concepts, `AdalComponent` contracts, `Trainer.fit`\u002F`diagnose`, few-shot\u002Ftext-gradient optimizer setup, and `optimize_anything`.\n- [Datasets and benchmarks](references\u002Fdatasets-and-benchmarks.md): loader classes, fields, optional dependencies, downloads, and benchmark constraints.\n- [Troubleshooting](references\u002Ftroubleshooting.md): evaluator shape errors, metric extras, dataset misses, checkpoint\u002Fresume, validation-set requirements, provider cost, runtime limits, and multiprocessing issues.\n- [Metric smoke script](scripts\u002Fevaluation_metrics_smoke.py): service-free exact\u002Ffuzzy\u002FF1\u002Fretriever metric check.\n\n## Safe operating defaults\n\n1. **Score before optimizing.** Build a deterministic metric with `AnswerMatchAcc` or `RetrieverEvaluator` and run a tiny local smoke case before any model-backed judge or optimizer.\n2. **Treat live judging and text gradients as expensive.** `LLMasJudge`, `LLMAsTextLoss`, backward engines, teacher generators, and `TGDOptimizer` can invoke model providers. Require explicit provider setup, credentials, budget, sample limits, and cache\u002Fcheckpoint choices.\n3. **Use bounded data.** Dataset loaders may download external data and optional packages. Prefer `size=...` or pre-existing local data for first tests.\n4. **Make `AdalComponent` contracts explicit.** `prepare_task` returns `(callable, kwargs)` for one sample; `prepare_eval` returns `(eval_fn, kwargs)` for inference scoring; `prepare_loss` returns `(loss_fn, kwargs)` with `Parameter` inputs and correct `eval_input` values.\n5. **Prefer diagnosis before training.** `Trainer.diagnose` is useful for identifying failing samples and prompt\u002Foutput issues. Use `num_workers=1` for minimal debugging, then raise concurrency only after stable.\n6. **Do not imply full benchmark training is safe.** Benchmark scripts and full prompt optimization runs are optional, network\u002Fprovider\u002Fdataset dependent, and should be reduced to a tiny user-approved run before scaling.\n\n## Minimal decision checklist\n\n- What is being scored: final answer strings, structured labels, retrieved contexts\u002FIDs, or model-judged quality?\n- Is the metric service-free, optional-package-backed, or provider-backed?\n- Are predictions raw values, `GeneratorOutput.data`, or `Parameter.full_response.data`?\n- Are trainable values marked with the right `ParameterType` (`PROMPT` for instructions, `DEMOS` for examples)?\n- Does the training data provide stable `id` fields and a real validation set?\n- What are the maximum samples, optimizer steps, worker count, and checkpoint\u002Fresume plan?\n","# AdalFlow evaluation and optimization\n\nUse this sub-skill when the user wants to score task outputs, select or load AdalFlow benchmark datasets, make `Parameter` objects trainable, wrap a task in `AdalComponent`, run `Trainer.diagnose` or `Trainer.fit`, configure few-shot\u002Ftext-gradient optimizers, or use `optimize_anything` for bounded text-artifact search.\n\n## Route boundaries\n\nStay here for:\n\n- `AnswerMatchAcc`, `RetrieverEvaluator`, and cautious `LLMasJudge` evaluation.\n- Dataset loaders for GSM8K, TREC, HotpotQA, and BBH-style tasks, including small-size and skip-expensive guidance.\n- `Parameter`, `ParameterType`, `GradComponent`, `EvalFnToTextLoss`, `LLMAsTextLoss`, `BootstrapFewShot`, `TGDOptimizer`, `AdalComponent`, `Trainer`, and `optimize_anything` workflows.\n\nRoute away instead of duplicating guidance:\n\n- RAG construction, document splitting, local\u002Fvector retrievers, and index pipelines: use `retrieval-rag-and-data-pipelines`.\n- Provider model-client configuration, credentials, `Generator`, and fake provider clients: use `model-client-and-generator-workflows`.\n- Tracing\u002Flogging artifacts and callback storage: use `tracing-observability-and-configuration`.\n\n## Reference map\n\n- [Evaluation metrics](references\u002Fevaluation-metrics.md): metric APIs, input shapes, output objects, and no-network smoke examples.\n- [Optimization and training](references\u002Foptimization-and-training.md): `Parameter`\u002Fgradient concepts, `AdalComponent` contracts, `Trainer.fit`\u002F`diagnose`, few-shot\u002Ftext-gradient optimizer setup, and `optimize_anything`.\n- [Datasets and benchmarks](references\u002Fdatasets-and-benchmarks.md): loader classes, fields, optional dependencies, downloads, and benchmark constraints.\n- [Troubleshooting](references\u002Ftroubleshooting.md): evaluator shape errors, metric extras, dataset misses, checkpoint\u002Fresume, validation-set requirements, provider cost, runtime limits, and multiprocessing issues.\n- [Metric smoke script](scripts\u002Fevaluation_metrics_smoke.py): service-free exact\u002Ffuzzy\u002FF1\u002Fretriever metric check.\n\n## Safe operating defaults\n\n1. **Score before optimizing.** Build a deterministic metric with `AnswerMatchAcc` or `RetrieverEvaluator` and run a tiny local smoke case before any model-backed judge or optimizer.\n2. **Treat live judging and text gradients as expensive.** `LLMasJudge`, `LLMAsTextLoss`, backward engines, teacher generators, and `TGDOptimizer` can invoke model providers. Require explicit provider setup, credentials, budget, sample limits, and cache\u002Fcheckpoint choices.\n3. **Use bounded data.** Dataset loaders may download external data and optional packages. Prefer `size=...` or pre-existing local data for first tests.\n4. **Make `AdalComponent` contracts explicit.** `prepare_task` returns `(callable, kwargs)` for one sample; `prepare_eval` returns `(eval_fn, kwargs)` for inference scoring; `prepare_loss` returns `(loss_fn, kwargs)` with `Parameter` inputs and correct `eval_input` values.\n5. **Prefer diagnosis before training.** `Trainer.diagnose` is useful for identifying failing samples and prompt\u002Foutput issues. Use `num_workers=1` for minimal debugging, then raise concurrency only after stable.\n6. **Do not imply full benchmark training is safe.** Benchmark scripts and full prompt optimization runs are optional, network\u002Fprovider\u002Fdataset dependent, and should be reduced to a tiny user-approved run before scaling.\n\n## Minimal decision checklist\n\n- What is being scored: final answer strings, structured labels, retrieved contexts\u002FIDs, or model-judged quality?\n- Is the metric service-free, optional-package-backed, or provider-backed?\n- Are predictions raw values, `GeneratorOutput.data`, or `Parameter.full_response.data`?\n- Are trainable values marked with the right `ParameterType` (`PROMPT` for instructions, `DEMOS` for examples)?\n- Does the training data provide stable `id` fields and a real validation set?\n- What are the maximum samples, optimizer steps, worker count, and checkpoint\u002Fresume plan?\n",{"name":5,"license":14,"metadata":23,"description":7,"disable-model-invocation":25},{"disco-role":24},"operating",true,"main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","skills\u002Frepositories\u002Frepo-skills\u002Fadalflow\u002Fsub-skills\u002Fevaluation-and-optimization",27,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab","2026-09-03 10:25:21",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":25},[35,38,41],{"tool":36,"command":37},"npx","npx skills add VectorSpaceLab\u002FAREX-Skill --skill evaluation-and-optimization",{"tool":39,"command":40},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill evaluation-and-optimization",{"tool":42,"command":43},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill evaluation-and-optimization","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fadalflow\u002Fsub-skills\u002Fevaluation-and-optimization\u002FSKILL.md",[46,52,56,60,64],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"references\u002Fdatasets-and-benchmarks.md","REFERENCE","参考文档",6324,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fadalflow\u002Fsub-skills\u002Fevaluation-and-optimization\u002Freferences\u002Fdatasets-and-benchmarks.md",{"path":53,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":54,"sourceHtmlUrl":55},"references\u002Fevaluation-metrics.md",6866,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fadalflow\u002Fsub-skills\u002Fevaluation-and-optimization\u002Freferences\u002Fevaluation-metrics.md",{"path":57,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":58,"sourceHtmlUrl":59},"references\u002Foptimization-and-training.md",11789,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fadalflow\u002Fsub-skills\u002Fevaluation-and-optimization\u002Freferences\u002Foptimization-and-training.md",{"path":61,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":62,"sourceHtmlUrl":63},"references\u002Ftroubleshooting.md",8857,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fadalflow\u002Fsub-skills\u002Fevaluation-and-optimization\u002Freferences\u002Ftroubleshooting.md",{"path":65,"resourceType":66,"resourceTypeDesc":67,"sizeBytes":68,"sourceHtmlUrl":69},"scripts\u002Fevaluation_metrics_smoke.py","SCRIPT","脚本",4098,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fadalflow\u002Fsub-skills\u002Fevaluation-and-optimization\u002Fscripts\u002Fevaluation_metrics_smoke.py",[71,79,86,93,100,107],{"skillNo":72,"name":73,"nameZh":6,"description":74,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":75,"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.","Apache-2.0","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":75,"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":75,"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":75,"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":75,"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":75,"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",[],[]]