[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78T0J3A7A5248ZWF4SCAXPF":3,"skill-same:01M78T0J3A7A5248ZWF4SCAXPF":66},{"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},"01M78T0J3A7A5248ZWF4SCAXPF","differentiation-core",null,"Routes scalar and array differentiation, higher-order derivatives, and gradient checking for Autograd's core AD operators.","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\u002Fautograd\u002Fsub-skills\u002Fdifferentiation-core\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: differentiation-core\ndescription: \"Routes scalar and array differentiation, higher-order derivatives,\n  and gradient checking for Autograd's core AD operators.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\nlicense: MIT\n---\n\n# Differentiation Core\n\nUse this sub-skill when a task is about Autograd's core differentiation operators: scalar gradients, Jacobians, directional derivatives, Hessians, VJP\u002FJVP pullbacks, generalized Gauss-Newton products, complex holomorphic derivatives, checkpointing, or numerical gradient checks.\n\n## Trigger examples\n\n- \"I need `grad` or `value_and_grad` for a scalar loss.\"\n- \"I need `jacobian` or `elementwise_grad` for a vector-valued output.\"\n- \"I need a Hessian-vector product, `make_vjp`, or `make_jvp`.\"\n- \"I want to verify a derivative with `check_grads`.\"\n- \"My function has `if`\u002F`while` branches and I want to know whether Autograd can still differentiate it.\"\n\n## Read first\n\n- `references\u002Fapi-reference.md` for operator selection, return conventions, and mode relationships.\n- `references\u002Fworkflows.md` for scalar-loss, vector-output, higher-order, complex, and gradient-checking recipes.\n- `references\u002Ftroubleshooting.md` for scalar-output errors, argnum confusion, shape mismatches, holomorphic warnings, and checkpoint tradeoffs.\n- `scripts\u002Fdifferentiation_smoke.py` for a runnable smoke that exercises the core operators without plotting.\n\n## Boundary routing\n\n- Wrapper behavior, NumPy\u002FSciPy support, `A.dot(B)` \u002F in-place \u002F container pitfalls, and optional dependency questions belong in `..\u002Fnumpy-scipy-primitives\u002FSKILL.md`.\n- `primitive`, `defvjp`, `defjvp`, and custom VJP\u002FJVP mechanics belong in `..\u002Fextend-primitives\u002FSKILL.md`.\n- Optimizer wiring, flattening, and `fixed_point` belong in `..\u002Foptimization-workflows\u002FSKILL.md`.\n\n## How to choose an operator\n\n- Use reverse mode when the output is a real scalar and the input is large: `grad`, `value_and_grad`, or `make_vjp`.\n- Use forward mode when the input is scalar or you want one directional derivative: `deriv` or `make_jvp`.\n- Use `jacobian` when you need the full output-to-input sensitivity tensor.\n- Use `elementwise_grad` when the function is real-valued and elementwise so the Jacobian is diagonal or column-summed.\n- Use `hessian`, `hessian_tensor_product`, `hessian_vector_product`, or `make_ggnvp` when you need curvature without writing the matrix explicitly.\n- Use `holomorphic_grad` only for complex-holomorphic functions.\n- Use `checkpoint` when the function is deep and memory is the bottleneck.\n- Use `grad_and_aux` when the function returns a scalar loss plus extra side information.\n\n## Core rules\n\n- `grad` and `value_and_grad` require a real scalar output.\n- `jacobian` works on scalar or array outputs and returns a tensor with output shape first and input shape last.\n- `elementwise_grad` is not a general Jacobian helper; it returns the column-sum\u002Fdiagonal behavior used for elementwise real functions.\n- Control flow is transparent to Autograd: the executed Python path matters, not the unexecuted branches.\n- `make_vjp` exposes the reverse-mode pullback; `make_jvp` exposes the forward-mode pushforward.\n- Higher-order derivatives are usually formed by composing these operators.\n- `grad_and_aux` differentiates only the first output and returns the auxiliary payload unchanged.\n- `checkpoint` recomputes forward intermediates during the backward pass to trade time for memory.\n\n## Typical route\n\n1. Start with `grad` for a scalar loss.\n2. If the output is vector-valued, switch to `jacobian` or `elementwise_grad`.\n3. If you need both the value and the gradient, use `value_and_grad`.\n4. If you need a directional derivative, use `deriv` or `make_jvp`.\n5. If you need a pullback from output cotangents, use `make_vjp`.\n6. If you need curvature information, use `hessian` or a Hessian-vector product helper.\n7. If the function is complex-valued and holomorphic, use `holomorphic_grad`.\n8. If the function is deep and memory-bound, wrap it with `checkpoint`.\n9. Validate the chosen operator with `check_grads` before trusting a new workflow.\n\n## Reliability notes\n\n- Use `check_grads` for a single function and `combo_check` when you want the same assertion to cover several positional or keyword combinations.\n- When a derivative shape looks wrong, inspect the primal input and output shapes before changing the operator.\n- When the function is intentionally branchy, choose a test point away from the branch threshold.\n- When the failure message names a missing VJP or JVP rule for a primitive, route to `..\u002Fextend-primitives\u002FSKILL.md` instead of trying to solve it here.\n\n## What this sub-skill does not cover\n\n- Built-in NumPy\u002FSciPy wrapper behavior and compatibility traps live in `..\u002Fnumpy-scipy-primitives\u002FSKILL.md`.\n- Defining new primitives or VJP\u002FJVP rules lives in `..\u002Fextend-primitives\u002FSKILL.md`.\n- Optimizer flattening and structured parameter updates live in `..\u002Foptimization-workflows\u002FSKILL.md`.\n\n## Fastest sanity check\n\nRun `scripts\u002Fdifferentiation_smoke.py` when you need a quick end-to-end check that the bundled differentiation guidance still matches the installed package.\n","# Differentiation Core\n\nUse this sub-skill when a task is about Autograd's core differentiation operators: scalar gradients, Jacobians, directional derivatives, Hessians, VJP\u002FJVP pullbacks, generalized Gauss-Newton products, complex holomorphic derivatives, checkpointing, or numerical gradient checks.\n\n## Trigger examples\n\n- \"I need `grad` or `value_and_grad` for a scalar loss.\"\n- \"I need `jacobian` or `elementwise_grad` for a vector-valued output.\"\n- \"I need a Hessian-vector product, `make_vjp`, or `make_jvp`.\"\n- \"I want to verify a derivative with `check_grads`.\"\n- \"My function has `if`\u002F`while` branches and I want to know whether Autograd can still differentiate it.\"\n\n## Read first\n\n- `references\u002Fapi-reference.md` for operator selection, return conventions, and mode relationships.\n- `references\u002Fworkflows.md` for scalar-loss, vector-output, higher-order, complex, and gradient-checking recipes.\n- `references\u002Ftroubleshooting.md` for scalar-output errors, argnum confusion, shape mismatches, holomorphic warnings, and checkpoint tradeoffs.\n- `scripts\u002Fdifferentiation_smoke.py` for a runnable smoke that exercises the core operators without plotting.\n\n## Boundary routing\n\n- Wrapper behavior, NumPy\u002FSciPy support, `A.dot(B)` \u002F in-place \u002F container pitfalls, and optional dependency questions belong in `..\u002Fnumpy-scipy-primitives\u002FSKILL.md`.\n- `primitive`, `defvjp`, `defjvp`, and custom VJP\u002FJVP mechanics belong in `..\u002Fextend-primitives\u002FSKILL.md`.\n- Optimizer wiring, flattening, and `fixed_point` belong in `..\u002Foptimization-workflows\u002FSKILL.md`.\n\n## How to choose an operator\n\n- Use reverse mode when the output is a real scalar and the input is large: `grad`, `value_and_grad`, or `make_vjp`.\n- Use forward mode when the input is scalar or you want one directional derivative: `deriv` or `make_jvp`.\n- Use `jacobian` when you need the full output-to-input sensitivity tensor.\n- Use `elementwise_grad` when the function is real-valued and elementwise so the Jacobian is diagonal or column-summed.\n- Use `hessian`, `hessian_tensor_product`, `hessian_vector_product`, or `make_ggnvp` when you need curvature without writing the matrix explicitly.\n- Use `holomorphic_grad` only for complex-holomorphic functions.\n- Use `checkpoint` when the function is deep and memory is the bottleneck.\n- Use `grad_and_aux` when the function returns a scalar loss plus extra side information.\n\n## Core rules\n\n- `grad` and `value_and_grad` require a real scalar output.\n- `jacobian` works on scalar or array outputs and returns a tensor with output shape first and input shape last.\n- `elementwise_grad` is not a general Jacobian helper; it returns the column-sum\u002Fdiagonal behavior used for elementwise real functions.\n- Control flow is transparent to Autograd: the executed Python path matters, not the unexecuted branches.\n- `make_vjp` exposes the reverse-mode pullback; `make_jvp` exposes the forward-mode pushforward.\n- Higher-order derivatives are usually formed by composing these operators.\n- `grad_and_aux` differentiates only the first output and returns the auxiliary payload unchanged.\n- `checkpoint` recomputes forward intermediates during the backward pass to trade time for memory.\n\n## Typical route\n\n1. Start with `grad` for a scalar loss.\n2. If the output is vector-valued, switch to `jacobian` or `elementwise_grad`.\n3. If you need both the value and the gradient, use `value_and_grad`.\n4. If you need a directional derivative, use `deriv` or `make_jvp`.\n5. If you need a pullback from output cotangents, use `make_vjp`.\n6. If you need curvature information, use `hessian` or a Hessian-vector product helper.\n7. If the function is complex-valued and holomorphic, use `holomorphic_grad`.\n8. If the function is deep and memory-bound, wrap it with `checkpoint`.\n9. Validate the chosen operator with `check_grads` before trusting a new workflow.\n\n## Reliability notes\n\n- Use `check_grads` for a single function and `combo_check` when you want the same assertion to cover several positional or keyword combinations.\n- When a derivative shape looks wrong, inspect the primal input and output shapes before changing the operator.\n- When the function is intentionally branchy, choose a test point away from the branch threshold.\n- When the failure message names a missing VJP or JVP rule for a primitive, route to `..\u002Fextend-primitives\u002FSKILL.md` instead of trying to solve it here.\n\n## What this sub-skill does not cover\n\n- Built-in NumPy\u002FSciPy wrapper behavior and compatibility traps live in `..\u002Fnumpy-scipy-primitives\u002FSKILL.md`.\n- Defining new primitives or VJP\u002FJVP rules lives in `..\u002Fextend-primitives\u002FSKILL.md`.\n- Optimizer flattening and structured parameter updates live in `..\u002Foptimization-workflows\u002FSKILL.md`.\n\n## Fastest sanity check\n\nRun `scripts\u002Fdifferentiation_smoke.py` when you need a quick end-to-end check that the bundled differentiation guidance still matches the installed package.\n",{"name":5,"license":14,"metadata":23,"description":7,"disable-model-invocation":25},{"disco-role":24},"operating",true,"main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","skills\u002Frepositories\u002Frepo-skills\u002Fautograd\u002Fsub-skills\u002Fdifferentiation-core",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 differentiation-core",{"tool":39,"command":40},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill differentiation-core",{"tool":42,"command":43},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill differentiation-core","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fautograd\u002Fsub-skills\u002Fdifferentiation-core\u002FSKILL.md",[46,52,56,60],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"references\u002Fapi-reference.md","REFERENCE","参考文档",5040,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fautograd\u002Fsub-skills\u002Fdifferentiation-core\u002Freferences\u002Fapi-reference.md",{"path":53,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":54,"sourceHtmlUrl":55},"references\u002Ftroubleshooting.md",3987,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fautograd\u002Fsub-skills\u002Fdifferentiation-core\u002Freferences\u002Ftroubleshooting.md",{"path":57,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":58,"sourceHtmlUrl":59},"references\u002Fworkflows.md",7058,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fautograd\u002Fsub-skills\u002Fdifferentiation-core\u002Freferences\u002Fworkflows.md",{"path":61,"resourceType":62,"resourceTypeDesc":63,"sizeBytes":64,"sourceHtmlUrl":65},"scripts\u002Fdifferentiation_smoke.py","SCRIPT","脚本",4227,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fautograd\u002Fsub-skills\u002Fdifferentiation-core\u002Fscripts\u002Fdifferentiation_smoke.py",[67,75,82,89,96,103],{"skillNo":68,"name":69,"nameZh":6,"description":70,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":71,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"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":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":71,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"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":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":71,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"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":90,"name":91,"nameZh":6,"description":92,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":71,"licenseFileUrl":6,"sourceHtmlUrl":93,"categories":94,"keywords":95,"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":97,"name":98,"nameZh":6,"description":99,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":71,"licenseFileUrl":6,"sourceHtmlUrl":100,"categories":101,"keywords":102,"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":104,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":71,"licenseFileUrl":6,"sourceHtmlUrl":107,"categories":108,"keywords":109,"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",[],[]]