vectorspacelab/arex-skill/aggregate-influence-pruning
Claude Code and agent skill by vectorspacelab/arex-skill
- community
Checking that the SKILL.md file is still in the repository.
Select removable data subsets by constraining aggregate influence vector norm instead of ranking independent scalar scores.
Install
npx vskill@latest install vectorspacelab/arex-skill/aggregate-influence-pruningSKILL.md
The same SKILL.md was published earlier as vectorspacelab/arex-skill/aggregate-influence-pruning.
Aggregate Influence Pruning
Use this skill after per-example influence vectors have been estimated. It implements the paper's central subset-selection idea: maximize removable cardinality while the norm of the sum of removed influences remains below epsilon.
Inputs
- Per-example influence vectors with equal dimensionality.
epsilonfor generalization-guaranteed pruning orcardinalityfor fixed-size pruning.- Bounded search mode; exhaustive search is appropriate only for small recovery cases.
Outputs
- Selected removal indices and binary mask.
- Aggregate influence vector and norm.
- Feasibility and objective metadata.
Workflow
- Validate influence vector dimensions and finite values.
- Enumerate candidate masks for tiny recovery runs, or use a deterministic bounded heuristic for larger runs.
- Prefer larger feasible subsets; break ties by lower aggregate norm.
- Save enough metadata to prove joint cancellation was considered.
Validation
Run python scripts/prune_by_influence.py --demo and python tests/test_prune_by_influence.py from this skill directory.
Limitations
The exhaustive implementation is for bounded recovery and testing. Large datasets should use a time-limited heuristic such as simulated annealing while preserving the same objective contract.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/task-oriented/PaperBench/skills/lbcs/skill/generalization_influence_pruning/aggregate_influence_pruning/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/vectorspacelab/arex-skill/aggregate-influence-pruning)