davila7 avatar

Pyvene Interventions

This brings Stanford NLP's pyvene library to Claude, letting you run causal interventions on PyTorch models without writing boilerplate. If you're doing activation patching, causal tracing (ROME-style), or interchange intervention training, this handles the setup through a dict-based config system. The main win is reproducibility: experiments become shareable artifacts you can push to HuggingFace instead of scattered notebooks. It works with any PyTorch architecture, not just transformers, which matters if you're testing causal hypotheses beyond language models. The library has 840+ stars and a NAACL 2024 paper backing it, so the abstractions are research-tested rather than someone's weekend project.

skills.sh installs
287
GitHub stars
33.2K
last updated
Aug 21, 2026

Install

$ npx -y skills add davila7/claude-code-templates --skill pyvene-interventions --agent claude-code

Installs into .claude/skills of the current project.

Files

This brings Stanford NLP's pyvene library to Claude, letting you run causal interventions on PyTorch models without writing boilerplate. If you're doing activation patching, causal tracing (ROME-style), or interchange intervention training, this handles the setup through a dict-based config system. The main win is reproducibility: experiments become shareable artifacts you can push to HuggingFace instead of scattered notebooks. It works with any PyTorch architecture, not just transformers, which matters if you're testing causal hypotheses beyond language models. The library has 840+ stars and a NAACL 2024 paper backing it, so the abstractions are research-tested rather than someone's weekend project.

Browse the repository on GitHub

Sponsors

Details

Repository
github.com/davila7/claude-code-templates
Categories
Testing & QA, Data Science & ML
First seen
Jun 3, 2026

More from davila7/claude-code-templates

All 618
1h1d avataresp32-serial-commandsh1d/agent-skills-esp32Send commands to ESP32 and microcontrollers via serial port to emulate button presses and user actions. Use for automated testing, stress testing, and debugging287skills.sh installs
2gramiojs avatargramiogramiojs/documentationInvoke for ANY Telegram bot code — `gramio`/`@gramio/*` imports, `new Bot()`, `bot.command`/`bot.callbackQuery`/`bot.on`/`bot.updates.on`, scenes/FSM, inline & 286skills.sh installs
3feiskyer avatarautonomous-skillfeiskyer/claude-code-settingsExecute long-running, multi-session tasks autonomously using Claude Code headless mode or in-session hook-based loops. Supports structured task decomposition (f284skills.sh installs
4postplusai avataredit-enhancement-packagerpostplusai/postplus-skillsPackage beat-level edit enhancement instructions from a B-roll plan, subtitle chunks, and optional style context. Use this when the goal is to turn B-roll match282skills.sh installs
5patricio0312rev avatarquality-gates-enforcerpatricio0312rev/skillsEnforces minimum quality thresholds in CI including code coverage, linting, type checking, and security scanning. Provides required checks, PR rules, and automa280skills.sh installs
6ulpi-io avatarbrowseulpi-io/skillsDrive a REAL browser and native apps the way a user would — not fetch raw HTML — through the persistent `browse` CLI: one long-lived headless Chromium (plus iOS280skills.sh installs