activation-steering / README.md
apehex's picture
Use the psai-ops boilerplate with the matching app
8a582a2

A newer version of the Gradio SDK is available: 6.1.0

Upgrade
metadata
pinned: false
title: Contrastive Steering
emoji: 🎯
colorFrom: yellow
colorTo: red
sdk: gradio
sdk_version: 5.47.2
app_file: app.py
app_build_command: pip install -r requirements.txt
license: agpl-3.0
short_description: Steer the model outputs with an activation delta.
models:
  - openai/gpt-oss-20b

psAI-Ops: Contrastive Steering apehex logo

Compute the difference in activation between two prompts and use it to steer the model outputs.

Licensed under the aGPLv3.