The Assistant Axis is a direction in activation space that captures how "Assistant-like" a model's behavior is. Models can drift away from the Assistant during conversations—sometimes toward bizarre or harmful personas. This repo contains a pipeline for generating the Assistant Axis and notebooks for monitoring and steering with it.
☆182Jan 20, 2026Updated 9 months ago
Alternatives and similar repositories for assistant-axis
Users that are interested in assistant-axis are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.
Sorting:
- Persona Vectors: Monitoring and Controlling Character Traits in Language Models☆474Apr 22, 2026Updated 6 months ago
- ☆14Feb 12, 2026Updated 8 months ago
- Training LLMs to Report Their Learned Behaviors☆31Apr 28, 2026Updated 6 months ago
- ☆19Jun 3, 2026Updated 5 months ago
- Zero-dependency local web app for browsing Claude Code session transcripts☆15Oct 27, 2026Updated last week
- Virtual machines for every use case on DigitalOcean • AdGet dependable uptime with 99.99% SLA, simple security tools, and predictable monthly pricing with DigitalOcean's virtual machines, called Droplets.
- Official Code for our paper: "Language Models Learn to Mislead Humans via RLHF""☆20Oct 11, 2024Updated 2 years ago
- ☆67Jul 4, 2025Updated last year
- ☆104Sep 26, 2026Updated last month
- Benchmark dataset for evaluating trusted monitors on AI agent transcripts☆19May 16, 2026Updated 5 months ago
- Framework for generating behavioral evaluations of frontier AI models.☆45Updated this week
- Inference API for many LLMs and other useful tools for empirical research☆142May 29, 2026Updated 5 months ago
- This was designed for interp researchers who want to do research on or with interp agents to give quality of life improvements and fix …☆146Feb 8, 2026Updated 8 months ago
- bloom - evaluate any behavior immediately 🌸🌱☆1,426May 7, 2026Updated 5 months ago
- Parameter Decomposition☆153Oct 20, 2026Updated 2 weeks ago
- Deploy to Railway using AI coding agents - Free Credits Offer • AdUse Claude Code, Codex, OpenCode, and more. Autonomous software development now has the infrastructure to match with Railway.
- Dataset of hackable TerminalBench-style tasks and exploit trajectories☆50Sep 30, 2026Updated last month
- ☆16Oct 13, 2025Updated last year
- Official code for the ICML 2026 paper "Modular Pretraining Enables Access Control" (GRAM).☆44Jul 8, 2026Updated 3 months ago
- Code repo for the model organisms and convergent directions of EM papers.☆90Sep 22, 2025Updated last year
- 🌳 Repository for the "Thought Branches: Interpreting LLM Reasoning Requires Resampling" paper.☆14Nov 5, 2025Updated 11 months ago
- Training-free feature discovery for LLM activations via streaming exemplar partitioning.☆14Sep 17, 2026Updated last month
- A toolkit that provides a range of model diffing techniques including a UI to visualize them interactively.☆88Oct 25, 2026Updated last week
- Code repository for "Eliciting Secret Knowledge from Language Models"☆24Mar 30, 2026Updated 7 months ago
- An alignment auditing agent capable of quickly exploring alignment hypothesis☆1,397Updated this week
- Wordpress hosting with auto-scaling - Free Trial Offer • AdFully Managed hosting for WordPress and WooCommerce businesses that need reliable, auto-scalable performance. Cloudways SafeUpdates now available.
- Extract residual-stream activations and apply steering vectors (including activation oracles) to any vLLM model during inference.☆140Oct 13, 2026Updated 3 weeks ago
- ☆343Jan 12, 2026Updated 9 months ago
- Repository for the Feature Alignment to Descriptions Evaluation (FADE) Paper☆11Feb 10, 2026Updated 8 months ago
- 🤖🪝Interpretability with tensordict and torch hooks.☆13Sep 8, 2026Updated last month
- ☆164Feb 10, 2026Updated 8 months ago
- ☆34Oct 8, 2026Updated 3 weeks ago
- James' cookbook of evaluations and finetuning experiments☆36Feb 19, 2026Updated 8 months ago
- Training Transformers with knowledge localization (SGTM)☆57Jan 11, 2026Updated 9 months ago
- Attribution-based Parameter Decomposition☆35Jun 11, 2025Updated last year
- 1-Click AI Models by DigitalOcean Gradient • AdDeploy popular AI models on DigitalOcean Gradient GPU virtual machines with just a single click. Zero configuration with optimized deployments.
- Code and results on finding one-shot steering vectors that mediate emergent misalignment☆12Apr 12, 2025Updated last year
- open source interpretability platform 🧠☆1,198Updated this week
- Companion code for the global workspace interpretability paper☆2,100Oct 10, 2026Updated 3 weeks ago
- ☆15Jul 8, 2025Updated last year
- Agent observability and replay tooling for AI safety & interpretability research.☆117Jun 19, 2026Updated 4 months ago
- ☆89Feb 25, 2025Updated last year
- Open Character Training☆106Apr 4, 2026Updated 6 months ago