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.
☆165Jan 20, 2026Updated 6 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☆457Apr 22, 2026Updated 3 months ago
- ☆95Apr 18, 2026Updated 3 months ago
- Repository for "Training Language Models To Explain Their Own Computations"☆32Jul 7, 2026Updated last month
- Inference API for many LLMs and other useful tools for empirical research☆136May 29, 2026Updated 2 months ago
- Training LLMs to Report Their Learned Behaviors☆28Apr 28, 2026Updated 3 months ago
- Managed hosting for WordPress and PHP on Cloudways • AdManaged hosting for WordPress, Magento, Laravel, or PHP apps, on multiple cloud providers. Deploy in minutes on Cloudways by DigitalOcean.
- Extract residual-stream activations and apply steering vectors (including activation oracles) to any vLLM model during inference.☆120Updated this week
- Code repo for the model organisms and convergent directions of EM papers.☆78Sep 22, 2025Updated 10 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 6 months ago
- ☆55Jul 4, 2025Updated last year
- bloom - evaluate any behavior immediately 🌸🌱☆1,387May 7, 2026Updated 3 months ago
- A toolkit that provides a range of model diffing techniques including a UI to visualize them interactively.☆79Jul 20, 2026Updated 3 weeks ago
- Code repository for "Eliciting Secret Knowledge from Language Models"☆24Mar 30, 2026Updated 4 months ago
- Unified access to Large Language Model modules using NNsight☆119Jul 28, 2026Updated 2 weeks ago
- ☆323Jan 12, 2026Updated 7 months ago
- Managed hosting for WordPress and PHP on Cloudways • AdManaged hosting for WordPress, Magento, Laravel, or PHP apps, on multiple cloud providers. Deploy in minutes on Cloudways by DigitalOcean.
- Repo for measuring whether using AI tools inhibits skill formation and development☆15Jan 3, 2026Updated 7 months ago
- ☆25Mar 30, 2026Updated 4 months ago
- ☆16Nov 14, 2025Updated 8 months ago
- [COLM 2026] Official implementation for "MonitorBench: A Comprehensive Benchmark for Chain-of-Thought Monitorability in Large Language Mo…☆20Apr 23, 2026Updated 3 months ago
- Code for Negation Neglect☆16May 22, 2026Updated 2 months ago
- A library for training crosscoders☆18May 28, 2025Updated last year
- Agent observability and replay tooling for AI safety & interpretability research.☆112Jun 19, 2026Updated last month
- ☆27Sep 5, 2024Updated last year
- Open-sourced evaluation suite from the Monitoring Monitorability paper☆89Jun 11, 2026Updated 2 months ago
- Managed hosting for WordPress and PHP on Cloudways • AdManaged hosting for WordPress, Magento, Laravel, or PHP apps, on multiple cloud providers. Deploy in minutes on Cloudways by DigitalOcean.
- An alignment auditing agent capable of quickly exploring alignment hypothesis☆1,287Updated this week
- Parameter Decomposition☆139Updated this week
- Datasets used in the paper "Reward hacking behavior can generalize across tasks"☆16Aug 17, 2025Updated 11 months ago
- ⚓️ Repository for the "Thought Anchors: Which LLM Reasoning Steps Matter?" paper.☆140Oct 27, 2025Updated 9 months ago
- ☆2,896Jul 18, 2026Updated 3 weeks ago
- open source interpretability platform 🧠☆1,101Jul 17, 2026Updated 3 weeks ago
- Repository for the "Chain-of-Thought Reasoning In The Wild Is Not Always Faithful" paper☆35Mar 31, 2026Updated 4 months ago
- ☆23Jan 30, 2026Updated 6 months ago
- ☆18Jul 9, 2025Updated last year
- Managed Kubernetes at scale on DigitalOcean • AdDigitalOcean Kubernetes includes the control plane, bandwidth allowance, container registry, automatic updates, and more for free.
- Stanford NLP Python library for benchmarking the utility of LLM interpretability methods☆211Mar 12, 2026Updated 5 months ago
- ☆114Aug 8, 2024Updated 2 years ago
- Code to generate figures of paper "When do spectral gradient updates help in deep learning?"☆16Dec 3, 2025Updated 8 months ago
- ☆155Feb 10, 2026Updated 6 months ago
- ☆1,224Jul 24, 2026Updated 2 weeks ago
- ☆86Feb 25, 2025Updated last year
- Open Character Training☆97Apr 4, 2026Updated 4 months ago