Alternatives and similar repositories for control-arena

Users that are interested in control-arena are comparing it to the libraries listed below

• METR / task-standard
  METR Task Standard
  ☆171Updated 11 months ago
• UKGovernmentBEIS / inspect_evals
  Collection of evals for Inspect AI
  ☆344Updated this week
• safety-research / safety-tooling
  Inference API for many LLMs and other useful tools for empirical research
  ☆101Updated this week
• TransformerLensOrg / CircuitsVis
  Mechanistic Interpretability Visualizations using React
  ☆315Updated last year
• ArthurConmy / Automatic-Circuit-Discovery
  ☆265Updated last year
• TransluceAI / docent
  ☆79Updated last month
• anthropics / sleeper-agents-paper
  Contains random samples referenced in the paper "Sleeper Agents: Training Robustly Deceptive LLMs that Persist Through Safety Training".
  ☆124Updated last year
• TransluceAI / observatory
  A toolkit for describing model features and intervening on those features to steer behavior.
  ☆227Updated last month
• callummcdougall / sae_vis
  Create feature-centric and prompt-centric visualizations for sparse autoencoders (like those from Anthropic's published research).
  ☆238Updated last year
• redwoodresearch / alignment_faking_public
  ☆82Updated 3 months ago
• ndif-team / nnterp
  Unified access to Large Language Model modules using NNsight
  ☆81Updated 2 weeks ago
• callummcdougall / ARENA_2.0
  Resources for skilling up in AI alignment research engineering. Covers basics of deep learning, mechanistic interpretability, and RL.
  ☆238Updated 5 months ago
• GraySwanAI / circuit-breakers
  Improving Alignment and Robustness with Circuit Breakers
  ☆255Updated last year
• steering-vectors / steering-vectors
  Steering vectors for transformer language models in Pytorch / Huggingface
  ☆138Updated 11 months ago
• ajobi-uhc / seer
  This was designed for interp researchers who want to do research on or with interp agents to give quality of life improvements and fix …
  ☆116Updated last week
• EleutherAI / delphi
  Delphi was the home of a temple to Phoebus Apollo, which famously had the inscription, 'Know Thyself.' This library lets language models …
  ☆235Updated last week
• google-deepmind / dangerous-capability-evaluations
  ☆65Updated 2 weeks ago
• safety-research / safety-examples
  ☆24Updated 2 months ago
• ndif-team / nnsight
  The nnsight package enables interpreting and manipulating the internals of deep learned models.
  ☆782Updated this week
• timaeus-research / devinterp
  Tools for studying developmental interpretability in neural networks.
  ☆124Updated 3 weeks ago
• goodfire-ai / scribe
  ☆62Updated 2 weeks ago
• METR / RE-Bench
  ☆133Updated 3 months ago
• nrimsky / LM-exp
  LLM experiments done during SERI MATS - focusing on activation steering / interpreting activation spaces
  ☆100Updated 2 years ago
• ai-safety-foundation / sparse_autoencoder
  Sparse Autoencoder for Mechanistic Interpretability
  ☆289Updated last year
• anthropics / evals
  ☆328Updated last year
• andyrdt / refusal_direction
  Code and results accompanying the paper "Refusal in Language Models Is Mediated by a Single Direction".
  ☆333Updated 7 months ago
• saprmarks / feature-circuits
  ☆204Updated 3 months ago
• aypan17 / machiavelli
  ☆144Updated 6 months ago
• science-of-finetuning / diffing-toolkit
  A toolkit that provides a range of model diffing techniques including a UI to visualize them interactively.
  ☆50Updated last week
• neelnanda-io / 1L-Sparse-Autoencoder
  ☆132Updated 2 years ago