Keeping language models honest by directly eliciting knowledge encoded in their activations.
☆217Updated this week
Alternatives and similar repositories for elk
Users that are interested in elk are comparing it to the libraries listed below
Sorting:
- ☆284Mar 2, 2024Updated last year
- Erasing concepts from neural representations with provable guarantees☆243Jan 27, 2025Updated last year
- Investigating the generalization behavior of LM probes trained to predict truth labels: (1) from one annotator to another, and (2) from e…☆28May 23, 2024Updated last year
- Interpreting how transformers simulate agents performing RL tasks☆90Oct 23, 2023Updated 2 years ago
- Tools for understanding how transformer predictions are built layer-by-layer☆567Aug 7, 2025Updated 6 months ago
- Machine Learning for Alignment Bootcamp☆82Apr 27, 2022Updated 3 years ago
- Code for the ICLR 2024 paper "How to catch an AI liar: Lie detection in black-box LLMs by asking unrelated questions"☆71Jun 19, 2024Updated last year
- Minimum Description Length probing for neural network representations☆20Jan 28, 2025Updated last year
- A library for mechanistic interpretability of GPT-style language models☆3,112Updated this week
- Code for our paper "Localizing Lying in Llama"☆13Apr 24, 2025Updated 10 months ago
- ☆44Nov 17, 2024Updated last year
- Understanding how features learned by neural networks evolve throughout training☆41Oct 24, 2024Updated last year
- Redwood Research's transformer interpretability tools☆15Apr 15, 2022Updated 3 years ago
- Sparse Autoencoder for Mechanistic Interpretability☆292Jul 20, 2024Updated last year
- A library for mechanistic anomaly detection☆22Jan 9, 2025Updated last year
- Vivaria is METR's tool for running evaluations and conducting agent elicitation research.☆134Feb 15, 2026Updated 2 weeks ago
- Improving Alignment and Robustness with Circuit Breakers☆258Sep 24, 2024Updated last year
- Pile Deduplication Code☆18May 15, 2023Updated 2 years ago
- Repo accompanying our paper "Do Llamas Work in English? On the Latent Language of Multilingual Transformers".☆80Mar 11, 2024Updated last year
- A repository for transformer critique learning and generation☆89Dec 7, 2023Updated 2 years ago
- Open source replication of Anthropic's Crosscoders for Model Diffing☆64Oct 27, 2024Updated last year
- Utilities for the HuggingFace transformers library☆75Jan 21, 2023Updated 3 years ago
- Steering Llama 2 with Contrastive Activation Addition☆212May 23, 2024Updated last year
- Inspect: A framework for large language model evaluations☆1,783Updated this week
- ☆271Oct 1, 2024Updated last year
- ☆330Jul 2, 2024Updated last year
- ☆19Jan 21, 2023Updated 3 years ago
- ☆20Feb 17, 2023Updated 3 years ago
- The nnsight package enables interpreting and manipulating the internals of deep learned models.☆825Updated this week
- ☆12Jul 12, 2024Updated last year
- Script for downloading GitHub.☆13Sep 24, 2020Updated 5 years ago
- Mechanistic Interpretability for Transformer Models☆53Jun 1, 2022Updated 3 years ago
- This is the starter kit for the Trojan Detection Challenge 2023 (LLM Edition), a NeurIPS 2023 competition.☆90May 19, 2024Updated last year
- A library for bridging Python and HTML/Javascript (via Svelte) for creating interactive visualizations☆205Dec 22, 2021Updated 4 years ago
- A collection of different ways to implement accessing and modifying internal model activations for LLMs☆20Oct 18, 2024Updated last year
- Create feature-centric and prompt-centric visualizations for sparse autoencoders (like those from Anthropic's published research).☆247Updated this week
- ☆11Jun 2, 2021Updated 4 years ago
- Finding trojans in aligned LLMs. Official repository for the competition hosted at SaTML 2024.☆116Jun 13, 2024Updated last year
- A python sdk for LLM finetuning and inference on runpod infrastructure☆19Feb 16, 2026Updated last week