google-deepmind / dangerous-capability-evaluationsLinks
☆65Updated last week
Alternatives and similar repositories for dangerous-capability-evaluations
Users that are interested in dangerous-capability-evaluations are comparing it to the libraries listed below
Sorting:
- Contains random samples referenced in the paper "Sleeper Agents: Training Robustly Deceptive LLMs that Persist Through Safety Training".☆127Updated last year
- ☆133Updated 3 months ago
- Investigating the generalization behavior of LM probes trained to predict truth labels: (1) from one annotator to another, and (2) from e…☆28Updated last year
- METR Task Standard☆173Updated last year
- ☆153Updated 5 months ago
- Inference API for many LLMs and other useful tools for empirical research☆104Updated this week
- Open source interpretability artefacts for R1.☆170Updated 9 months ago
- Vivaria is METR's tool for running evaluations and conducting agent elicitation research.☆134Updated this week
- Keeping language models honest by directly eliciting knowledge encoded in their activations.☆217Updated 2 weeks ago
- Open Source Replication of Anthropic's Alignment Faking Paper☆54Updated 10 months ago
- Functional Benchmarks and the Reasoning Gap☆89Updated last year
- ☆118Updated 3 weeks ago
- A toolkit for describing model features and intervening on those features to steer behavior.☆228Updated last month
- datasets from the paper "Towards Understanding Sycophancy in Language Models"☆102Updated 2 years ago
- Code for the ICLR 2024 paper "How to catch an AI liar: Lie detection in black-box LLMs by asking unrelated questions"☆71Updated last year
- ☆144Updated 6 months ago
- Notebooks accompanying Anthropic's "Toy Models of Superposition" paper☆135Updated 3 years ago
- ☆112Updated 11 months ago
- ☆107Updated last week
- ControlArena is a collection of settings, model organisms and protocols - for running control experiments.☆153Updated this week
- Fluid Language Model Benchmarking☆26Updated 4 months ago
- Attribution-based Parameter Decomposition☆33Updated 7 months ago
- Scaling is a distributed training library and installable dependency designed to scale up neural networks, with a dedicated module for tr…☆66Updated 2 months ago
- ☆34Updated 11 months ago
- Adversarial Attacks on GPT-4 via Simple Random Search [Dec 2023]☆43Updated last year
- we got you bro☆37Updated last year
- Delphi was the home of a temple to Phoebus Apollo, which famously had the inscription, 'Know Thyself.' This library lets language models …☆241Updated 2 weeks ago
- Sparse Autoencoder Training Library☆56Updated 9 months ago
- Redwood Research's transformer interpretability tools☆15Updated 3 years ago
- ☆259Updated 3 weeks ago