Dan Hendrycks

Author of MMLU; Executive Director at Center for AI Safety

Authored Projects (3)

Starred by

Johannes Hagemann

Johannes Hagemann(Cofounder of Prime Intellect) and

Jinze Bai

Jinze Bai(Research Scientist at Alibaba Qwen).

apps by hendrycks

Dataset for measuring coding challenge competence (NeurIPS 2021)

Training and evaluation code for assessing programming skills.
Contains 10,000 coding problems with varying difficulty levels.
Fine-tuned GPT-2 1.5B and GPT-Neo 2.7B weights available.
Available in Hugging Face datasets.

Created 4 years ago

Updated 1 year ago

Starred by

Zack Li

Zack Li(Cofounder of Nexa AI),

Simon Willison

Simon Willison(Coauthor of Django),

Jiayi Pan

Jiayi Pan(Author of SWE-Gym; MTS at xAI),

Chip Huyen

Chip Huyen(Author of "AI Engineering", "Designing Machine Learning Systems"), and

4 more.

test by hendrycks

Research paper for measuring multitask language understanding

Evaluates language models (OpenAI API) across humanities, social sciences, STEM, and other categories.
Includes a leaderboard of model results, such as Chinchilla, Gopher, and GPT-3.
Draws from the ETHICS dataset for alignment with human values.

Created 5 years ago

Updated 2 years ago

ethics by hendrycks

ICLR 2021 research paper on aligning AI with human values

Fine-tuning scripts for tasks in the ETHICS benchmark.
Interactive script to probe commonsense morality and utilitarianism models.
Datasets and weights for morality models included.

Created 5 years ago

Updated 2 years ago

Starred Projects (3)

nanoGCG by GraySwanAI

PyTorch implementation of the Greedy Coordinate Gradient (GCG) algorithm

Created 1 year ago

Updated 4 months ago

L1B3RT4S by elder-plinius

AI jailbreak prompts

chiphuyen:

vincentweisser:

elder-plinius:

Created 1 year ago

Updated 5 days ago

HarmBench by centerforaisafety

Evaluation framework for LLM red teaming and defense

Created 1 year ago

Updated 1 year ago

Feedback? Help us improve.