thestephencasper

benchmarking_interpretability

everything-you-need

we got you bro

explore_establish_exploit_llms

gpt4_bs

Examples of prompts that cause ChatGPT-4 to hallucinate.

latent_adversarial_training

feature_level_adv

Demo code for the paper: One Thing to Fool them All: Generating Interpretable, Universal, and Physically-Realizable Adversarial Features

sd_cycle_consistency

mechanistic_interpretability_challenge

local_specialization

Detecting modularity in deep neural networks

snafue

white_box_rarl

common_claim

GRIT

Genome Recoding Informatics Toolbox

casper_igem_labs

Computer science labs used for the Havard iGEM club.

discovering_latent_knowledge

football

Check out the new game server:

lm_white_box_attacks

multiagent-competition

Code for the paper "Emergent Complexity via Multi-agent Competition"

trlx

A repo for distributed training of language models with Reinforcement Learning via Human Feedback (RLHF)