Search Results for author:
Found 3 papers, 3 papers with code
Exploiting Novel GPT-4 APIs
Language model attacks typically assume one of two extreme threat models: full white-box access to model weights, or black-box access limited to a text generation API.
Codebook Features: Sparse and Discrete Interpretability for Neural Networks
In this setting, our approach overcomes the superposition problem by assigning states to distinct codes, and we find that we can make the neural network behave as if it is in a different state by activating the code for that state.
imitation: Clean Imitation Learning Implementations
imitation provides open-source implementations of imitation and reward learning algorithms in PyTorch.
Cannot find the paper you are looking for? You can
Submit a new open access paper.
