Goto

Collaborating Authors

 Large Language Model

Spectral Editing of Activations for Large Language Model Alignment

Neural Information Processing Systems

We also extend our method to non-linear editing using feature functions.

681fe4ec554beabdc9c84a1780cd5a8a-Paper-Datasets_and_Benchmarks_Track.pdf

680da2fd0331deecc2e5b7cf0e55e832-Paper-Conference.pdf

6808f2c57d9564a2639a4710e3bbd9b9-Paper-Conference.pdf

6801fa3fd290229efc490ee0cf1c5687-Paper-Conference.pdf

Causal language modeling can elicit search and reasoning capabilities on logic puzzles

Neural Information Processing Systems

Causal language modeling using the Transformer architecture has yielded remarkable capabilities in Large Language Models (LLMs) over the last few years.

67b0e7c7c2a5780aeefe3b79caac106e-Paper-Conference.pdf

An Efficient Recipe for Long Context Extension via Middle-Focused Positional Encoding

Neural Information Processing Systems

Transformer-based Large Language Models (LLMs) are typically pre-trained with a fixed context window size, e.g ., 4K tokens in Touvron et al. [2023a].

665bb142d4b9f55660cb89bb56a66fe1-Paper-Conference.pdf

Memorization with Compression