Search Results for author:
Found 6 papers, 4 papers with code
Testing the Limits of Jailbreaking Defenses with the Purple Problem
The rise of "jailbreak" attacks on language models has led to a flurry of defenses aimed at preventing undesirable responses.
A Safe Harbor for AI Evaluation and Red Teaming
Independent evaluation and red teaming are critical for identifying the risks posed by generative AI systems.
Repetition Improves Language Model Embeddings
In this work, we address an architectural limitation of autoregressive models: token embeddings cannot contain information from tokens that appear later in the input.
Understanding Catastrophic Forgetting in Language Models via Implicit Inference
We lack a systematic understanding of the effects of fine-tuning (via methods such as instruction-tuning or reinforcement learning from human feedback), particularly on tasks outside the narrow fine-tuning distribution.
Provably Bounding Neural Network Preimages
Most work on the formal verification of neural networks has focused on bounding the set of outputs that correspond to a given set of inputs (for example, bounded perturbations of a nominal input).
CELESTIAL: Classification Enabled via Labelless Embeddings with Self-supervised Telescope Image Analysis Learning
To solve this problem, we establish CELESTIAL-a self-supervised learning pipeline for effectively leveraging sparsely-labeled satellite imagery.
