Hello, I'm Manley. I'm a research scientist at Abacus.AI, working on research in (mostly) robustness of modern LLMs. Previously, I completed my MS in Machine Learning at Carnegie Mellon University while conducting research in the ACMI Lab. Prior to that, I completed my BS in Computer Science at Georgia Tech.
A Natural Experiment on LLM Data Contamination in Code Generation
ICBINB Workshop at NeurIPS 2023, Contributed Talk
Roberts, Thakur, Herlihy, White, and Dooley
Giraffe: Adventures in Expanding Context Lengths in LLMs
Arxiv preprint
Pal, Karkhanis, Roberts, Dooley, Sundararajan, and Naidu
Unsupervised Learning under Latent Label Shift
NeurIPS 2022
SCIS Workshop at ICML 2022
Roberts*, Mani*, Garg, and Lipton