Jesse DodgeResearch Scientist, MSL · MetaResearcher↗
Researcher
MetaResearch Scientist
Williams helped create MultiNLI, a widely used dataset for testing whether models understand relationships between sentences. She later co-developed Dynabench, where people continually find examples that fool models, replacing static tests with an evolving evaluation that exposes real weaknesses.
Presented Llama 2 and its chat models, including training methods and capability evaluations.
Based on public sources. If accuracy is important, verify details using the links or references.
For corrections, updates, or requests to remove information, contact us.
Reviewed August 7, 2026 · Checked for accuracy and new information.