Scholar
Anthony Hartshorn
Google Scholar ID: F3SdygkAAAAJ
Machine Learning Engineer, Genie AI
Citations & Impact
Citations
30,761
H-index
15
i10-index
15
Publications
17
Co-authors
0
Resume
Academic Achievements
- Published 'What Characterizes Effective Reasoning? Revisiting Length, Review, and Structure of CoT'
- Published 'Calibrating Verbal Uncertainty as a Linear Feature to Reduce Hallucinations'
- Co-developed 'HalluLens: LLM Hallucination Benchmark'
- Contributed to 'The Llama 3 Herd of Models'
- Contributed to 'Llama 2: Open Foundation and Fine-Tuned Chat Models'
- Published 'Improving clinical trial design using interpretable machine learning based prediction of early trial termination'
- Contributed to 'Galactica: A Large Language Model for Science'
- Published 'Assessing Robustness of Text Classification through Maximal Safe Radius Computation'
Research Experience
- Meta (2022–present): Research Engineer, contributing to Llama 2, 3, 4, and Galactica projects.
- Genie AI (2018–2022): Machine Learning Lead, leading multiple collaborative research initiatives.
- Two-year InnovateUK collaboration with Marta Kwiatkowska’s team at Oxford: focused on explainability and robustness in NLP; developed a game-theoretic framework for explaining Transformers, a Monte Carlo tree search method for adversarial robustness assessment, and causal explanation techniques.
- Two-year InnovateUK collaboration with the Computational Privacy Group at Imperial College: developed methods to assess vulnerability to authorship attribution attacks on text and potential mitigation strategies.
- Signal AI: researched indirect reported speech detection in news articles (identifying unquoted paraphrased statements).
- Advised startups AimHi and Lindus Health on machine learning.
- Previously founded a legal tech startup through Entrepreneur First (EF, LD10 cohort); gained valuable experience despite the venture’s eventual shutdown.
- Early career: Java developer at IBlocks and software engineer at TNG Technology Consulting in Germany.