Alex Murphy @alexandermurphy.bsky.social · 01/04/2025- Faithful decoding + ReAct consistently boosts performance on answering and recall across datasets and models - They enhance ReAct's ability to follow the crucial Thought-Action-Observation format - We achieved a significant Answer F1 improvement of up to 13.1 (using the DoLa decoder on HotpotQA) 010
Alex Murphy @alexandermurphy.bsky.social · 01/04/2025🚀 Thrilled to share our new preprint, "An Analysis of Decoding Methods for LLM-based Agents for Faithful Multi-Hop Question Answering"! 📄 Dive into the paper: arxiv.org/abs/2503.23415 #AI #MachineLearning #LLM #NLP #Research #QuestionAnswering #Retrieval 121
Reposted by Alex MurphyAdriana Guatame-García @adriguatame.bsky.social · 06/03/2025A friend included this anti-acknowledgement section on her PhD thesis. She also added the proposition: “Systematic bullying and undermining of girls and women in STEM starts early on and is the reason why they do not stay in science and related fields.” Absurd we still need to go through this 🧪👩🔬 12365112319