Summary of Datelogicqa: Benchmarking Temporal Biases in Large Language Models, by Gagan Bhatia et al.
DateLogicQA: Benchmarking Temporal Biases in Large Language Models
by Gagan Bhatia, MingZe Tang, Cristina Mahanta, Madiha Kazi
First submitted to arxiv on: 17 Dec 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
Summary difficulty | Written by | Summary |
---|---|---|
High | Paper authors | High Difficulty Summary Read the original abstract here |
Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed DateLogicQA benchmark assesses language models’ capabilities in temporal reasoning, featuring 190 questions covering various date formats, contexts, and reasoning types. The Semantic Integrity Metric is introduced to evaluate tokenization quality, and two biases are analyzed: Representation-Level Bias affecting embeddings and Logical-Level Bias influencing reasoning outputs. The study provides a comprehensive evaluation of large language models’ strengths and limitations in handling temporal data accurately. |
Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper introduces DateLogicQA, a benchmark with 190 questions about dates, times, and logic. It also explains two biases that affect how well models work: one that affects how words are represented and another that affects how models make logical conclusions. The study shows what language models can and cannot do when working with time-related data. |
Keywords
» Artificial intelligence » Tokenization