Summary of Will Llms Replace the Encoder-only Models in Temporal Relation Classification?, by Gabriel Roccabruna et al.
Will LLMs Replace the Encoder-Only Models in Temporal Relation Classification?
by Gabriel Roccabruna, Massimo Rizzoli, Giuseppe Riccardi
First submitted to arxiv on: 14 Oct 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 paper investigates the performance of Large Language Models (LLMs) in detecting temporal relations among events. It finds that LLMs with in-context learning significantly underperform smaller encoder-only models based on RoBERTa in the Temporal Relation Classification task. The study uses explainable methods to delve into the possible reasons for this gap, concluding that LLMs are limited by their autoregressive nature and focus on the last part of the sequence. Additionally, it compares word embeddings between LLMs and RoBERTa-based models to understand pre-training differences. This research has implications for temporal reasoning tasks such as temporal question answering. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper looks at how good Large Language Models (LLMs) are at finding connections between events that happen over time. It finds that some LLMs don’t do very well in this task, and it tries to figure out why by looking more closely at what these models are doing. The results suggest that there might be a problem with how the LLMs process information, causing them to focus on just one part of the sequence instead of the whole thing. This research could help us better understand how LLMs work and how we can use them for tasks like answering questions about time. |
Keywords
» Artificial intelligence » Autoregressive » Classification » Encoder » Question answering
