Summary of Large Language Model As a Teacher For Zero-shot Tagging at Extreme Scales, by Jinbin Zhang et al.
Large Language Model as a Teacher for Zero-shot Tagging at Extreme Scales
by Jinbin Zhang, Nasib Ullah, Rohit Babbar
First submitted to arxiv on: 13 Jun 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 introduces LMTX (Large language Model as Teacher for eXtreme classification), a novel framework that tackles the extreme zero-shot multi-label text classification (EZ-XMC) challenge. This challenge involves selecting relevant labels from a vast label set without annotated data, making it crucial for addressing cold-start problems in large-scale recommendation and categorization systems. LMTX bridges the gap between state-of-the-art methods like MACLR and RTS, which rely on suboptimal pseudo labels, and LLM-based approaches like ICXML, which are computationally expensive. LMTX uses an LLM to identify high-quality pseudo labels during training and a lightweight bi-encoder for efficient inference, eliminating the need for LLMs at inference time. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper is about finding the right words or labels that match what someone says, but without any help from humans. This is important because it can make things like music recommendations or classifying news articles much better. The problem with current methods is they use fake labels that aren’t always accurate. A new approach called LMTX tries to fix this by using a special kind of computer model to find the right labels, while also being fast and efficient. |
Keywords
» Artificial intelligence » Classification » Encoder » Inference » Large language model » Text classification » Zero shot
