Summary of Routerdc: Query-based Router by Dual Contrastive Learning For Assembling Large Language Models, By Shuhao Chen et al.
RouterDC: Query-Based Router by Dual Contrastive Learning for Assembling Large Language Models
by Shuhao Chen, Weisen Jiang, Baijiong Lin, James T. Kwok, Yu Zhang
First submitted to arxiv on: 30 Sep 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
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 query-based Router by Dual Contrastive learning (RouterDC) method assembles off-the-shelf large language models (LLMs) to harness their complementary abilities. The model consists of an encoder and LLM embeddings, and uses two contrastive learning losses to train the router. Experimental results show that RouterDC outperforms individual top-performing LLMs and existing routing methods on both in-distribution (+2.76%) and out-of-distribution (+1.90%) tasks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary RouterDC is a new approach for combining large language models. It works by using two losses to train the model, which helps it choose the right model for each question. The results show that RouterDC does better than just using one of the best individual models or existing methods. This could be useful for natural language processing tasks. |
Keywords
* Artificial intelligence * Encoder * Natural language processing
