Summary of On the Multi-turn Instruction Following For Conversational Web Agents, by Yang Deng et al.
On the Multi-turn Instruction Following for Conversational Web Agents
by Yang Deng, Xuan Zhang, Wenxuan Zhang, Yifei Yuan, See-Kiong Ng, Tat-Seng Chua
First submitted to arxiv on: 23 Feb 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 This paper presents a new challenge for web agents powered by Large Language Models (LLMs): effectively engaging with sequential user instructions in real-world scenarios through Conversational Web Navigation. To address this task’s complexities, the authors develop a novel framework called Self-reflective Memory-Augmented Planning (Self-MAP), which leverages memory utilization and self-reflection techniques to overcome limitations of LLMs’ context length and conversational tasks’ context dependency. The authors also introduce a dataset, Multi-Turn Mind2Web (MT-Mind2Web), designed specifically for this task. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper shows how web agents can talk with people on the internet in a smart way. Right now, these agents are great at following instructions and doing tasks, but they don’t always understand what’s going on or remember things from earlier conversations. To fix this, scientists created a new way for agents to plan ahead using memories of past conversations. They also made a special set of examples, called MT-Mind2Web, that shows how well this new planning works. |
Keywords
» Artificial intelligence » Context length
