WaveWave 2

XLANG Lab

Welcome to the Executable Language Grounding (XLANG) Lab! We are part of the HKU NLP Group at the University of Hong Kong. We focus on developing grounded AI agents that empower users to use language to interact with digital and physical environments to carry out real-world tasks. Our systems ground language and perception into code and actions executable in the corresponding environments, including databases (data/coding agent), computers (computer use agent), and the physical world (robotic agent) etc,. Through these agents, we aim to enable non-experts to access complex systems such as databases, software, and robots while unlocking functionalities across existing applications and physical systems that dramatically expand AI capabilities.

xlang-overview

News

01/24/2025
OSWorld is used in benchmarking OpenAI Computer-Using Agent (Operator) performance, which scores 38.1% success rate.
12/24/2024
Introducing Aguvis - A unified vision-based strong agent model for autonomous GUI interaction across web, desktop & mobile platforms.
12/15/2024
Instructor embeddings recently hit 5 million downloads on huggingface!
11/15/2024
6 years after our Yale Spider 1.0, we're introducing Spider 2.0, the real-world enterprise agentic Text-to-SQL workflow challenge in the LLM era!
10/23/2024
Excited to see Anthropic using our OSWorld (NeurIPS'24) to benchmark their computer use!
04/11/2024
We have released  OSWorld,  A unified, real computer env for multimodal agents to evaluate open-ended computer tasks with arbitrary apps and interfaces on Ubuntu, Windows, & macOS!  
10/18/2023
We have released  💥OpenAgents💥,  an open platform designed for language agents in the wild! For more details, you can visit our paper and the code!  
10/13/2023
We have released  Lemur70B,  🚀 Open & SOTA Foundation Models for Language Agents! The closest open model to GPT-3.5 on 🤖15 agent tasks🤖! ! Check out our paper and feel free to download and use the model at  HuggingFace
09/28/2023
Introducing Text2Reward - Using LLMs to generate dense reward functions from natural language for robotic RL policy training!

Acknowledgements

We thank the following institutions for their funding support: Google Research, Amazon AWS, Salesforce Research, and UGC.

Google ResearchAmazon AWSSalesforce ResearchUGC
Xlang
© Copyright 2023 XLANG Lab. All right reserved.