Jun Wang

Professor of Computer Science, UCL

[elvis @omarsar0]

Fine-tuning LLM Agents without Fine-tuning LLMs

Catchy title and very cool memory technique to improve deep research agents.

Great for continuous, real-time learning without gradient updates.

Here are my notes:

[Jun Wang @seawan]

@omarsar0 Thanks for sharing our work! We felt an external memory and case based reasoning could revolutionise the way a LLM agent learns with minimum efforts. It is time to get rid of llm post training and gradient descent.

[Grace Hui Yang @GraceHuiYang]

I guess we will have a lot to talk about #ChatGPT in our upcoming #wsdm2023 tutorial “proactive conversational agents.” I don’t have any concern/complaint/resistance for this intelligent new being. I look forward to my own evolution via interactions w/ this true AI.

[UK Research and Innovation @UKRI_News]

Congratulations to UKRI Board members Professor Anthony Finkelstein (@profserious), who has received a knighthood, and @JohnFingleton1, who has received a CBE.

#NewYearHonours

[Jun Wang @seawan]

Check out our multiagent reinforcement learning library together with Shanghai Jiaotong: https://t.co/l6TjM5zt8V

[@[email protected] 🕊️🇺🇦 @arjenpdevries]

So proud! Our new hire @HarrieOos wins SIGIR best paper with a single-authored contribution, and former student @seawan has an honerable mention for the Test-of-time award for his paper on Portfolio theory for IR!! #SIGIR2021

[Jun Wang @seawan]

Our weird behavior during the pandemic is messing with AI models https://t.co/RGYX7te32H

[Jun Wang @seawan]

Beeswax Buys MediaGamma to Shore Up Programmatic Offering https://t.co/8yKnXyK6Ww via @Adweek

[Jun Wang @seawan]

https://t.co/DIwsSAH7qK

[Katja Hofmann @katjahofmann]

An exciting demonstration of curiosity driven learning in #minecraft @Project_Malmo

[Google DeepMind @GoogleDeepMind]

We provide a new theoretical analysis to understand how they happen, and what we can do about them: https://t.co/ehlfaZaX12

[Matthew Hutson @SilverJacket]

"Artificially Intelligent Players Invent Nonverbal 'Languages' to Win Card Games": My piece for @IEEESpectrum on bridge and Hanabi research by @seawan @ucl @j_foerst @UniofOxford @DeepMindAI https://t.co/JMKiBTRhDw

[Jun Wang @seawan]

See our recent work on AI playing the Bridge Card game and how AI agents can learn communication language secretly.

[Matthew Hutson @SilverJacket]

"Artificially Intelligent Players Invent Nonverbal 'Languages' to Win Card Games": My piece for @IEEESpectrum on bridge and Hanabi research by @seawan @ucl @j_foerst @UniofOxford @DeepMindAI https://t.co/JMKiBTRhDw

[UCL Computer Science @uclcs]

Today @UCL launches #MadeAtUCL to showcase our ground-breaking research and discoveries changing people's lives. Know some amazing UCL research that deserves the spotlight? Submit it here: https://t.co/7sjZMl3r17 #MadeAtUni

[Jun Wang @seawan]

From finance and mobile communications to digital advertising https://t.co/7cudunH5MN

[Jun Wang @seawan]

Our recent work on AI playing Contract Bridge card game.

[Jun Wang @seawan]

We explain the difference through late 20th century mainstream song ... https://t.co/2ZjmQUFoHN

[Jun Wang @seawan]

Check out "Inaugural Lecture: On the Possibility and Engineering of Machine Consciousness " https://t.co/v8y2ZKK2vR @EventbriteUK

[Jun Wang @seawan]

Open ETL for Real Time Decision Making with Dr. Shuai Yuan (MediaGamma) https://t.co/GYHwI3BaUc via @YouTube

[Jun Wang @seawan]

New DeepMind professorship at UCL to push frontiers of AI https://t.co/eaenjHkp3V via @uclnews