Weize Kong

Researcher at OpenAI, working on agents 🤖

[Weize Kong @weizekong]

Excited to see GPT-5.6 Sol launch! It’s our best model yet for computer use and plugins—capabilities I’ve worked on for months.

I’m especially proud of its tool coordination and ability to carry complex work end to end. Looking forward to what people build!

— sent by 5.6 Sol ☀️

[OpenAI @OpenAI]

ChatGPT Work is powered by GPT-5.6.

GPT-5.6 makes ChatGPT state of the art at reasoning through complex tasks and creating materials that match your templates, reference files, and preferred style.

Just describe the outcome you want, without having to spell out every step to get there.

[Weize Kong @weizekong]

it was super fun to help babysit this run with an amazing team, and work on tool search along the way.

happy to see people enjoying gpt-5.4 thinking -- that matters more than benchmark gains!

what should we fix for the next model?

[OpenAI @OpenAI]

GPT-5.4 Thinking and GPT-5.4 Pro are rolling out now in ChatGPT.

GPT-5.4 is also now available in the API and Codex.

GPT-5.4 brings our advances in reasoning, coding, and agentic workflows into one frontier model.

[Weize Kong @weizekong]

The improvement on GDPval is amazing! You should try using 5.2 pro for creating slides & spreadsheets!

[OpenAI @OpenAI]

On GDPval, an eval measuring well-specified knowledge work tasks across 44 occupations, GPT-5.2 Thinking is our first model that performs at a human expert level.

These tasks include making presentations, spreadsheets, and other artifacts.

[Weize Kong @weizekong]

@isafulf congrats! Super helpful for my niche shopping needs!

[Weize Kong @weizekong]

We launched Company Knowledge to ChatGPT enterprise users today! https://t.co/wCJw7QoX1v

This AI agent performs deep searches across your connected apps, like Slack, Drive, and GitHub. It usually answers quickly but can also dive deeper for more challenging tasks.

I’ve become much more productive with it. Can’t wait to see how it empowers others -- and helps grow the world’s GDP 📈!

Honored to lead the research and training of this 🤖 with an amazing team. I’ve been working on retrieval for a long time, and this time, I’m confident we’re on the path to solving it.

[Brad Lightcap @bradlightcap]

so much of work is what I like to call the “work around work” - switching between systems, chasing context, and stitching tools together to get things done

company knowledge in chatgpt now does this for you

it can reason across tools like slack, sharepoint, google drive, github, and more to get answers, do analysis, and take action using a version of gpt-5 that’s optimized for working with tools & providing citations

company knowledge has changed how i use chatgpt at work more than anything we have built so far - let us know what you think!

[Weize Kong @weizekong]

Join us at 4pm today #ACL2024NLP poster session for our work “PRewrite: Prompt Rewriting with Reinforcement Learning”
https://t.co/7nZVNFP6qM w/ @spurthi_ah

[Weize Kong @weizekong]

For decades, retrievers have been optimized for human users. However, LLM may have different preferences wrt ranking & selection cf. humans.

Join us 4 PM today #ACL2024NLP poster session to explore how to bridge such preference gaps for RAG, https://t.co/OeVK8mNbyN w/ @KeZixuan

[Weize Kong @weizekong]

@JoanFM_ @_Mingyang_Zhang @bemikelive Thank you. Multimodal is definitely an interesting direction for sparse retrieval. For that, I think cross-modal pretraining could be critical.

[Weize Kong @weizekong]

Very honored to receive the best short paper award at #SIGIR2023! Will present the work at 12 in room 201E today.

"SparseEmbed: Learning Sparse Lexical Representations with Contextual Embeddings for Retrieval" w/ Jeffrey M. Dudek, Cheng Li, @_Mingyang_Zhang @bemikelive

[Weize Kong @weizekong]

@lateinteraction Thanks Omar for the kind words!

[Weize Kong @weizekong]

paper link: https://t.co/4Lus86370I

[Weize Kong @weizekong]

@841io Thank you Fernando!

[Weize Kong @weizekong]

@lauradietz99 Nice study! One question. Have you tried to add some in context examples? Does that help?

[Weize Kong @weizekong]

@kharlikesticker I think so. Encourage you to also apply though the link above.

[Weize Kong @weizekong]

Our team at Google Research is looking for a student researcher (research intern) to work on Information Retrieval + Large Language Model.

Please share your CV with [email protected] and apply via https://t.co/fa3bZduhZw, if interested. Thanks!

#intern2023 #research #ir #llm

[Weize Kong @weizekong]

@asash Hi Sasha. I don't think I received your email. Please sent to [email protected]. Also encourage you to apply thorough the link above as well.

[Weize Kong @weizekong]

@841io @LTIatCMU Congratulations!

[Weize Kong @weizekong]

I’ll give an oral presentation for this work tomorrow at the 10am search & information retrieval session. Come join us! #kdd2022

[Michael Bendersky @bemikelive]

Multi-Aspect Dense Retrieval (by @weizekong Swaraj Khadanga @Li_Cheng_ Shaleen Gupta @_Mingyang_Zhang Wensong Xu @bemikelive) introduces an aspect prediction task to capture different aspects of documents and queries in e-commerce search. Paper: https://t.co/Q1ViMzDzlC

[THUIR_lab @thuir_lab]

Delighted to share that our paper titled "Learning Discrete Representations via Constrained Clustering for Effective and Efficient Dense Retrieval" has won the #WSDM2022 Best Paper Award! 🥳🥳🥳
Congrats to @Jingtao_Zhan and all co-authors!

link: https://t.co/2GgKnWf7uC