Log inSign up
Xing Han Lu
2,658 posts
Image
user avatar
Xing Han Lu
@xhluca
Vibe agents @Mila_Quebec @McGill_NLP
The Wired
xinghanlu.com
Joined December 2017
354
Following
2,360
Followers

New to X?

Sign up now to get your own personalized timeline!

Create account

By signing up, you agree to the Terms of Service and Privacy Policy, including Cookie Use.

Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
Don't miss what's happening
People on X are the first to know.
Log inSign up
  • Pinned
    user avatar
    Xing Han Lu
    @xhluca
    Apr 10
    Frontier LLMs can navigate complex websites, but are expensive and can't run locally. At the same time, small open models can't match the capabilities of commercial APIs. Can we close this gap with synthetic data? To answer this, we built Agent-as-Annotators (A3): a framework
    Image
    4.2K
  • user avatar
    Xing Han Lu
    @xhluca
    Apr 22, 2024
    Delighted to release ✨Llama-3-8B-Web✨, the most capable agent built for web navigation by following instructions and replying💬. It surpasses GPT-4V* by 18% on WebLINX, a benchmark for web navigation with dialogue. Model: huggingface.co/McGill-NLP/Lla… Code: github.com/McGill-NLP/web…
    Image
    Image
    563K
  • user avatar
    Xing Han Lu
    @xhluca
    Jun 18, 2024
    Announcing ⚡BM25S, a fast lexical retrieval library! 🏎️ Up to 500x faster than the most popular Python lib, matches @elastic search (BM25 default) 🤗 First BM25 library that is directly integrated with @huggingface hub: load or save in 1 line! GitHub: github.com/xhluca/bm25s
    Image
    Image
    Image
    91K
  • user avatar
    Xing Han Lu
    @xhluca
    Mar 17, 2021
    Now that @huggingface's transformers 4.4.0 is out, I'm happy to release dl-translate, a library for text translation between 50 languages and built on top of 🤗 transformers and @facebookai's mBART Repo: git.io/dl-translate Docs: dl-translate.rtfd.io
    Image
  • user avatar
    Xing Han Lu
    @xhluca
    Jun 8, 2024
    Another solid work by the Android agents team at @GoogleDeepMind. 800+ apps interactions is massive. It would probably cover most of the use cases we can see on Android (so one can spend less time on thinking about generalization and focus on doing well on validation).
    Image
    40K
  • user avatar
    Xing Han Lu
    @xhluca
    Jun 14, 2023
    CTRL (arxiv.org/abs/1909.05858) by Keskar+ is such an under-appreciated paper, considering it introduce the idea of control code/prompts for controllable generation (based on GPT-2) almost 3 years before InstructGPT.
    23K
  • user avatar
    Xing Han Lu
    @xhluca
    Mar 27, 2024
    Is this the DSPy moment of text-to-image generation? Congratulations @oscmansan @Piovrasca et al!
    Image
    Image
    user avatar
    AK
    @_akhaliq
    Mar 27, 2024
    Improving Text-to-Image Consistency via Automatic Prompt Optimization Impressive advances in text-to-image (T2I) generative models have yielded a plethora of high performing models which are able to generate aesthetically appealing, photorealistic images. Despite the progress,
    13K
  • user avatar
    Xing Han Lu
    @xhluca
    Dec 12, 2021
    I think this @huggingface blog post by @PatrickPlaten is by far my favorite resource on post-hoc text generation
    Image
    How to generate text: using different decoding methods for language generation with Transformers
    From huggingface.co
  • user avatar
    Xing Han Lu
    @xhluca
    Mar 25, 2024
    I'm seriously surprised I am not seeing more hype around @valence_ai @PrescientDesign @IsomorphicLabs Those groups are clearly working on the most impactful applications of ML and are at the forefront of biotech breakthroughs.
    14K
  • user avatar
    Xing Han Lu
    @xhluca
    Feb 12, 2024
    I'm going to make a thread later on why I think WebLINX (mcgill-nlp.github.io/weblinx) is really exciting, but I want to take a moment to say how amazing the @huggingface ecosystem has become. You can host the crucial parts of a project all at the same place. Let's take a look 👇
    43K
  • user avatar
    Xing Han Lu
    @xhluca
    Jun 23, 2024
    Twitter threads are temporary, but blog posts are forever. So I combined this thread and the readme into a blog post posted to @huggingface community blog:
    user avatar
    Xing Han Lu
    @xhluca
    Jun 18, 2024
    Announcing ⚡BM25S, a fast lexical retrieval library! 🏎️ Up to 500x faster than the most popular Python lib, matches @elastic search (BM25 default) 🤗 First BM25 library that is directly integrated with @huggingface hub: load or save in 1 line! GitHub: github.com/xhluca/bm25s
    Image
    Image
    Image
    Image
    BM25 for Python: Achieving high performance while simplifying dependencies with *BM25S*⚡
    From huggingface.co
    7.9K
  • user avatar
    Xing Han Lu
    @xhluca
    Apr 12, 2021
    The @huggingface hub is still new but it's already super convenient for downloading transformers. I was looking for a way to download libraries + models so git.io/dl-translate could be used offline. In total it took 6 lines (4 in terminal + 2 in python) thanks to the hub ⬇️
    https://git.io/dl-translate#offline-usage
  • user avatar
    Xing Han Lu
    @xhluca
    Jul 14, 2023
    This is probably the nicest looking poster at #ACL2023NLP (and also very interesting + insightful!)
    Poster about Neural Machine Translation for the indigenous languages of the Americas
    5.8K
  • user avatar
    Xing Han Lu
    @xhluca
    Jan 25, 2024
    Replying to @srush_nlp and @BlancheMinerva
    The 6's only weakness was: The model still has a quadratic memory requirement during training like Transformers. Except it doesn't...
    4.4K
This post is unavailable.