1. X
  2. Mitchell Wortsman
Log inSign up
Mitchell Wortsman
418 posts
Mitchell Wortsman profile banner
user avatar

Mitchell Wortsman

@Mitchnw
mitchellnw.github.io
Joined October 2011
1,068
Following
1,902
Followers
RepliesRepliesMediaMedia
  • Pinned
    user avatar
    Mitchell Wortsman
    @Mitchnw
    Sep 28, 2023
    Sharing some highlights from our work on small-scale proxies for large-scale Transformer training instabilities: arxiv.org/abs/2309.14322 With fantastic collaborators @peterjliu, @Locchiu, @_katieeverett, many others (see final tweet!), @hoonkp, @jmgilmer, @skornblith! (1/15)
    Image
  • user avatar
    Mitchell Wortsman
    @Mitchnw
    Jun 28, 2023
    Check out OpenFlamingo V2!
    user avatar
    Anas Awadalla
    @anas_awadalla
    Jun 28, 2023
    We are excited to announce OpenFlamingo V2 🦩! We are releasing five new multimodal models, across the 3B, 4B, and 9B scales, that outperform our previous model. w/ @irena_gao. Repo: github.com/mlfoundations/… Demo: huggingface.co/spaces/openfla… Blog: laion.ai/blog/open-flam…
  • user avatar
    Mitchell Wortsman
    @Mitchnw
    Apr 28, 2023
    Excited it’s finally here — check out DataComp!
    user avatar
    Gabriel Ilharco
    @gabriel_ilharco
    Apr 28, 2023
    Introducing DataComp, a new benchmark for multimodal datasets! We release 12.8B image-text pairs, 300+ experiments and a 1.4B subset that outcompetes compute-matched CLIP runs from OpenAI & LAION 📜 arxiv.org/abs/2304.14108 🖥️ github.com/mlfoundations/… 🌐 datacomp.ai
    Image
  • user avatar
    Mitchell Wortsman
    @Mitchnw
    Apr 26, 2023
    Thanks to the OpenCLIP team, tutorial on OpenCLIP is good to go - check it out github.com/mlfoundations/…
    user avatar
    Mitchell Wortsman
    @Mitchnw
    Apr 26, 2023
    Sharing our project on 1) accelerating and 2) stabilizing training for large language-vision models 1) Towards accelerating training, we introduce SwitchBack, a linear layer for int8 quantized training which matches bfloat16 within 0.1 for CLIP ViT-Huge arxiv.org/abs/2304.13013
    Image
  • user avatar
    Mitchell Wortsman
    @Mitchnw
    Apr 26, 2023
    Sharing our project on 1) accelerating and 2) stabilizing training for large language-vision models 1) Towards accelerating training, we introduce SwitchBack, a linear layer for int8 quantized training which matches bfloat16 within 0.1 for CLIP ViT-Huge arxiv.org/abs/2304.13013
    Image

Log in or sign up for X

See what’s happening and join the conversation

Continue with phone
or
Log in with username or email
Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
Advertisement
Advertisement