🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Checkout our #ICML2024 poster on a framework for knowledge localization and model editing in t2i models (including SDXL)! Unfortunately, I am not attending ICML - but my lab mates will be presenting our poster! Do check it out on 24th July (1:30pm local Vienna Time).

Excited to share Lottery Ticket Adaptation (LoTA)! We propose a sparse adaptation method that finetunes only a sparse subset of the weights. LoTA mitigates catastrophic forgetting and enables model merging by breaking the destructive interference between tasks. 🧵👇

Glad to see Idefics-3 use our PixelProse captioning dataset for training their model 😀 Dataset Link on HF - huggingface.co/datasets/tomg-… IDEFICS 3 - arxiv.org/abs/2408.12637

Super happy to announce that our work has been accepted at #NeurIPS2024. tl;dr: our simple yet effective goldfish loss mitigates memorization even after training a 7B LLM for 100 epochs by masking tokens during loss computation. See you all in Vancouver in December! :)

Today, we’re introducing Ideogram Canvas, an infinite creative board for organizing, generating, editing, and combining images. Bring your face or brand visuals to Ideogram Canvas and use industry-leading Magic Fill and Extend to blend them with creative, AI-generated content.

📢I’ll be admitting multiple PhD students this winter to Columbia University 🏙️ in the most exciting city in the world! If you are interested in dissecting modern deep learning systems to probe how they work, advancing AI safety, or automating data science, apply to my group.

Checkout our #emnlp2024 presentation today at 11am EST (Session 02) in Miami, on upgrading CLIP for improved compositionality using feedback from diffusion models! Unfortunately, I cannot be there in-person but my colleague Soumya Suvra Ghosal will be presenting for us!

Very excited to share our work on building Edify Image - a family of diffusion models for various image generation applications. Please check it out.

Speedy-Splat: Fast 3D Gaussian Splatting with Sparse Pixels and Sparse Primitives Contributions: 1. SnugBox: A precise algorithm for computing Gaussiantile bounding box intersections. 2. AccuTile: An extension of SnugBo for computing exact Gaussian-tile intersections. 3. Soft

Speedy-Splat: Fast 3D Gaussian Splatting with Sparse Pixels and Sparse Primitives Alex Hanson, Allen Tu, Geng Lin, Vasu Singla, Matthias Zwicker, Tom Goldstein tl;dr: SnugBox+AccuTile->precisely localize Gaussians; Soft+Hard Pruning arxiv.org/abs/2412.00578

My team at Google DeepMind is hiring. If you are passionate about robust ML, the provenance of synthetic media, and the trustworthiness of data, consider applying: boards.greenhouse.io/deepmind/jobs/…

New open source reasoning model! Huginn-3.5B reasons implicitly in latent space 🧠 Unlike O1 and R1, latent reasoning doesn’t need special chain-of-thought training data, and doesn't produce extra CoT tokens at test time. We trained on 800B tokens 👇

Ok, so I can finally talk about this! We spent the last year (actually a bit longer) training an LLM with recurrent depth at scale. The model has an internal latent space in which it can adaptively spend more compute to think longer. I think the tech report ...🐦‍⬛

📢 We're hiring a Postdoctoral Associate to research 3D scene reconstruction, novel view synthesis, and inverse rendering. Join our team and contribute to cutting-edge projects in computer vision! 🔗 go.umd.edu/PostDoc2-2025

Looking at the reviews in ICML, I am noticing more and more that some reviewers are assuming knowledge or rumors that may or may not exist in industry labs. This isn't great for open research

Attention sinks in LLMs are weird. There’s ~20% of heads that don’t seem to do anything. Do these heads matter? Turns out that if we get rid of them, benchmark scores don’t change.

wait for it..... wait for it......... now, there it is. Something fun is cooking!