stay hydrated
Amsterdam
Joined January 2017
- Tweets179
- Following257
- Followers447
- Likes1.1K
mechcoder retweeted
I’m presenting “Agents’ Room: Narrative Generation through Multi-step Collaboration” at the #ICLR2025 poster session at 3pm today.
come chat about multi-agent systems for complex writing tasks!
link: iclr.cc/virtual/2025/poster/…
paper: arxiv.org/abs/2410.02603
mechcoder retweeted
🔥Excited to introduce RINS - a technique that boosts model performance by recursively applying early layers during inference without increasing model size or training compute flops! Not only does it significantly improve LMs, but also multimodal systems like SigLIP.
(1/N)
mechcoder retweeted
Want your VLM to reflect the world's rich diversity 🌍? We’re very excited to share our recent research on this topic. TLDR: to build truly inclusive models that work for everyone, don’t filter by English, and check out our recommended evaluation benchmarks. (1/7)
mechcoder retweeted
Happy to introduce GERALD - our new VLM that recognizes 6M+ entities, an exciting step towards Web-scale visual entity recognition!
Predictions are simply made by auto-regressively decoding a code representing the entity name.
Check out our CVPR24 paper: arxiv.org/abs/2403.02041
mechcoder retweeted
Excited to share our #ICLR2024 paper, focused on reducing bias in CLIP models. We study the impact of data balancing and come up with some recommendations for how to apply it effectively. Surprising insights included! Here are 3 main takeaways.
mechcoder retweeted
More AI "doomer" pushback. Thanks to @kchonyc for talking to me for this one. venturebeat.com/ai/top-ai-re…
mechcoder retweeted
Dual PatchNorm
Manoj Kumar, Mostafa Dehghani, Neil Houlsby.
Action editor: Yunhe Wang.
openreview.net/forum?id=jgMq…
#layernorm #layernorms #patchnorm
mechcoder retweeted
1/ There is a huge headroom for improving capabilities of our vision models and given the lessons we've learned from LLMs, scaling is a promising bet. We are introducing ViT-22B, the largest vision backbone reported to date: arxiv.org/abs/2302.05442
mechcoder retweeted
Dual PatchNorm is one of those ideas that is simple (adding literally two lines of code), yet consistently effective. There is also an interesting backstory to it.
nitter.gq/mechcoder/status/16215…
While we await GPT-4 which is expected to have a trillion parameters, here are 3072 parameters, that can make your Vision Transformer better.
Paper: arxiv.org/abs/2302.01327
Joint work w/ @neilhoulsby @m__dehghani
mechcoder retweeted
Understanding how a person perceives image quality is necessary for the success of many computer vision technologies. Here we discuss an experimental study comparing the accuracy of ImageNet classifiers and their ability to capture perceptual similarity. goo.gle/3Se08u0
mechcoder retweeted
The weather space stands to benefit in a transformative way from deep learning! I am thrilled to share our latest results with MetNet-2:
Blog: ai.googleblog.com/2021/11/me…
Paper: arxiv.org/abs/2111.07470
mechcoder retweeted
Continuing our work on #NeuralWeatherModels, we present an improved model that uses data from radar and satellite images, as well as information about the atmosphere, to quickly create a state-of-the-art forecast up to 12 hours ahead ↓ goo.gle/3HmZru1
mechcoder retweeted
A few weeks ago, we open-sourced SCENIC, a JAX library/codebase that we like it a lot and wanted to share our joy with the community.
GitHub: github.com/google-research/s…
Paper: arxiv.org/abs/2110.11403
ALT Scenic: A Jax Library for Computer Vision Research and Beyond
mechcoder retweeted
New paper! IDF++: analyzing and improving integer discrete flows for lossless compression, with @agritsenko @m__dehghani @CasperKaae @TimSalimans
paper: arxiv.org/abs/2006.12459
mechcoder retweeted
Me in a month: *breaks down in tears of joy*
Cashier: sir this is a Wendy’s
Me: I KNOW
mechcoder retweeted
Can deep neural nets forecast weather accurately? Check out our #NeuralWeatherModel, MetNet, which outperforms physical models at up to eight hour forecasts and runs in a matter of seconds. Learn more in the blog post below! goo.gle/2Jkm9pH
mechcoder retweeted
VideoFlow: A Flow-Based Generative Model for Video extends Glow to a new algorithm for multi-frame video prediction with normalizing flows. Tractable & scalable! Work w/ Manoj Kumar, @babaeizadeh, @chelseabfinn, @svlevine, @laurent_dinh and @dpkingma.
arxiv.org/abs/1903.01434
mechcoder retweeted
"Parallel Architecture and Hyperparameter Search via Successive Halving and Classification," Kumar et al.: arxiv.org/abs/1805.10255