André Susano Pinto

Machine learning research @GoogleAI, Opinions mine.

[André Susano Pinto @ASusanoPinto]

@mtschannen 🚀Gemma4 12B🚀
We made it great by training a simpler model.
No vision or audio encoders. Easier said than done.

Running exploratory experiments to a final model is always interesting. Joint work with @mtschannen @AndreasPSteiner @confusezius @kmisiunas and the whole Gemma team.

[André Susano Pinto @ASusanoPinto]

@mtschannen 🚀Gemma4 12B🚀
We made it great by training a simpler model.
No vision or audio encoders. Easier said than done.

Running exploratory experiments to a final model is always interesting. Joint work with @mtschannen @AndreasPSteiner @confusezius @kmisiunas and the whole Gemma team.

[Michael Tschannen @mtschannen]

Check out our detailed report about *Jet* 🌊 - a simple, transformer-based normalizing flow architecture without bells and whistles.

Jet is an important part of JetFormer's engine ⚙️ As a standalone model it is very tame and behaves predictably (e.g. when scaling it up).

[André Susano Pinto @ASusanoPinto]

Making new simple things requires attention to detail. From numeric precision and unexpected bugs deep in the stack. But now there is a precedent which includes paper, numbers and code.

Hope it helps people go hammer some nails🔨

[André Susano Pinto @ASusanoPinto]

Jet, the tool in JetFormer. A coupling normalizing flow where the blocks are powered by ViT.

Simple, scalable and it works!

[Alexander Kolesnikov @__kolesnikov__]

With some delay, JetFormer's *prequel* paper is finally out on arXiv: a radically simple ViT-based normalizing flow (NF) model that achieves SOTA results in its class.

Jet is one of the key components of JetFormer, deserving a standalone report. Let's unpack: 🧵⬇️

[merve @mervenoyann]

Welcome PaliGemma 2! 🤗

Google released PaliGemma 2, best vision language model family that comes in various sizes: 3B, 10B, 28B, based on Gemma 2 and SigLIP, comes with transformers support day-0 🎁

Saying this model is amazing would be an understatement, keep reading ✨

[Andreas Steiner @AndreasPSteiner]

🚀🚀PaliGemma 2 is our updated and improved PaliGemma release using the Gemma 2 models and providing new pre-trained checkpoints for the full cross product of {224px,448px,896px} resolutions and {3B,10B,28B} model sizes.

1/7

[André Susano Pinto @ASusanoPinto]

@AndreasPSteiner: Let's just train and see where it goes... Well far... We had to write a 30+ page tech report for all of you to enjoy :)

[Andreas Steiner @AndreasPSteiner]

🚀🚀PaliGemma 2 is our updated and improved PaliGemma release using the Gemma 2 models and providing new pre-trained checkpoints for the full cross product of {224px,448px,896px} resolutions and {3B,10B,28B} model sizes.

1/7

[André Susano Pinto @ASusanoPinto]

@YugeTen @__kolesnikov__ We already knew we would like it. But we didn't know how :) The NF comes with two properties: invertible and computable logdet. together they don't allow to cheat to map all latents to a trivial point and then obtain a perfect loss on the AR to model that trivial output.

[André Susano Pinto @ASusanoPinto]

Did you try to get an auto-regressive transformer to operate in a continuous latent space which is not fixed ahead of time but learned end to end from scratch?

Enter JetFormer: https://t.co/NaQzHGvezm -- joint work in a dream team: @mtschannen and @__kolesnikov__

[Michael Tschannen @mtschannen]

Have you ever wondered how to train an autoregressive generative transformer on text and raw pixels, without a pretrained visual tokenizer (e.g. VQ-VAE)?

We have been pondering this during summer and developed a new model: JetFormer 🌊🤖

https://t.co/ngvPzZvUYW

A thread 👇

1/

[André Susano Pinto @ASusanoPinto]

Feels great to start adding diversity to the available pre-trained visual representations. Especially when it has considerable impact for problems with a smaller number of examples available or hard to collect.

[Maxim Neumann @neu_maxim]

We've looked into representation learning for #RemoteSensing with different datasets and fine-tuning using in-domain data. See paper with datasets and models included 🔋: https://t.co/TheaeAssWm with @ASusanoPinto, @XiaohuaZhai and @neilhoulsby.

[Google AI @GoogleAI]

We’re pleased to release the Visual Task Adaptation Benchmark (VTAB), a diverse, realistic, and challenging protocol to measure progress towards universal visual representations. Learn all about it below. https://t.co/PbORwSFPAg

[André Susano Pinto @ASusanoPinto]

@lc0d3r We started with the longer one, but it looks by now #tfhub is winning. Shorter it is.

[André Susano Pinto @ASusanoPinto]

#TensorFlowHub helping fast experimentation and making ML models that go to space.

[Sergii 🇺🇦 @lc0d3r]

Amazing article showing how accessible #DeepLearning is becoming. Model trained with transfer learning and "#TensorFlow For Poets" codelab +#tfhub. Converted to #TFLite and now deployed on International
Space Station🚀 - TensorFlow Lite is Going to Space - https://t.co/Jb8ykVxtRN

[Andy Brock @ajmooch]

BigGAN-deep pretrained models are now publicly available for download on TFHub!

https://t.co/dZuYLwr8cJ

[TensorFlow @TensorFlow]

A new, multilingual version of the Universal Sentence Encoder (USE) model is now available on #TFHub!

Check it out here → https://t.co/N1JzuuX4MR

[Google DeepMind @GoogleDeepMind]

The BigGAN generators from our paper https://t.co/QUYlE9IBsE are now available on TF Hub (https://t.co/GHM9pIgQPw). Try the Colab demo at: https://t.co/Ynyb9T9AAD

[ACM-W womENcourage @ACMwomENcourage]

Enjoying the Workshop by Google engineer Elizabeth Kemp: Transfer Learning with TensorFlow Hub 👩‍💻
#womencourage18 #ML #TensorFlow

[André Susano Pinto @ASusanoPinto]

Our team hopes the new frontend helps more people find and use cutting-edge research modules :) #TensorFlowHub #transferlearning

[TensorFlow @TensorFlow]

We are launching a new web experience for TensorFlow Hub! Check out https://t.co/T8COqipES0 and explore our modules, including some new additions like the FasterRCNN for object detection.

Learn more on the post ↓ https://t.co/Et5NjpoW8X

[André Susano Pinto @ASusanoPinto]

Great to have image embedding modules trained on datasets other than just ImageNet.

[TensorFlow @TensorFlow]

Winners of the @inaturalist Challenge 2017 released their model on #TensorflowHub showcasing advantages of transfer learning! #tfhub #transferlearning

Check it out here ↓ https://t.co/ELrBIBWJUn