Rafael Barbolo

[Rafael Barbolo @barbolo]

@GoogleResearch @grok does it work with any CGM data or is it something bound to a specific hardware?

[Rafael Barbolo @barbolo]

@leerob Do cursor plans have higher tokens/dollar than supergrok plans? I love both, but will choose a single one to keep.

[Rafael Barbolo @barbolo]

@tibor_tee @tibor_tee @poteto please, check DM + form. Thank you.

[Rafael Barbolo @barbolo]

@oalanicolas Cruza os dados dos dois links (depende do que você quer fazer com o modelo):
- https://t.co/OHNPGGYT3E
- https://t.co/4P47WmObzs

[Federico Baldassarre @BaldassarreFe]

Say hello to DINOv3 🦖🦖🦖

A major release that raises the bar of self-supervised vision foundation models.
With stunning high-resolution dense features, it’s a game-changer for vision tasks!

We scaled model size and training data, but here's what makes it special 👇

[TensorFlow @TensorFlow]

💃 Strike a pose with #MoveNet!

The next-generation human pose model is now available as #TFLite versions on #TFHub, ready for fast on-device inference!

Check it out → https://t.co/zSuioI1DSO

[Rafael Barbolo @barbolo]

@diraol @pait Aí volta pro sistema ruim que tinha antes ou investe na nuvem USP 🤷‍♂️

[Rafael Barbolo @barbolo]

@pekayatt É possível que o aumento de transações cripto gere muita inovação em energia. E colocar culpa de floresta queimando em cripto é sacanagem.

[Rafael Barbolo @barbolo]

@Suhail I don’t think IPs from different regions will make any difference. Having traffic distributed through many IPs might help. reCAPTCHA is a very broken piece of software IMO. You should probably talk to Google if Mighty is having low scores.

[Mingxing Tan @tanmingxing]

Happy to introduce EfficientNetV2: Smaller Models and Faster Training

Achieved faster training and inference speed, AND also with better parameters efficiency.

Arxiv: https://t.co/YHWEb8pHmR

Thread 1/4

[AI at Meta @AIatMeta]

We're introducing an optimizer for deep learning, MADGRAD. This method matches or exceeds the performance of the Adam optimizer across a varied set of realistic large-scale deep learning training problems. https://t.co/6NGmJzHlZp

[Rafael Barbolo @barbolo]

@josevalim I hope you don’t mind if I keep saying it’s a Brazilian language. 😜 These days we have so little to be proud of.

[Rafael Barbolo @barbolo]

@josevalim @josevalim do you consider Elixir a Brazilian or Polish language?

[Soham De @sohamde_]

Releasing NFNets: SOTA on ImageNet. Without normalization layers!

https://t.co/eUPSNle0r2
Code: https://t.co/YuwULNqpTR

This is the third paper in a series that began by studying the benefits of BatchNorm and ended by designing highly performant networks w/o it.

A thread:

1/8

[Yann LeCun @ylecun]

Facebook AI & NYU Langone Health are open-sourcing a system that can predict if a Covid patient will require intensive care, from chest x-rays, up to four days in the future.

Paper: https://t.co/zA1Ig3i6WA

Github: https://t.co/mSe18J1lLk

Blog post: https://t.co/5yl2KJ1QHp

[Rafael Barbolo @barbolo]

@fernandoulrich Servidores locais em 2021 não faz o menor sentido. Não apenas pelos custos serem imensamente maiores, você precisaria de diversos datacenters para ter o mesmo nível de disponibilidade que cloud oferece. Existem ótimas opções cloud de empresas americanas, chinesas e europeias.

[Andrej Karpathy @karpathy]

Impressive and surprising https://t.co/9QpeWbhQFU I use those words sparingly

[Rafael Barbolo @barbolo]

@lucaslago Não costumo fazer muitos loops dentro de loops. E costumo usar iteradores sobre objetos. Dito isso, acho que uma opção boa é usar um nome de variável mais semântico (index, count, max, position, cell, row, column).

[Rafael Barbolo @barbolo]

@andrefatala Onde é esse lugar?