Hacker News: barbolo

New comment by barbolo in "A History of IDEs at Google"

barbolo — Wed, 13 May 2026 19:17:55 +0000

It was a sad surprise last week when we tried to upgrade the workspace AI plan for some of our team members to Ultra and it was gone. We're moving to Claude/Codex.

New vibe coding experience – Google AI Studio

barbolo — Thu, 19 Mar 2026 16:55:07 +0000

Article URL: https://twitter.com/GoogleAIStudio/status/2034654985850659149

Comments URL: https://news.ycombinator.com/item?id=47442421

Points: 6

# Comments: 0

New comment by barbolo in "DINOv3"

barbolo — Thu, 14 Aug 2025 22:31:42 +0000

That's awesome. DINOv2 was the best image embedder until now.

New comment by barbolo in "Breaking the 4Chan CAPTCHA"

barbolo — Sat, 30 Nov 2024 10:14:30 +0000

https://github.com/lwthiker/curl-impersonate

New comment by barbolo in "Vision Transformers Need Registers"

barbolo — Sun, 12 May 2024 00:39:09 +0000

I’ve been using DinoV2 for some months now. I’ve tried the models with 4 register tokens along with CLS + patch tokens. I’ve several embeddings (tokens) from previous model (no registers) which are part of my solution, so I didn’t adopt the newer “register” models because the CLS tokens are not aligned between 0 registers and 4 registers models. It would be nice if the CLS and patch tokens were somehow aligned between those models.

Code for EfficientNetV2 (Deep Learning model) released

barbolo — Sat, 22 May 2021 23:54:27 +0000

Article URL: https://github.com/google/automl/tree/master/efficientnetv2

Comments URL: https://news.ycombinator.com/item?id=27251294

Points: 2

# Comments: 0

New comment by barbolo in "Giant leak exposes data from almost all Brazilians"

barbolo — Mon, 25 Jan 2021 23:30:06 +0000

It's much worse than that. The leak contains dozens of datasets (relatives, addresses, jobs (+ linkedin), schools, vehicles, income, debts, pictures of faces, companies).

New comment by barbolo in "Giant leak exposes data from almost all Brazilians"

barbolo — Mon, 25 Jan 2021 23:25:51 +0000

This leak is much more harmful. If the data really comes from Serasa Experian, they have more accurate and structured data from people/companies/assets in Brazil than anyone else.

Detect Headless – Run several tests against your browser

barbolo — Mon, 24 Dec 2018 15:02:10 +0000

Article URL: https://github.com/infosimples/detect-headless

Comments URL: https://news.ycombinator.com/item?id=18752404

Points: 1

# Comments: 0

Google Cloud Functions with Python (beta)

barbolo — Wed, 25 Jul 2018 12:52:10 +0000

Article URL: https://cloud.google.com/functions/docs/concepts/python-runtime

Comments URL: https://news.ycombinator.com/item?id=17608834

Points: 5

# Comments: 0

ReCAPTCHA v3 Beta – The new way to stop bots

barbolo — Fri, 08 Jun 2018 13:09:04 +0000

Article URL: https://www.google.com/recaptcha/intro/v3beta.html

Comments URL: https://news.ycombinator.com/item?id=17264889

Points: 1

# Comments: 0

Semi-supervised image classification explained

barbolo — Tue, 19 Dec 2017 22:27:29 +0000

Article URL: https://thecuriousaicompany.com/mean-teacher/

Comments URL: https://news.ycombinator.com/item?id=15965018

Points: 2

# Comments: 0

New comment by barbolo in "Developer preview of TensorFlow Lite"

barbolo — Wed, 15 Nov 2017 08:53:01 +0000

Thanks, @rasmi. I have a feedback for you guys. The pricing for predictions inference in GCP is not very fair. If I deploy a small model (like a SqueezeNet or Mobilenet) I pay almost the same price of someone deploying large models (like Resnet or VGG). That’s why I’m deploying my models on serverless environments and paying about 5 dollars for 1 million inferences.

The pricing of GCP is: $0.10 per thousand predictions, plus $0.40 per hour. That’s more than 100 dollars for 1 million inferences.

New comment by barbolo in "Developer preview of TensorFlow Lite"

barbolo — Wed, 15 Nov 2017 08:41:40 +0000

Thanks for the answer. Currently I’m using AWS Lambda to deploy my TensorFlow models. But it’s pretty hard and hacky. I need to remove a considerable portion of the code base that is not needed for inference only routines. I do that so the code loads faster and to fit the deployment package size limit. If TensorFlow Lite is already a compressed code, then it may be much easier to deploy it to a serverless environment. I’ll be trying it in my next deployments.

New comment by barbolo in "Developer preview of TensorFlow Lite"

barbolo — Tue, 14 Nov 2017 22:23:16 +0000

Would that be a viable option to deploy TensorFlow models on serverless environments (Lambda, Functions)?

New comment by barbolo in "RCN is much more data efficient than traditional Deep Neural Networks"

barbolo — Tue, 31 Oct 2017 08:18:03 +0000

I’m reading over and over again since the last weekend. And I’m checking the code. And I’m still not understanding it.

New comment by barbolo in "Uncaptcha: Defeating Google's audio reCaptcha with 85% accuracy"

barbolo — Mon, 30 Oct 2017 19:11:50 +0000

Automating searches on a government website that decided to use reCAPTCHA just because it wants to look modern. There are dozens of them in Brazil for example.

RCN is much more data efficient than traditional Deep Neural Networks

barbolo — Mon, 30 Oct 2017 10:41:32 +0000

Article URL: https://www.vicarious.com/2017/10/26/common-sense-cortex-and-captcha/

Comments URL: https://news.ycombinator.com/item?id=15584086

Points: 235

# Comments: 46

New comment by barbolo in "Uncaptcha: Defeating Google's audio reCaptcha with 85% accuracy"

barbolo — Sun, 29 Oct 2017 14:37:51 +0000

Blocking audio means blocking visually impaired people from accessing websites with reCAPTCHA.

Hacking reCAPTCHA is not only for bad people. There are several use cases where solving reCAPTCHA automatically is needed.

New comment by barbolo in "Uncaptcha: Defeating Google's audio reCaptcha with 85% accuracy"

barbolo — Sat, 28 Oct 2017 20:25:32 +0000

Another interesting fact is that TensorFlow 1.4 supports native MFCC spectrogram tensors.