7 users. We’re blowing up haha
| Website | https://www.davederiso.com |
| https://www.linkedin.com/in/dmderiso | |
| Scholar | https://scholar.google.com/citations?user=FkGFk18AAAAJ&hl=en |
Dave Deriso
- 43 Followers
- 33 Following
- 45 Posts
AI Researcher @ BAIR, Prev: Computational & Mathematical Engineering @ Stanford, Neuroscience @ Oxford
Research: self-supervised learning and visualization of sequential data
Work: Taught Data Visualization @ Stanford (visualization book in progress), Built Vehicle Routing system @ Caviar (sold to square), Founded Bluue Co. (med device in FDA approvals process)
Hobbies: photography, algorithmic art, acrylic painting, music production with ableton
I made a thing! It’s an AR app that highlights Al generated content on the browser page. http://www.bladerunner.ai #gpt #ai #ar #ml
Media Helping MediaHow can editors distinguish between text generated by #AI and that created by humans? The open-source GLTR (Giant Language model Test Room) says its model can improve human detection of fake text from 54% to 72% without any prior training. GLTR has a live demo for testing text you have written. I tried it with the intro for the About page on Media Helping Media. #media #journalism #fakenews #artificialintelligence #editing
@bramus Hey Bramus! I’m working on an AI based extension for google chrome. I saw that you’re a dev advocate and I wanted to reach out to chat about how to make the project better.
John LeonardPyTorch admins warn of malicious dependency
Package maintainers for #opensource framework #PyTorch have said a malicious dependency imitating one of its own was available on a prominent code repository during the holiday season.
#infosec #sbom
https://www.computing.co.uk/news/4062224/pytorch-admins-warn-malicious-torchtriton-dependency
@rachelandrew Hey Rachel! I’m working on an AI based extension for google chrome called Bladerunner. It’s getting some media attention and it’s definitely good for users. Do you have any insight into the chrome extension approvals process?
3 hours lost in the bucket today realizing that SGD is not the right way to rigidly align two point clouds. In the end it was the SVD rotation matrices that worked. If linear algebra was a noir nightclub, the SVD would be a mobster lurking in the far back “so you come to me again for a favor”
What’s your favorite autoencoder model on #huggingface or #replicate?
John Conway