DavidFerris

5 ms·

1.
by DavidFerris 8mo ago
The initial ABC dataset is from public Onshape files -- clearly some people had a reason to design a dog model parametrically!
2.
by DavidFerris 8mo ago
There's a pretty big bias for mechanical engineering components in the dataset- very few organic forms. It's one of the limitations we call out in the dataset card. There are a few though! Try "dog" or "cookie cutte
3.
by DavidFerris 8mo ago
This isn't meant to be a commercially useful search engine- just a demonstration. You'll only be able to search for terms that the VLM could directly discern. From the blog post: Our search demo proves that it works quite well. As
4.
We rendered and embedded one million CAD files(cad-search-three.vercel.app)
136 points by DavidFerris 8mo ago|49 comments
5.
by DavidFerris 8mo ago
We rendered the one million part ABC dataset from Deep Geometry, and open-sourced the data. We also built a fun demo with the following pipeline: CAD > render > caption > embed. Open-sourced dataset: https://huggingface.
6.
AI Realist Predictions for the Future(david-ferris.medium.com)
2 points by DavidFerris 2y ago|0 comments
7.
by DavidFerris 3y ago
Super cool! btw I love the name "Greptile" :)
8.
by DavidFerris 3y ago
Inspired by Ian Hickon's reflection on his 18 years at Google, I reflect on how Google got into their current slump- and what they can do to bounce back.
9.
Can Google Bounce Back?(david-ferris.medium.com)
2 points by DavidFerris 3y ago|1 comments
10.
by DavidFerris 3y ago
Interesting idea! One of the problems with any primary research (surveys included) is the delay in collecting responses, which can take hours to weeks depending on sample, IR, incentives, etc. This would solve that! It's not surprising
11.
YC vs. CDL(david-ferris.medium.com)
1 points by DavidFerris 3y ago|0 comments
12.
by DavidFerris 4y ago
Very cool! I constantly struggle trying to do things in spreadsheets that are easy in Python. But I/O makes it annoying to write one-off scripts for a 30 second op. This would solve that pain point! I would love a Google Sheets integra
13.
Taxis, Web3, and a $26,000 Fraud(david-ferris.medium.com)
3 points by DavidFerris 4y ago|1 comments
14.
by DavidFerris 4y ago
I wrote about a bad thing that happened to me recently and what the web3 people can learn from it.
15.
by DavidFerris 4y ago
> "However, the algorithm is only valuable for detecting scams accurately after they have been executed." Forgive me, but isn't it only useful to detect scams before they happen?
16.
by DavidFerris 6y ago
I've been using Superpowered for the past 2 weeks and can honestly say the experience is great. I don't live in the browser and the Mac calendar app is pretty awful, so this has been a definitely workflow improvement for me. I pro
17.
by DavidFerris 6y ago
A survey platform designed around voice and video. https://phonic.ai It turns out that a LOT of podcasts are recorded asynchronously since it's easier to schedule.
18.
by DavidFerris 6y ago
"I don't know if camera based computer vision by itself will ever be enough for full self driving." Rebuttal: If you had a 3D camera sitting in the driver's seat of a car and hooked up the control to a remote controller,