Pinned Tweet
Doing lesson 15 of the @fastdotai course; deducing how to rearrange convolutions as a matrix product
1
1
7
13,625
Salman // 萨尔曼 retweeted
NVIDIA released Foundation4D. Give it multi-camera driving video and a single forward pass returns depth, camera poses, 3D boxes, semantic occupancy and Gaussians you can render. One shared backbone, a decoder per task.
5
59
562
22,999
Early exits
Exploring generating routes for maps, and I'm gradually whipping together an animation to help me understand BFS better, inspired from @johnowhitaker visualization in SolveIt lesson 8
0
0
0
21
Salman // 萨尔曼 retweeted
A recent UX pattern I've started using, I don't know if I got this from someone else, is I ask AI to allow me to highlight text in an HTML such that the highlight remains and it gets saved to a JSON file so that I can communicate with AI about the things I highlighted. I've found easier and quicker for me to get through a lot of text or data during AI-review. I haven't looked at actual numbers, but I would assume it also saves tokens.
0
2
5
279
Exploring generating routes for maps, and I'm gradually whipping together an animation to help me understand BFS better, inspired from @johnowhitaker visualization in SolveIt lesson 8
0
0
1
53
Homogenouity in tensors means every dimension is of the same length.
Homogenous: [[1,2],[3,4]]
Nonhomogenous: [1,[2,3]]
Homogenous: [[1,2],[3,4]]
Nonhomogenous: [1,[2,3]]
0
0
0
26
I don't have to use the modulo operator. I can pass an argument to the step parameter in range
1
0
0
1
Observing the output, I see that to have a centered triangle, odd rows can only be output. I can use the modulo operator for this
1
0
0
4
Here's my deduction for printing a centered triangle in Python. Do you know of a cooler way?
🧵/
🧵/
1
0
0
28



















