@salltweets Might it be beneficial to take this argument a step further?
Is the govt saying that the employer (and the police?) are unable to determine whether someone has childbearing potential. Maybe only judges and the govt can do that? 🤡
@measure_plan I trained a computer vision model to classify and edit beach volleyball footage using your iPhone’s GPU
MN3 for detection and TCN for temporal smoothing.
Runs at 20x watch speed on 1080p footage
App Store: Cutshot
@elonmusk@grok I still choose Gemini over grok for coding because when it comes to implementing changes to my files, Gemini is smart enough, but dumb enough to still rewrite an entire file when I want it to. Grok can not be convinced to rewrite an entire replacement file for my convenience
@fr0st_e I love the transition of detail, where you see branches, then trees at all different scales and angles. Same with the interaction of the clouds with the mountains. If anything I want to see more pixel art 3d weather
@_360Genius_ Surely every ‘thing’ is 3d, but has a sort of post-processing so that it is simplified to the right ‘billboard’ for your distance and angle.
@burkov@tibo_maker I should make a cursor extension for this with checkboxes on each file and it goes to your clipboard, but surely someone already did this?
@levelsio I’ve been playing around with my prototypes in this space for ages. Right now I’m building it directly in MPS Graph as a swift package using Grok3 and Cursor.
PyTorch and Keras just didn’t work for me to train a model on MPS.
Otherwise Yolov10 exports to CoreML pretty well