CUDA became the de facto framework for ML compute because nothing else came close to its ease of use and maturity. I wish AMD could muster similar strength because competition is sorely needed in the space.

https://www.nytimes.com/2023/08/21/technology/nvidia-ai-chips-gpu.html?unlocked_article_code=eP2wLDddF2qWM74gKImwSVSE9p_1A6-Obct1xF56ciPvG4ztVV2b3wX5eJ01YyAjl1-XY732tQ41r-CWKKyK3gFEY4pg7W0oi6dv9M3Mkc9a9wYOA0y33pZC50wuvcnZf5XB3SpKukjT4gKhFvFjj1FJfp0v7b1vz_zv0yFInC3RdWkqd3mXrDUW7VG8E_-pjvZxa7vb_nH5bL7ZpAibSIZGCv0AJuw8r5ea_m6eJ2q4RBT9-hN38jwqGSVY7CaLXO0kOMJuoCFm2YVutKPO7ysayASs-Zn0WEbl8f74IweunLMTvxd5pp6DQZApGK6RkPJk5UuKlGlcdoy9Bd-Apw

Any reason they need 2052 bits of entropy in their verification code?

I hate unnecessarily long urls