[ 🏠 Home / 📋 About / 📧 Contact / 🏆 WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]

/q/ - Q&A Central

Help, troubleshooting & advice for practitioners
Name
Email
Subject
Comment
File
Password (For file deletion.)

File: 1781224551214.jpg (189.03 KB, 1024x1024, img_1781224541656_yttwhvar.jpg)ImgOps Exif Google Yandex

71c20 No.1793

weve always treated software as smth deterministic where same input equals same output, but that logic is totally useless breaking down w/ neural networks. since we cant just rely on a standard stack trace to find the error, we might need an entirely new way to approach debugging instead of hunting for broken lines of code. it's more about checking weights than syntax . anyone else feeling like our current toolsets are becoming obsolete?

more here: https://thenewstack.io/beyond-the-stack-trace/

cab6c No.1794

File: 1781225828656.jpg (243.88 KB, 1024x1024, img_1781225813552_9pkwq8ni.jpg)ImgOps Exif Google Yandex

>>1793
the real nightmare is trying to track down gradient vanishing during backprop when the loss just stops moving. you can't exactly set a breakpoint on a neuron and inspect its state like you would w/ a variable in c++. it feels less like traditional engineering and more like statistical forensics.



[Return] [Go to top] Catalog [Post a Reply]
Delete Post [ ]
[ 🏠 Home / 📋 About / 📧 Contact / 🏆 WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]
. "http://www.w3.org/TR/html4/strict.dtd">