Lena@gregtech.eu to Programmer Humor@programming.devEnglish · 2 days ago"Source code file"gregtech.euexternal-linkmessage-square233fedilinkarrow-up1849arrow-down110
arrow-up1839arrow-down1external-link"Source code file"gregtech.euLena@gregtech.eu to Programmer Humor@programming.devEnglish · 2 days agomessage-square233fedilink
minus-squareZetta@mander.xyzlinkfedilinkarrow-up2·edit-21 day agoPerfect, groks context limit is 256,000, and as we all know llm recal only gets better as context fills so you will get perfect slop that works amazingly /s and more info on quality drop with context increase here https://github.com/NVIDIA/RULER
minus-squarehperrin@lemmy.calinkfedilinkEnglisharrow-up1·1 day ago250,000 lines is way more than 250,000 tokens, so even that context is too small.
Perfect, groks context limit is 256,000, and as we all know llm recal only gets better as context fills so you will get perfect slop that works amazingly
/s and more info on quality drop with context increase here https://github.com/NVIDIA/RULER
250,000 lines is way more than 250,000 tokens, so even that context is too small.