DaisyRose-27B (radixark.com)
woah 4 hours ago
charcircuit 3 hours ago
It already happened. https://github.com/karpathy/autoresearch
its_down_again 2 hours ago
dimes 2 hours ago
anon373839 2 hours ago
harry8 an hour ago
boardendgrain 14 minutes ago
0xWTF an hour ago
StilesCrisis an hour ago
jacquesm 4 hours ago
gardnr 3 hours ago
jacquesm 3 hours ago
nblintao 3 hours ago
jacquesm 3 hours ago
reissbaker 3 hours ago
jacquesm 3 hours ago
And no, that won't get you to AGI, that is a completely different goal.
banghua13 3 hours ago
jacquesm 3 hours ago
mik1998 2 hours ago
jacquesm an hour ago
pkoiralap 2 hours ago
https://huggingface.co/sabicap/sabi-noema
Also HF model here: https://huggingface.co/RadixArk/DaisyRose-27B
xlayn 5 hours ago
"18 days on 144 H200 GPUs" insane.. didn't thought so much hardware would be needed to train Qwen.
drdaeman 4 hours ago
smallerize 3 hours ago
drdaeman 3 hours ago
mips_avatar 2 hours ago
robertkarl 5 hours ago
XCSme 5 hours ago
octoberfranklin 4 hours ago
You might be able to shave that down using a different flavor of H200 (like PCIe) but it would take more hours, so benefits are unclear.
lumost 4 hours ago
octoberfranklin 4 hours ago
But seriously. If there were a cheaper and reliable way, I'm sure the authors would have mentioned it instead of H200 SXM modules. They're trying to brag about cheap training.
Maybe you wanted to use an opportunity to shill tranium.
phs318u 3 hours ago
This was unnecessary and against HN guidelines to engage in good faith.
octoberfranklin an hour ago
Once you realize people are doing it you see it everywhere.
mips_avatar 2 hours ago