New research paper, R1 type reasoning models can be drastically improved in quality
#19
by
krustik
- opened
This is reminder or who haven't heard yet.
Long story short: R1 not ideal, missing opportunities and thinking not deep enough. Changing logic algorithm can improve it.
Paper is on huggingface here - https://huggingface.co/papers/2501.18585
P.S.: in other news proposed new law project bill in USA which sentence for 20 years with $1 mln fines for using, promoting any digital products and esp AI originated from China or chinese nationals (same sanctions exactly like against Russia which working already).
Just curious, are u American?