#1 paper of the day, week, and month for @gensynai on @huggingface. Just 2 years ago, advocating for decentralised training strategies was non-consensus! 👇
Gabriel P. Andrade
@gab_p_andrade
11 Sep 2025
Is the whole greater than the sum of its parts? In decentralized RL post-training the answer is YES. 🐸🐸🐸 Swarm sAmpling Policy Optimization (SAPO) 🐸🐸🐸

Sep 12, 2025 · 2:08 PM UTC

8
20
101
11,437
Sort replies: Relevant Recent Liked
Replying to @harrygrieve @gensynai @huggingface
👏
1
118
Replying to @harrygrieve @gensynai @huggingface
Impressive and you are just getting started
1
97
Replying to @harrygrieve @gensynai @huggingface
This is just the beginning, we believe in you...
1
131
Replying to @harrygrieve @gensynai @huggingface
Crazy how fast research adoption curves move in AI — what’s the next ‘non-consensus’ idea you think will break through?
1
67
Replying to @harrygrieve @gensynai @huggingface
build！
3
96
Replying to @harrygrieve @gensynai @huggingface
Two years ago it was fringe, now it’s becoming reality. Pretty cool
3
Replying to @harrygrieve @gensynai @huggingface
Look at us now, proving the skeptics wrong! Decentralized is the new black.
5
Replying to @harrygrieve @eaodonald @gensynai @huggingface
Her's my comment ,
41