Amrith Setlur (@setlur_amrith) 's Twitter Profile
Amrith Setlur

@setlur_amrith

Phd Student at CMU.

ID: 1248012740423307266

linkhttps://ars22.github.io/ calendar_today08-04-2020 22:20:08

111 Tweet

692 Takipçi

214 Takip Edilen

Amrith Setlur (@setlur_amrith) 's Twitter Profile Photo
Amrith Setlur
@setlur_amrith

7 months ago

Scaling test-time compute is fine πŸ˜’ but are we making good use of it? πŸ€” We try to answer this question in our new work: arxiv.org/pdf/2503.07572 TLDR; πŸš€ *Optimizing* test-time compute = RL with dense (progress) rewards = minimizing regret over long CoT episodes 😲 🧡‡️

Scaling test-time compute is fine πŸ˜’ but are we making good use of it? πŸ€” We try to answer this question in our new work: arxiv.org/pdf/2503.07572 TLDR; πŸš€ *Optimizing* test-time compute = RL with dense (progress) rewards = minimizing regret over long CoT episodes 😲 🧡‡️