Scaling LLM Test-Time Compute Optimally Can Be More Effective than Scaling Model Parameters
Snell et al., 2025, Scaling LLM Test-Time Compute Optimally Can Be More Effective than Scaling Model Parameters (The Thirteenth International Conference on Learning Representations (ICLR 2025))
Subscriber deck — free during early access
1 cards in “5.2 Search Methods Against a PRM” · whole deck
5.2 Search Methods Against a PRM
- Compare best-of-N, beam search and lookahead search against a PRM.Best-of-N samples whole solutions and picks the best. Beam search samples candidates step by step and keeps the top ones by PRM score. Lookahead search scores each step by rolling out \(k\) more steps first, a deterministic form of MCTS, and costs \(N(k+1)\) samples.
- Snell et al., 2025, Scaling LLM Test-Time Compute Optimally Can Be More Effective than Scaling Model Parameters (The Thirteenth International Conference on Learning Representations (ICLR 2025)), §5.2
Source
- Charlie Snell, Jaehoon Lee, Kelvin Xu and Aviral Kumar. 2025. Scaling LLM Test-Time Compute Optimally Can Be More Effective than Scaling Model Parameters. In The Thirteenth International Conference on Learning Representations (ICLR 2025). https://arxiv.org/abs/2408.03314