Dark | Light
[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

![WenSun1 Avatar](https://lunarcrush.com/gi/w:24/cr:twitter::824609918.png) Wen Sun [@WenSun1](/creator/twitter/WenSun1) on x XXX followers
Created: 2025-07-18 03:06:27 UTC

How can small LLMs match or even surpass frontier models like DeepSeek R1 and o3 Mini in math competition (AIME & HMMT) reasoning? Prior work seems to suggest that ideas like PRMs do not really work or scale well for long context reasoning.  @kaiwenw_ai will reveal how a novel value model (not the usual PRMs!) can be trained to enable massive search at inference time.


XXXXX engagements

![Engagements Line Chart](https://lunarcrush.com/gi/w:600/p:tweet::1946043864097169620/c:line.svg)

**Related Topics**
[lays](/topic/lays)
[inference](/topic/inference)
[o3](/topic/o3)
[r1](/topic/r1)
[wen](/topic/wen)

[Post Link](https://x.com/WenSun1/status/1946043864097169620)

[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

WenSun1 Avatar Wen Sun @WenSun1 on x XXX followers Created: 2025-07-18 03:06:27 UTC

How can small LLMs match or even surpass frontier models like DeepSeek R1 and o3 Mini in math competition (AIME & HMMT) reasoning? Prior work seems to suggest that ideas like PRMs do not really work or scale well for long context reasoning. @kaiwenw_ai will reveal how a novel value model (not the usual PRMs!) can be trained to enable massive search at inference time.

XXXXX engagements

Engagements Line Chart

Related Topics lays inference o3 r1 wen

Post Link

post/tweet::1946043864097169620
/post/tweet::1946043864097169620