iiinova — where your digital self keeps evolving

Shangshang Wang

@shangshang-wang.bsky.social

shangshang-wang.github.io Phd student in CS + AI @usc.edu. CS undergrad, master at ShanghaiTech. LLM reasoning, RL, AI4Science.

Posts Replies Media

Shangshang Wang @shangshang-wang.bsky.social · 12/06/2025

Sparse autoencoders (SAEs) can be used to elicit strong reasoning abilities with remarkable efficiency. Using only 1 hour of training at $2 cost without any reasoning traces, we find a way to train 1.5B models via SAEs to score 43.33% Pass@1 on AIME24 and 90% Pass@1 on AMC23.

110

Shangshang Wang @shangshang-wang.bsky.social · 23/04/2025

😃 Want strong LLM reasoning without breaking the bank? We explored just how cost-effectively RL can enhance reasoning using LoRA! [1/9] Introducing Tina: A family of tiny reasoning models with strong performance at low cost, providing an accessible testbed for RL reasoning. 🧵

183

Shangshang Wang @shangshang-wang.bsky.social · 19/02/2025

🔍 Diving deep into LLM reasoning? From OpenAI's o-series to DeepSeek R1, from post-training to test-time compute — we break it down into structured spreadsheets. 🧵

152

Reposted by Shangshang Wang

Ollie Liu @oliu-io.bsky.social · 06/01/2025

Introducing METAGENE-1🧬, an open-source 7B-parameter metagenomics foundation model pretrained on 1.5 trillion base pairs. Built for pandemic monitoring, pathogen detection, and biosurveillance, with SOTA results across many genomics tasks. 🧵1/

2266