Haowei Lin

@AndyLin2001

Undergrad and PhD student @PKU1898 LLM Agent | AI4Science | Generative Models | Continual Learning. Intern @ Tencent Hy agent team

Beijing, China

linhaowei1.github.io

Joined May 2023

187 Following

374 Followers

Replies Reposts Media

Pinned
Haowei Lin
@AndyLin2001
Nov 2
Our thoughts for the next half of AutoResearch. With this blog, we hope to help usher in the era of AI4Science 3.0: 1.0: Scientific foundation models. (e.g., AlphaFold) 2.0: Agentic tool use. 3.0: Agentic tool improvement.
Tencent Hy
@TencentHunyuan
Nov 2
Introducing Hyra 2.0, the next evolution of Hunyuan Research Agent. 💡 From solving problems to building reusable tools, models, and workflows that accelerate future discoveries. Read our perspective on the next chapter of AutoResearch: hy.tencent.com/research/hyra2
1
Haowei Lin
@AndyLin2001
Oct 14
In Apr ’25, I discussed AI4AI as AI4Sci (“AI as science”) with my professor, using scaling law discovery (SLD) as an example. Now AI4AI is hot. SLDBench v2 is out: arxiv.org/abs/2610.16297… More on SLD: arxiv.org/pdf/2402.02314 arxiv.org/pdf/2507.21184 arxiv.org/pdf/2604.22753
arxiv.org
SLDBench v2: Active Scaling Law Discovery
Scaling laws predict large-model performance from small experiments, but finding a reliable law itself requires costly experiments. SLDBench tests whether language-model agents can discover...
Haowei Lin
@AndyLin2001
Sep 26
So happy to have this paper accepted!
Lin Shi
@LinShi592021
Sep 25
Harbor Adapters and Harbor-Index (previously named as harbor-mix) accepted at NeurIPS E&D! Congrats and special thanks to all of our contributors, advisors, and funding partners!
Haowei Lin
@AndyLin2001
Sep 7
Our Harbor Adapters & Harbor-Index paper (the first version) is finally out:
arxiv.org
Harbor Adapters and Harbor-Index: Infrastructure and a Curated...
Evaluating agents on the growing number of agentic benchmarks is challenging because they often require complex environments and agent integrations. We introduce Harbor Adapters, a unified...
Haowei Lin
@AndyLin2001
Aug 19
Some interesting math results just dropped... though not entirely surprising, given the current surge in math discoveries.
Tencent Hy
@TencentHunyuan
Aug 19
Hyra keeps pushing the frontier. Since our last math update, four more advances have landed: 📐 3D Blaschke–Lebesgue: raised the universal lower bound for constant-width bodies from 0.380799w³ → 0.411040w³ (now 97.9% of the conjectured Meissner optimum). ∫ Beurling–Ahlfors:

Haowei Lin

Haowei Lin’s posts