1 3 5

Yihua Zhang

NormalUhr

https://www.yihua-zhang.com

AI & ML interests

None yet

Recent Activity

published an article 26 days ago

A Role Shift for AI Infra: From Foundational Support to a Core Engine of Innovation

liked a model 30 days ago

deepseek-ai/DeepSeek-V3.2-Exp

commented on their article 3 months ago

MLA: Redefining KV-Cache Through Low-Rank Projections and On-Demand Decompression

View all activity

Organizations

published an article 26 days ago

Article

A Role Shift for AI Infra: From Foundational Support to a Core Engine of Innovation

•

26 days ago

liked a model 30 days ago

deepseek-ai/DeepSeek-V3.2-Exp

Text Generation • 685B • Updated 20 days ago • 102k • • 751

commented on MLA: Redefining KV-Cache Through Low-Rank Projections and On-Demand Decompression 3 months ago

The issues have been fixed. Thanks for letting me know. This is because the math engine in HuggingFace is different from the one we usually use. It uses "\(" and "\)" to wrap in-line equations, rather than "(" and ")". Obviously, I was not aware of that before.

published an article 3 months ago

Article

Re-understanding KL Approximation from an RL-for-LLM Lens: Notes on “Approximating KL Divergence”

•

Aug 11

published an article 3 months ago

Article

From GRPO to DAPO and GSPO: What, Why, and How

•

Aug 9

• 47

liked a model 3 months ago

openai/gpt-oss-20b

Text Generation • 22B • Updated Aug 26 • 4.77M • • 3.82k

upvoted a paper 3 months ago

Understanding R1-Zero-Like Training: A Critical Perspective

Paper • 2503.20783 • Published Mar 26 • 56

published an article 5 months ago

Article

Decorators in Machine Learning

•

Jun 8

upvoted a paper 5 months ago

Reinforcing Multi-Turn Reasoning in LLM Agents via Turn-Level Credit Assignment

Paper • 2505.11821 • Published May 17 • 14

published an article 8 months ago

Article