Zixun Huang

Alexupenn1123

5

·

AI & ML interests

None yet

Recent Activity

upvoted an article 3 days ago

Score Function Centering: Your Stale Rollout Does Not Have to Be Importance Sampled and Stabilize Your Async RL Training with Trainer-Generator KL

upvoted an article 3 days ago

The Most Intuitive Way to Understand Score Function Centering: Understand Trainer-Generator Mismatch in Async RL from the Score Function

upvoted a paper about 1 month ago

FlowBalance: Verifier-Grounded Self-Improvement from On-Policy Reasoning Experience

View all activity

Organizations

None yet

upvoted 2 articles 3 days ago

Article

Score Function Centering: Your Stale Rollout Does Not Have to Be Importance Sampled and Stabilize Your Async RL Training with Trainer-Generator KL

NormalUhr

•

6 days ago

• 2

Article

The Most Intuitive Way to Understand Score Function Centering: Understand Trainer-Generator Mismatch in Async RL from the Score Function

NormalUhr

•

7 days ago

• 9

upvoted a paper about 1 month ago

FlowBalance: Verifier-Grounded Self-Improvement from On-Policy Reasoning Experience

Paper • 2609.03241 • Published Sep 3 • 50

upvoted a paper 3 months ago

Recursive Synthesis for Long-Horizon Terminal Tasks

Paper • 2608.05466 • Published Aug 7 • 252

upvoted a collection 3 months ago

Recursive Synthesis for Long-Horizon Terminal Tasks

CC BY 4.0 datasets and SFT/RL checkpoints for recursive task synthesis; base-model and third-party terms still apply. • 6 items • Updated Aug 10 • 18