On-Policy Parameter Update Direction Underlies Generalization in LLM Post-Training
Paper • 2609.36659 • Published • 84
zhou
jack234dsaffsdf
AI & ML interests
None yet
Recent Activity
upvoted a paper 2 days ago
On-Policy Parameter Update Direction Underlies Generalization in LLM Post-Training upvoted a paper 5 days ago
Agent Priors-guided Policy LearningOrganizations
None yet
upvoted a paper 2 days ago
upvoted 2
papers 5 days ago
upvoted a paper 6 days ago
upvoted 2
papers 7 days ago
upvoted 3
papers 8 days ago
CoWindow Attention: Full Causal Coverage Is a Collective Property
Paper • 2609.32704 • Published • 67
How Far Are We from Removing the Visual Encoder? Scaling Laws for Encoder-Free Multimodal Pretraining
Paper • 2609.35457 • Published • 66
MassAlloc Attention: Let Attention Allocate Its Own Compute
Paper • 2609.32712 • Published • 75
upvoted 5
papers about 1 month ago
Iris: Climbing to the Search Frontier
Paper • 2609.04304 • Published • 73
Random Attention: Rethinking KV Cache Eviction for Efficient Reasoning
Paper • 2609.03430 • Published • 188
It Takes Two to Match: Co-Evolving Generative Retriever with Reinforcement Learning
Paper • 2609.00638 • Published • 66
SMELT: Scaling Laws for Compute-Matched MoE Looped Transformers
Paper • 2609.01343 • Published • 92
UrbanGround: From Local Perception to Spatial Agency in a Real-Scale City
Paper • 2608.27456 • Published • 88
liked a
Space about 1 month ago
upvoted 4
papers about 2 months ago
Co-RL: Unsupervised Reasoning Emerges from Diverse Cohort in Multi-agent RL
Paper • 2608.17253 • Published • 100
Agentic ESOpt: Fine-Tuning Long-Horizon LLM Agents with Minimal GPU Requirements
Paper • 2608.17310 • Published • 110
LLMRouter: Unified Infrastructure for Developing, Evaluating, and Deploying LLM Routers
Paper • 2608.06867 • Published • 114
How Can Rhetoric Reward-Hack AI Reviewers? Dissecting Rhetorical Sensitivity in AI-Based Peer Review
Paper • 2608.08975 • Published • 48
upvoted a paper 4 months ago