The Thinking Leakage in Reinforcement Learning for LLMs
Paper • 2610.14270 • Published • 20
Li Nanbo
Obnan
AI & ML interests
sequence modelling, self-improving systems, and world models --- structural computation as the unifying theme.
Recent Activity
upvoted a paper about 4 hours ago
The Thinking Leakage in Reinforcement Learning for LLMs upvoted a paper about 4 hours ago
DiSCO: Defending text-to-image generation through distribution-guided contrastive prompt optimization upvoted a paper about 4 hours ago
Architect-Ant: Editable Automatic Furnishing of Architectural Floor PlansOrganizations
upvoted 4
papers about 4 hours ago
DiSCO: Defending text-to-image generation through distribution-guided contrastive prompt optimization
Paper • 2608.17067 • Published • 23
Architect-Ant: Editable Automatic Furnishing of Architectural Floor Plans
Paper • 2606.10953 • Published • 58
APEX-Harm: Evaluating Agent Safety in Shared Professional Workspaces
Paper • 2610.13623 • Published • 11
submitted a
paper to Daily Papers about 5 hours ago
upvoted 2
papers about 5 hours ago
authored 7
papers 1 day ago
Structural Prognostic Event Modeling for Multimodal Cancer Survival Analysis
Paper • 2512.01116 • Published • 1
Object-Centric Representation Learning with Generative Spatial-Temporal Factorization
Paper • 2111.05393 • Published • 1
History Replays Itself: Unleashing Long-Context Capability in Linear Attention
Paper • 2610.15715 • Published • 9
Learning Object-Centric Representations of Multi-Object Scenes from Multiple Views
Paper • 2111.07117 • Published
FACTS: A Factored State-Space Framework For World Modelling
Paper • 2410.20922 • Published • 1
Self-Improvements in Modern Agentic Systems: A Survey
Paper • 2607.13104 • Published • 32
Huxley-Gödel Machine: Human-Level Coding Agent Development by an Approximation of the Optimal Self-Improving Machine
Paper • 2510.21614 • Published • 22
upvoted 2
papers 1 day ago
upvoted a paper 3 months ago