acceptodds: the papers of ICLR 2027
Abstract
The safe for work edition of acceptodds: the submissions, without the markets. Search them, browse them on a map laid out by what they are about, follow the ones you care about, keep reading lists with your group, and get recommendations from what you read. There are no odds, no bets and no reputation here. acceptodds is not affiliated with any organisation or institution. acceptodds is an open source project, open to contributions, at github.com/PotentialHedge/acceptodds-sfw.1 Follow for updates on X.
[PDF]CARE: Cross-Domain Alignment by Reweighting Edges for Multi-View Graph Domain Adaptation▲6.5k3.33
[PDF]Remembering How to Act through Greedy Action Consolidation in Continual Reinforcement Learning▲1.7k3.33
[PDF]From Representation to Readout: Mechanistic Analysis of Membership Signals in Language Models▲6.5k3.33
[PDF]Muon with Finite Newton–Schulz: The Smoothing Benefit in Nonsmooth Nonconvex Optimization▼2373.33
[PDF]Refinement Is Inherently Editable: Training-Free Prompt-to-Prompt Image Editing with Generative Refinement Network▲16k3.33
[PDF]SpeakingRobot: Semantic-Event-Driven Robot-Space Co-Speech Generation with Lower-Body Coordination▲1.7k3.33
[PDF]Transformers Can Implement Preconditioned Richardson Iteration for In-Context Gaussian Kernel Regression▲1.7k3.33
[PDF]Thresholds, Not Normalization: A Homeostatic Spiking Layer at Parity Without the Overhead▲6.5k3.33
[PDF]Bandit Submodular Maximization under Matroid Constraints: Learning Compressed Exchange Policy▲233.33
[PDF]Learning from What Users Do Not Say: Self-Evolving Assistants by Internalizing Latent Thoughts▲1.7k3.33
[PDF]SPARK: Adaptive Parametric Knowledge Participation for Faithful Retrieval-Augmented Generation▲233.33
[PDF]FocusDreamer: Task-Directed Model-Based Reinforcement Learning Guided by Return Relevance▲6.5k3.33
[PDF]EviCon: How Evidence Quality Shapes Beneficial and Harmful Conformity in LLM-Based Multi-Agent Systems▲6.5k3.33
1 If this project is interesting to you, reach out at [email protected].