Skip the Compressor: Coding Agents Fix More Issues When They Pick Whole Files
Paper • 2610.21073 • Published • 3
AI & ML interests
Building interpretable, auditable and trustworthy AI models.
Recent Activity
View all activity
Papers
Skip the Compressor: Coding Agents Fix More Issues When They Pick Whole Files
Resisting Harmful Instructions through Training-Free Weight Editing
xiaocong01
submitted 3
papers to Daily Papers about 20 hours ago
xiaocong01
submitted a
paper to Daily Papers about 21 hours ago
xiaocong01
authored 8
papers 1 day ago
Generative Interpretability via Scalable Neuro-Symbolic Models
Paper • 2609.13529 • Published
DAGFormer: Global Topology Prediction for Transformer Cross-Layer Connections
Paper • 2610.19625 • Published • 5
Sparse Autoencoders Are a Shared Coordinate System for Symbolic Matching
Paper • 2610.19636 • Published • 4
Learning by Analogy: Enhancing Few-Shot Prompting for Math Word Problem Solving with Computational Graph-Based Retrieval
Paper • 2411.16454 • Published
Parameter-Efficient Tuning with Special Token Adaptation
Paper • 2210.04382 • Published • 2
NLP From Scratch Without Large-Scale Pretraining: A Simple and Efficient Framework
Paper • 2111.04130 • Published • 1
EVA: An Open-Domain Chinese Dialogue System with Large-Scale Generative Pre-Training
Paper • 2108.01547 • Published
Cascade Speculative Drafting for Even Faster LLM Inference
Paper • 2312.11462 • Published • 10