Less is More: Recursive Reasoning with Tiny Networks
Paper • 2510.04871 • Published • 513
read
-
-
Cache-to-Cache: Direct Semantic Communication Between Large Language Models
Paper • 2510.03215 • Published • 99 -
When Thoughts Meet Facts: Reusable Reasoning for Long-Context LMs
Paper • 2510.07499 • Published • 49 -
StreamingVLM: Real-Time Understanding for Infinite Video Streams
Paper • 2510.09608 • Published • 52 -
LiteStage: Latency-aware Layer Skipping for Multi-stage Reasoning
Paper • 2510.14211 • Published • 9 -
Every Attention Matters: An Efficient Hybrid Architecture for Long-Context Reasoning
Paper • 2510.19338 • Published • 117 -
LightMem: Lightweight and Efficient Memory-Augmented Generation
Paper • 2510.18866 • Published • 115 -
Glyph: Scaling Context Windows via Visual-Text Compression
Paper • 2510.17800 • Published • 69 -
DeepSeek-OCR: Contexts Optical Compression
Paper • 2510.18234 • Published • 93 -
Deep Self-Evolving Reasoning
Paper • 2510.17498 • Published • 12 -
Latent Sketchpad: Sketching Visual Thoughts to Elicit Multimodal Reasoning in MLLMs
Paper • 2510.24514 • Published • 22 -
The End of Manual Decoding: Towards Truly End-to-End Language Models
Paper • 2510.26697 • Published • 119 -
Exploring Conditions for Diffusion models in Robotic Control
Paper • 2510.15510 • Published • 40 -
Kimi Linear: An Expressive, Efficient Attention Architecture
Paper • 2510.26692 • Published • 132 -
Delta Attention: Fast and Accurate Sparse Attention Inference by Delta Correction
Paper • 2505.11254 • Published • 49