Rethinking Shrinkage Bias in LLM FP4 Pretraining: Geometric Origin, Systemic Impact, and UFP4 Recipe Paper • 2606.20381 • Published 8 days ago • 9
The Shape of Addition: Geometric Structures of Arithmetic in Large Language Models Paper • 2606.03645 • Published 28 days ago • 5
LongLive-RAG: A General Retrieval-Augmented Framework for Long Video Generation Paper • 2606.02553 • Published 25 days ago • 19
LLMs as Noisy Channels: A Shannon Perspective on Model Capacity and Scaling Laws Paper • 2605.23901 • Published May 22 • 13
BitCPM-CANN Collection Full-pipeline ternary quantized model trained on CANN. • 12 items • Updated May 24 • 28
Mix-Quant: Quantized Prefilling, Precise Decoding for Agentic LLMs Paper • 2605.20315 • Published May 19 • 28