Submitted by Corrado Rainone 11 When Cloud Agents Meet Device Agents: Lessons from Hybrid Multi-Agent Systems Qualcomm 3
Submitted by Fabio Valerio Massoli 30 Memory-Efficient Looped Transformer: Decoupling Compute from Memory in Looped Language Models Qualcomm 3
Submitted by Raghavv Goel 10 Efficient Training-Free Multi-Token Prediction via Embedding-Space Probing Qualcomm 2
Submitted by Fabio Valerio Massoli 23 Reasoning as Compression: Unifying Budget Forcing via the Conditional Information Bottleneck Qualcomm 2
Submitted by Raghavv Goel 4 Skip to the Good Part: Representation Structure & Inference-Time Layer Skipping in Diffusion vs. Autoregressive LLMs Qualcomm 2
Submitted by Amir Habibian 4 PyramidalWan: On Making Pretrained Video Model Pyramidal for Efficient Inference Qualcomm 3
Submitted by Amir Habibian 5 ReHyAt: Recurrent Hybrid Attention for Video Diffusion Transformers Qualcomm 6