causal attention

4 papers

• CogVLA: Cognition-Aligned Vision-Language-Action Models via Instruction-Driven Routing & Sparsification
• LEDiT: Your Length-Extrapolatable Diffusion Transformer without Positional Encoding
• LoRATv2: Enabling Low-Cost Temporal Modeling in One-Stream Trackers
• Multiverse: Your Language Models Secretly Decide How to Parallelize and Merge Generation