Search papers, labs, and topics across Lattice.
2
0
6
10
ActionCache can accelerate inference for VLA models by up to 34.43脳 without compromising task success rates, revolutionizing real-time robotic manipulation.
LLM inference gets a 3.4x speed boost on PIM architectures thanks to a novel quantization method that cuts down on memory bottlenecks by quantizing activations directly in memory.
