Search papers, labs, and topics across Lattice.
Affiliation:
4
0
5
6
Achieve 4-bit quantization for diffusion LLMs without calibration data, slashing memory by 2.6x and inference by 2.2x while preserving accuracy.
Celty achieves up to 5.3x speedup in LLM inference by harnessing dual-sparsity, revolutionizing how we approach GPU kernel design for sparse workloads.
KronQ achieves a perplexity of 7.93 on LLaMA-3-70B with 2-bit quantization, outperforming existing methods that struggle to maintain quality.
OrbitQuant achieves state-of-the-art post-training quantization for diffusion transformers, enabling efficient image and video generation without the need for data-specific calibration.
