Search papers, labs, and topics across Lattice.
ment of Mechanical and Energy Engineering , Southern University of Science and Technology , Shenzhen , 518055 , China
3
0
5
A novel Action Chunking Transformer policy enables smooth, continuous autonomous navigation in the colon by fusing overlapping predicted action chunks, overcoming temporal inconsistencies common in existing methods.
DiT-Pruning achieves unprecedented image quality retention in Diffusion Transformers, maintaining a CLIP score loss of only 0.001 at 50% sparsity.
LLMs can still be easily fooled by simple prompt rewrites because current adversarial training doesn't adequately cover the data distribution, but a new method using diffusion models closes this gap.
