Search papers, labs, and topics across Lattice.
Affiliation:
3
0
6
4
Despite executing target instructions, LLMs often fail to deliver competitive performance in GPU kernel optimization, especially on complex tasks.
Robustness, a critical but often overlooked property of program analyses, can be formally defined and achieved using a categorical framework.
Open-source LLMs can now autonomously optimize AI accelerator kernels, matching the performance of proprietary models at a fraction of the cost.
