Search papers, labs, and topics across Lattice.
University of Science and Technology of China
2
0
6
Forget noisy, biased LLM evaluators: CDRRM distills preference insights into compact rubrics, letting a frozen judge model leapfrog fully fine-tuned baselines with just 3k training samples.
Agentic RL can now beat proprietary LLMs and torch.compile in the challenging domain of CUDA kernel generation, achieving up to 40% speedups on hard tasks.
