Search papers, labs, and topics across Lattice.
Xiamen University
3
0
6
LLMs can compress context better than dedicated compression modules, simply by prompting them to "think" about the task.
Forget O(N^5) scaling: this new tensor decomposition approach slashes the computational cost of second-order perturbation theory down to O(N^3) without sacrificing accuracy.
By reflecting on its own reasoning, ReflectRM achieves a +10.2 improvement in mitigating positional bias compared to leading generative reward models, making it a far more stable evaluator.
