Search papers, labs, and topics across Lattice.
University of Electronic Science and Technology of China
3
0
6
2
DataFlex-RL, an evaluation platform for comparing choices under a common GRPO recipe, is introduced, finding that changing the data policy measurably changes the training process but does not produce a reproducible improvement over uniform training.
GraspLLM achieves unprecedented zero-shot generalization on Text-Attributed Graphs, outperforming existing methods by effectively merging graph structure with LLM semantics.
DataFlex makes data-centric LLM training dramatically easier, unifying disparate methods for data selection, mixing, and reweighting into a single, efficient, and reproducible framework.
