Search papers, labs, and topics across Lattice.
3
0
5
1
AMVL eliminates the critical train-inference mismatch in MLLMs, leading to substantial improvements in reasoning performance across multimodal tasks.
G2DP achieves a remarkable 7.2-point improvement in reactive driving performance by integrating dense environmental constraints directly into the planning process.
N-GRPO achieves superior performance in mathematical reasoning by mixing embeddings, leading to diverse yet semantically consistent solution paths.
