Search papers, labs, and topics across Lattice.
Shanghai Jiaotong University
3
0
5
Reconstruct a high-fidelity, full-head 3D avatar from a single image in under one second, finally breaking the quality-speed tradeoff.
Current research agent benchmarks miss critical flaws, as MiroEval reveals that process quality is a reliable predictor of research outcome, and multimodal tasks expose weaknesses invisible to output-level metrics.
Current VLMs can ace image quizzes, but completely fumble when asked to stack blocks in a physically plausible way, revealing a critical gap in understanding real-world physics.
