data-optimization

PROOF-Gen presents a per-scenario reflective optimization method to recover failed trajectories for distilling tool-calling capabilities, enhancing data quality and model performance in distillation pipelines.

PaddleOCR-VL-1.6 improves document parsing by identifying and refining under-optimized regions via targeted data optimization and progressive post-training, achieving state-of-the-art 96.33% on OmniDocBench v1.6.