Yicheng Liu

Eagle 2.5: boosting long-context post-training for frontier vision-language models

CG-Bench: Clue-grounded Question Answering Benchmark for Long Video Understanding