AdaptToken: Entropy-based Adaptive Token Selection for MLLM Long Video Understanding
Open paper
Benchmark
Model
Rank
Results
video-question-answering-on-mlvu
Qwen2.5-VL + AdaptToken
#14
Accuracy: 76.8
video-question-answering-on-lvbench
Qwen2.5-VL + AdaptToken
#24
Accuracy: 54.8
Rank counts only results with a code link.