Datasets:

Wiselnn
/

VideoRoPE

Tasks:

Video-Text-to-Text

ArXiv:

License:

Dataset card Files Files and versions

VideoRoPE / vision_niah_d /niah_output /Qwen2-VL-m_rope-128frames-16card_8k-context-330k-llava-video

50.6 kB

2 contributors

History: 1 commit

Wiselnn's picture

Upload folder using huggingface_hub (#1)

4eedc1a verified 7 months ago

all_accuracies.json

8.16 kB

Upload folder using huggingface_hub (#1) 7 months ago
avg_accuracy.txt

37 Bytes

Upload folder using huggingface_hub (#1) 7 months ago
heatmap.png

42.4 kB
xet

Upload folder using huggingface_hub (#1) 7 months ago