Xnhyacinth/ResAdapt-Checkpoints
Video-Text-to-Text • 8B • Updated • 18
ResAdapt
ResAdapt checkpoint, multimodal training mixture, and video-QA evaluation data.
-
Xnhyacinth/Resadapt-Mix
Viewer • Updated • 374k • 188Note Training mixture and media archives. Start with data/train_lens.parquet and data/TSPO-10K/val.parquet; materialize media before training.
Xnhyacinth/FBench
Viewer • Updated • 1.9k • 58Note Video-QA evaluation tables: longvideobench/validation and videomme/test.