Views
No views yet
VAM_Learn_from_Human_Video for the RobotWin paired v3 all-tasks multi-view motion VAM-ICL run.step-100000.safetensors1source: /cpfs/huangsq/VAM_Learn_from_Human_Video/src/vam/models/train/vam_icl/paired_v3_alltasks_mv_mot_ti2v5b_16g_100k_v2_refdrop50
2checkpoint: step-100000.safetensors
3reference dropout: 0.5
4steps: 100000
5backbone: Wan2.2-TI2V-5B