VAM-Cross MimicVideo fused Video2World DiT
This public repository contains the Bridge-initialized Video2World DiT
after fusing the trainable-only LoRA from iteration 1060 of
v2w_widowx_target_bridgeinit_lora_r256. The fused checkpoint is
checkpoints/video_backbone/iter_000001060_fused.pt.
The Bridge backbone and LoRA adapter are no longer separate at load time. This is a fused DiT checkpoint, not a complete inference bundle: the compatible MimicVideo code, video tokenizer, and T5 encoder are still required.
Provenance
- MimicVideo commit:
e3355dbc93132b576c02f920a59b4fc18a4f5906 - Bridge checkpoint repository:
jonpai/mimic-video - Bridge checkpoint revision:
f28339034831e3c2374be075e622e1ff38ebe0f8 - Source dataset revision:
dreamdifferent/vam-cross-target-widowx250-native@44c28a3bada463f06845daee8a1dd87607d32956 - Cameras:
observation.images.corner_cam, observation.images.front_cam - Video rate: 5 Hz
The private training dataset is not included. Users must comply with the upstream MimicVideo, NVIDIA Cosmos, and base-checkpoint terms.