MCPcopy Create free account

hub / github.com/boundless-large-model/boundless-world-model / types & classes

Types & classes14 in github.com/boundless-large-model/boundless-world-model

↓ 2 callersClassLoadCobotAction
wan_video_action/data/operators.py:341
↓ 1 callersClassImageCropAndResize
wan_video_action/data/operators.py:162
↓ 1 callersClassLoadGIFChunk
wan_video_action/data/operators.py:117
↓ 1 callersClassLoadVideoChunk
wan_video_action/data/operators.py:71
↓ 1 callersClassResolvePromptEmbPath
wan_video_action/data/operators.py:63
↓ 1 callersClassRoboTwinUnifiedDataset
wan_video_action/data/unified_dataset.py:8
↓ 1 callersClassRouteByKeyExtension
Applies a given operator to a specific key in a dictionary. Args: key: The dictionary key containing the file path to route.
wan_video_action/data/operators.py:30
↓ 1 callersClassToAbsolutePathByKeyExtension
wan_video_action/data/operators.py:53
↓ 1 callersClassToVideoTensor
Convert loaded video frames to float tensor in (V, C, T, H, W), range [-1, 1]. This operator converts a list of PIL Images or list of lists (for
wan_video_action/data/operators.py:211
↓ 1 callersClassWanTrainingModule
scripts/train.py:13
↓ 1 callersClassWanVideoActionEncoder
wan_video_action/models/wan_video_action_encoder.py:6
↓ 1 callersClassWanVideoUnit_ActionEmbedder
wan_video_action/pipelines/wan_video_action.py:289
↓ 1 callersClassWanVideoUnit_ImageEmbedderFused
Encode the conditioning frame directly into latents for Wan2.2 TI2V.
wan_video_action/pipelines/wan_video_action.py:497
↓ 1 callersClassWanVideoUnit_InputVideoEmbedder
Input frame embedder aligned to target history-conditioning behavior: - For short input (<=1 or < num_frames), skip VAE-conditioned noise inj
wan_video_action/pipelines/wan_video_action.py:322