Skip to content

vllm.model_executor.models.kimi_vl

Classes:

KimiVLImagePixelInputs

Bases: TensorSchema

Dimensions: - nc: Number of channels - np: Number of patches - ps: Patch size - ni: Number of images

Source code in vllm/model_executor/models/kimi_vl.py 
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
class KimiVLImagePixelInputs(TensorSchema):
    """Dimensions:
    - nc: Number of channels
    - np: Number of patches
    - ps: Patch size
    - ni: Number of images
    """

    type: Literal["pixel_values"] = "pixel_values"

    pixel_values: Annotated[
        torch.Tensor | list[torch.Tensor],
        TensorShape("np", 3, "ps", "ps"),
    ]

    image_grid_hws: Annotated[torch.Tensor, TensorShape("ni", 2)]