Migrate AyaVisionImagePixelInputs to TensorSchema for shape validation (#21622)

Signed-off-by: Benji Beck <benjibeck@meta.com>

vllm/model_executor/models/aya_vision.py

@@ -2,7 +2,7 @@
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 # Adapted from https://github.com/huggingface/transformers/tree/main/src/transformers/models/aya_vision
 from collections.abc import Iterable, Mapping, Sequence
-from typing import Literal, Optional, TypedDict, Union, cast
+from typing import Annotated, Literal, Optional, Union, cast
 
 import torch
 from torch import nn
@@ -29,6 +29,7 @@ from vllm.multimodal.processing import (BaseMultiModalProcessor,
                                         PromptUpdateDetails)
 from vllm.multimodal.profiling import BaseDummyInputsBuilder
 from vllm.sequence import IntermediateTensors
+from vllm.utils.tensor_schema import TensorSchema, TensorShape
 
 from .interfaces import MultiModalEmbeddings, SupportsMultiModal, SupportsPP
 from .siglip import SiglipVisionModel
@@ -37,18 +38,28 @@ from .utils import (AutoWeightsLoader, WeightsMapper, flatten_bn,
                     merge_multimodal_embeddings)
 
 
-class AyaVisionImagePixelInputs(TypedDict):
-    type: Literal["pixel_values"]
-    pixel_values: torch.Tensor
+class AyaVisionImagePixelInputs(TensorSchema):
     """
-    Shape: `(num_patches_total, num_channels, height, width)`
-
-    `num_patches_total` is the total number of patches over each image over each
-    prompt in the batch.
+    Dimensions:
+        - np: The total number of patches over each image over each prompt in
+              the batch
+        - c: Number of channels
+        - h: Height of each image patch
+        - w: Width of each image patch
+        - bn: Batch size * number of images
     """
 
-    num_patches: torch.Tensor
-    """Shape: `(batch_size * num_images)`"""
+    type: Literal["pixel_values"]
+
+    pixel_values: Annotated[
+        torch.Tensor,
+        TensorShape("np", 3, "h", "w"),
+    ]
+
+    num_patches: Annotated[
+        torch.Tensor,
+        TensorShape("bn"),
+    ]
 
 
 class AyaVisionMultiModalProjector(nn.Module):
@@ -383,21 +394,6 @@ class AyaVisionForConditionalGeneration(nn.Module, SupportsMultiModal,
             e.flatten(0, 2) for e in image_embeds.split(num_patches.tolist())
         ]
 
-    def _validate_pixel_values(self, data: torch.Tensor) -> torch.Tensor:
-        h = w = self.config.vision_config.image_size
-        expected_dims = (3, h, w)
-
-        def _validate_shape(d: torch.Tensor):
-            if d.shape != expected_dims:
-                raise ValueError(
-                    "The expected shape of pixel values per image per batch "
-                    f"is {expected_dims}. You supplied {tuple(d.shape)}.")
-
-        for d in data:
-            _validate_shape(d)
-
-        return data
-
     def _parse_and_validate_image_input(
             self, **kwargs: object) -> Optional[AyaVisionImagePixelInputs]:
         pixel_values = kwargs.pop("pixel_values", None)
@@ -405,22 +401,17 @@ class AyaVisionForConditionalGeneration(nn.Module, SupportsMultiModal,
         image_embeds = kwargs.pop("image_embeds", None)
         assert image_embeds is None, "Aya Vision does not support image_embeds."
 
-        if not isinstance(pixel_values, (torch.Tensor, list)):
-            raise ValueError("Incorrect type of pixel values. "
-                             f"Got type: {type(pixel_values)}")
-        if num_patches is not None and not isinstance(num_patches,
-                                                      (torch.Tensor, list)):
-            raise ValueError("Incorrect type of num_patches. "
-                             f"Got type: {type(num_patches)}")
-
-        pixel_values = flatten_bn(pixel_values, concat=True)
-        num_patches = flatten_bn(num_patches, concat=True)
+        if pixel_values is None:
+            return None
 
         return AyaVisionImagePixelInputs(
             type="pixel_values",
-            pixel_values=self._validate_pixel_values(pixel_values),
-            num_patches=num_patches,
-        )
+            pixel_values=flatten_bn(pixel_values, concat=True),
+            num_patches=flatten_bn(num_patches, concat=True),
+            resolve_bindings={
+                "h": self.config.vision_config.image_size,
+                "w": self.config.vision_config.image_size,
+            })
 
     def get_language_model(self) -> torch.nn.Module:
         return self.language_model