LLaVA-NeXT: fix anyres shapes (#32314)

fix

src/transformers/models/llava_next/modeling_llava_next.py

@@ -660,7 +660,7 @@ class LlavaNextForConditionalGeneration(LlavaNextPreTrainedModel):
                 height = width = self.config.vision_config.image_size // self.config.vision_config.patch_size
                 if height * width != base_image_feature.shape[0]:
                     raise ValueError("The number of patches is not consistent with the image size.")
-                num_patch_width, num_patch_height = get_anyres_image_grid_shape(
+                num_patch_height, num_patch_width = get_anyres_image_grid_shape(
                     image_sizes[image_idx],
                     self.config.image_grid_pinpoints,
                     self.config.vision_config.image_size,

src/transformers/models/llava_next_video/modeling_llava_next_video.py

@@ -704,7 +704,7 @@ class LlavaNextVideoForConditionalGeneration(LlavaNextVideoPreTrainedModel):
                 height = width = self.config.vision_config.image_size // self.config.vision_config.patch_size
                 if height * width != base_image_feature.shape[0]:
                     raise ValueError("The number of patches is not consistent with the image size.")
-                num_patch_width, num_patch_height = get_anyres_image_grid_shape(
+                num_patch_height, num_patch_width = get_anyres_image_grid_shape(
                     image_sizes[image_idx],
                     self.config.image_grid_pinpoints,
                     self.config.vision_config.image_size,