fix upcast in slice attention (#1591)

* fix upcast in slice attention

* fix dtype

* add test

* fix test

src/diffusers/models/attention.py

@@ -649,9 +649,9 @@ class CrossAttention(nn.Module):
                 key_slice = key_slice.float()
 
             attn_slice = torch.baddbmm(
-                torch.empty(slice_size, query.shape[1], key.shape[1], dtype=query.dtype, device=query.device),
-                query[start_idx:end_idx],
-                key[start_idx:end_idx].transpose(-1, -2),
+                torch.empty(slice_size, query.shape[1], key.shape[1], dtype=query_slice.dtype, device=query.device),
+                query_slice,
+                key_slice.transpose(-1, -2),
                 beta=0,
                 alpha=self.scale,
             )

tests/pipelines/stable_diffusion_2/test_stable_diffusion_v_pred.py

@@ -265,6 +265,25 @@ class StableDiffusion2VPredictionPipelineIntegrationTests(unittest.TestCase):
         expected_slice = np.array([0.0567, 0.057, 0.0416, 0.0463, 0.0433, 0.06, 0.0517, 0.0526, 0.0866])
         assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
 
+    def test_stable_diffusion_v_pred_upcast_attention(self):
+        sd_pipe = StableDiffusionPipeline.from_pretrained(
+            "stabilityai/stable-diffusion-2-1", torch_dtype=torch.float16
+        )
+        sd_pipe = sd_pipe.to(torch_device)
+        sd_pipe.enable_attention_slicing()
+        sd_pipe.set_progress_bar_config(disable=None)
+
+        prompt = "A painting of a squirrel eating a burger"
+        generator = torch.Generator(device=torch_device).manual_seed(0)
+        output = sd_pipe([prompt], generator=generator, guidance_scale=7.5, num_inference_steps=20, output_type="np")
+
+        image = output.images
+        image_slice = image[0, 253:256, 253:256, -1]
+
+        assert image.shape == (1, 768, 768, 3)
+        expected_slice = np.array([0.0461, 0.0483, 0.0566, 0.0512, 0.0446, 0.0751, 0.0664, 0.0551, 0.0488])
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
+
     def test_stable_diffusion_v_pred_euler(self):
         scheduler = EulerDiscreteScheduler.from_pretrained("stabilityai/stable-diffusion-2", subfolder="scheduler")
         sd_pipe = StableDiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-2", scheduler=scheduler)