Upload src/pipeline.py with huggingface_hub

src/pipeline.py

@@ -42,14 +42,13 @@ def load_pipeline() -> Pipeline:
         transformer=transformer,
         local_files_only=True,
         torch_dtype=torch.bfloat16,
-    )
+    ).to("cuda")
 
     pipeline.transformer.to(memory_format=torch.channels_last)
-    pipeline.vae.to(memory_format=torch.channels_last)
+    # pipeline.vae.to(memory_format=torch.channels_last)
     # quantize_(pipeline.vae, int8_weight_only())
     pipeline.vae = torch.compile(pipeline.vae, fullgraph=True, mode="max-autotune")
-    pipeline.to("cuda")
-    
+      
     for _ in range(2):
         pipeline("cat", num_inference_steps=4)