Merge branch 'main' of https://huggingface.co/spaces/diffusers-internal-dev/world-engine-zero

app.py

@@ -113,6 +113,9 @@ def create_gpu_game_loop(command_queue: Queue):
         """
         pipe.to("cuda")
         pipe.blocks.sub_blocks['before_denoise'].sub_blocks['setup_kv_cache']._setup_kv_cache(pipe.transformer, pipe.device, torch.bfloat16)
+        pipe.transformer.apply_inference_patches()
+        #pipe.transformer.quantize("fp8")
+        #aoti_load_(pipe.transformer, "diffusers-internal-dev/world-engine-aot", "transformer-fp8.pt2")
         aoti_load_(pipe.transformer, "diffusers-internal-dev/world-engine-aot", "transformer-inference-patch.pt2")
         
         pipe.vae.bake_weight_norm()