Update app.py

app.py

@@ -1,12 +1,15 @@
 import spaces
 import torch
 from diffusers import HunyuanVideoPipeline, HunyuanVideoTransformer3DModel
+from para_attn.first_block_cache.diffusers_adapters import apply_cache_on_pipe
 from PIL import Image
 import numpy as np
 import gradio as gr
 import os
 
 
+device = "cuda" if torch.cuda.is_available() else "cpu"
+
 # model_id = "hunyuanvideo-community/HunyuanVideo"
 
 model_id = "FastVideo/FastHunyuan-diffusers"
@@ -26,7 +29,11 @@ def generate(prompt, width=832, height=832, num_inference_steps=30, lora_id=None
     if lora_id and lora_id.strip() != "":
         pipe.unload_lora_weights()
         pipe.load_lora_weights(lora_id.strip())
-    pipe.to("cuda")
+    apply_cache_on_pipe(
+        pipe,
+        # residual_diff_threshold=0.2,
+    )
+    # pipe.to("cuda")
     torch.cuda.empty_cache()
     try:
         output = pipe(