Update app.py

app.py

@@ -195,7 +195,7 @@ def configure_sequence_parallel(sp_size):
     if sp_size > 1:
         init_sequence_parallel(sp_size)
 
-@spaces.GPU(duration=100)
+#@spaces.GPU(duration=100)
 def configure_runner(sp_size):
     config_path = os.path.join('./configs_3b', 'main.yaml')
     config = load_config(config_path)
@@ -211,7 +211,7 @@ def configure_runner(sp_size):
         runner.vae.set_memory_limit(**runner.config.vae.memory_limit)
     return runner
 
-@spaces.GPU(duration=100)
+#@spaces.GPU(duration=100)
 def generation_step(runner, text_embeds_dict, cond_latents):
     def _move_to_cuda(x):
         return [i.to(torch.device("cuda")) for i in x]
@@ -268,8 +268,8 @@ def generation_step(runner, text_embeds_dict, cond_latents):
 
     return samples
 
-@spaces.GPU(duration=100)
-def generation_loop(video_path='./test_videos', seed=666, fps_out=12, batch_size=1, cfg_scale=1.0, cfg_rescale=0.0, sample_steps=1, res_h=1280, res_w=720, sp_size=1):
+#@spaces.GPU(duration=100)
+def generation_loop(video_path='./test_videos', seed=666, fps_out=12, batch_size=1, cfg_scale=1.0, cfg_rescale=0.0, sample_steps=1, res_h=1280, res_w=720, sp_size=8):
     runner = configure_runner(1)
 
     def _extract_text_embeds():