Update run_wan_I2V_CP+FB.py

run_wan_I2V_CP+FB.py

@@ -47,7 +47,8 @@ end_loading = time.time()
 from para_attn.first_block_cache.diffusers_adapters import apply_cache_on_pipe
 apply_cache_on_pipe(pipe , residual_diff_threshold=0.1)
 
-pipe.enable_model_cpu_offload()
+# pipe.enable_model_cpu_offload(gpu_id=dist.get_rank())
+pipe.enable_vae_tiling()
 
 # torch._inductor.config.reorder_for_compute_comm_overlap = True
 # pipe.transformer = torch.compile(pipe.transformer, mode="max-autotune-no-cudagraphs")