Fix T5 double RAM use

nodes.py

@@ -387,7 +387,12 @@ class WanVideoTextEncode:
         params_to_keep = {'norm', 'pos_embedding', 'token_embedding'}
         for name, param in encoder.model.named_parameters():
             dtype_to_use = dtype if any(keyword in name for keyword in params_to_keep) else cast_dtype
-            set_module_tensor_to_device(encoder.model, name, device=device_to, dtype=dtype_to_use, value=encoder.state_dict[name])
+            value = encoder.state_dict[name] if hasattr(encoder, 'state_dict') else encoder.model.state_dict()[name]
+            set_module_tensor_to_device(encoder.model, name, device=device_to, dtype=dtype_to_use, value=value)
+        if hasattr(encoder, 'state_dict'):
+            del encoder.state_dict
+            mm.soft_empty_cache()
+            gc.collect()
 
         with torch.autocast(device_type=mm.get_autocast_device(device_to), dtype=encoder.dtype, enabled=encoder.quantization != 'disabled'):
             # Encode positive if not loaded from cache
@@ -411,6 +416,7 @@ class WanVideoTextEncode:
         if force_offload:
             encoder.model.to(offload_device)
             mm.soft_empty_cache()
+            gc.collect()
 
         prompt_embeds_dict = {
             "prompt_embeds": context,