fix weight loading bug

Author: HaoyiZhu

.gitignore

@@ -152,3 +152,5 @@ configs/local/default.yaml
 
 # Aim logging
 .aim
+
+evaluations/

README.md

@@ -204,6 +204,7 @@ Then please follow the instructions in the [CortexBench README](https://github.c
 
 <details>
 <summary><b>Camera Pose Evaluation</b></summary>
+
 To reproduce the camera pose evaluation, we have open-sourced the code in [evaluation/probe3d](evaluation/probe3d). Please first run `git submodule update --init --recursive` and `cd evaluation/probe3d`. Then follow the instructions in [probe3d](https://github.com/HaoyiZhu/probe3d/blob/main/data_processing/README.md) to prepare the **NAVI** dataset. Finally, run the following command to evaluate SPA:
 
 ```console

spa/models/components/spa.py

@@ -132,7 +132,8 @@ def load_pretrained(self, ckpt_name: str = None):
                 repo_id="HaoyiZhu/SPA", filename=f"{ckpt_name}.ckpt"
             )
             state_dict = torch.load(ckpt_file)["state_dict"]
-
+        # remove "model." prefix
+        state_dict = {k[6:]: v for k, v in state_dict.items()}
         self.load_state_dict(state_dict, strict=True)
 
     def init_data_processor(self):