Try adding initial load to CPU to see if runtime error goes away

app.py

@@ -27,7 +27,7 @@ import open_clip
 
 def load_model_from_config(config, ckpt, verbose=False):
     print(f"Loading model from {ckpt}")
-    pl_sd = torch.load(ckpt, map_location="cuda")
+    pl_sd = torch.load(ckpt, map_location="cpu")
     sd = pl_sd["state_dict"]
     model = instantiate_from_config(config.model)
     m, u = model.load_state_dict(sd, strict=False)