Upload 2 files

app.py

@@ -1,7 +1,7 @@
 import gradio as gr
 import torch
 import os
-from model_utils import load_model, generate_text
+from model_utils import load_model, generate_text, GPTConfig
 
 # Initialize model
 try:

model_utils.py

@@ -115,11 +115,10 @@ def load_model(model_path):
         
         # Create config from the saved dictionary
         config_dict = checkpoint['config']
-        if isinstance(config_dict, str):
-            # If config was saved as string, parse it to dict
-            import ast
-            config_dict = ast.literal_eval(config_dict)
-        config = GPTConfig(**config_dict)
+        if isinstance(config_dict, dict):
+            config = GPTConfig(**config_dict.__dict__)  # Convert dataclass to dict
+        else:
+            config = config_dict  # If config was already saved as GPTConfig instance
         
         model = GPT(config)
         model.load_state_dict(checkpoint['model_state_dict'])