Update app.py

app.py

@@ -17,6 +17,9 @@ from huggingface_hub import InferenceClient
 DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 MODEL_NAME = "microsoft/Phi-3-mini-4k-instruct"
 
+# Update transformers library
+!pip install --upgrade transformers
+
 # Load model and tokenizer, with specific handling for the Phi-3 model
 try:
     config = AutoConfig.from_pretrained(MODEL_NAME)
@@ -132,6 +135,12 @@ def model_inference(
         repetition_penalty,
         top_p,
 ):
+    if not isinstance(user_prompt, dict):
+        return "Invalid input format. Expected a dictionary."
+
+    if "files" not in user_prompt:
+        user_prompt["files"] = []
+
     if not user_prompt["files"]:
         if web_search:
             web_results = search(user_prompt["text"])
@@ -229,8 +238,10 @@ chatbot = gr.Chatbot(
 
 # Define Gradio interface
 def chat_interface(user_input, history, web_search, decoding_strategy, temperature, max_new_tokens, repetition_penalty, top_p):
+    # Ensure user_input is a dictionary
+    user_input_dict = {"text": user_input, "files": []}
     response = model_inference(
-        user_input,
+        user_input_dict,
         history,
         web_search,
         decoding_strategy,
@@ -264,4 +275,4 @@ interface = gr.Interface(
 )
 
 if __name__ == "__main__":
-    interface.launch()
+    interface.launch()