Update app.py

app.py

@@ -100,21 +100,6 @@ class StopOnTokens(StoppingCriteria):
         return False
 
 
-def user(message, history):
-    global source_text
-    global assess_rag
-    #For now, we only query the vector database once, at the start.
-    if len(history) == 0:
-        assess_rag = classification_chatrag(message)
-        if assess_rag:
-            source_text = vector_search(message)
-        else:
-            source_text = "Albert-Tchap n'utilise pas de sources comme votre requête n'a pas l'air d'en recueillir."
-    
-    history_transformer_format = history + [[message, ""]]
-
-    return history_transformer_format
-
 def predict(history_transformer_format):
 
     print(history_transformer_format)
@@ -144,6 +129,8 @@ def predict(history_transformer_format):
 
     messages = system_prompt + messages
 
+    print(messages)
+
     model_inputs = tokenizer([messages], return_tensors="pt").to("cuda")
     streamer = TextIteratorStreamer(tokenizer, timeout=10., skip_prompt=True, skip_special_tokens=True)
     generate_kwargs = dict(
@@ -165,6 +152,23 @@ def predict(history_transformer_format):
             yield partial_message
     return messages
 
+def user(message, history):
+    global source_text
+    global assess_rag
+    #For now, we only query the vector database once, at the start.
+    if len(history) == 0:
+        assess_rag = classification_chatrag(message)
+        if assess_rag:
+            source_text = vector_search(message)
+        else:
+            source_text = "Albert-Tchap n'utilise pas de sources comme votre requête n'a pas l'air d'en recueillir."
+    
+    history_transformer_format = history + [[message, ""]]
+
+    print(history_transformer_format)
+
+    return history_transformer_format
+
 # Define the Gradio interface
 title = "Tchap"
 description = "Le chatbot du service public"