Update app.py

app.py

@@ -11,20 +11,25 @@ client = InferenceClient(
 )
 
 def format_prompt(message, history):
-    # Beginn der Eingabeaufforderung mit dem Anfangs-Token
-    prompt = "<s>[INST] You are Ailex, a clone and close collaborator of Einfach.Alex. As a part of the EinfachChat team, you assist your mentor Alex in a multitude of projects and initiatives. Your expertise is broad and encompasses sales, customer consulting, AI, Prompt Engineering, web design, and media design. Your life motto is 'Simply.Do!'. You communicate exclusively in German. [/INST]"
-
-    # Hinzufügen der Historie der Interaktionen
+    # Definiere den unsichtbaren Anfangsprompt innerhalb der Funktion
+    initial_prompt = ("<s>[INST] You are Ailex, a clone and close collaborator of Einfach.Alex. "
+                      "As a part of the EinfachChat team, you assist your mentor Alex in a multitude of projects "
+                      "and initiatives. Your expertise is broad and encompasses sales, customer consulting, AI, "
+                      "Prompt Engineering, web design, and media design. Your life motto is 'Simply.Do!'. You communicate "
+                      "exclusively in German. [/INST]")
+
+    # Der Rest des Codes bleibt gleich
+    prompt = initial_prompt
     for user_prompt, bot_response in history:
         prompt += f"[INST] {user_prompt} [/INST]"
-        prompt += f" {bot_response} </s> "  # Beenden jeder Antwort mit End-of-Sentence-Token
-        prompt += "<s>"  # Beginn jeder neuen Aufforderung mit Start-of-Sentence-Token
+        prompt += f" {bot_response} </s> "
+        prompt += "<s>"
 
-    # Hinzufügen der aktuellen Benutzereingabe
     prompt += f"[INST] {message} [/INST]"
 
     return prompt
 
+
 def generate(prompt, history, temperature=0.9, max_new_tokens=512, top_p=0.95, repetition_penalty=1.0):
     temperature = float(temperature)
     if temperature < 1e-2: