Update app.py

app.py

@@ -1,25 +1,23 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
 import random
+from transformers import AutoTokenizer
 from mySystemPrompt import SYSTEM_PROMPT
 
 # Model which is used
 checkpoint = "mistralai/Mistral-7B-Instruct-v0.2"
 # Inference client with the model (And HF-token if needed)
 client = InferenceClient(checkpoint)
-
-
+tokenizer = AutoTokenizer.from_pretrained(checkpoint)
 
 def format_prompt(message,chatbot,system_prompt):
-    if system_prompt != "":
-        input_prompt = f"<s>[INST] <<SYS>>\n{system_prompt}\n<</SYS>>\n\n "
-    else:
-        input_prompt = f"<s>[INST] "
-    for interaction in chatbot:
-        input_prompt = input_prompt + str(interaction[0]) + " [/INST] " + str(interaction[1]) + " </s><s>[INST] "
-    input_prompt = input_prompt + str(message) + " [/INST] "
-    return input_prompt
-
+    messages = [{"role": "system","content": system_prompt}]
+    for user_message,bot_message in chatbot:
+        messages.append({"role": "user", "content":user_message})
+        messages.append({"role": "assistant", "content":bot_message})
+    messages.append({"role": "user", "content":message})
+    tokenized_chat = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_tensors="pt")
+    return tokenized_chat
 
 def inference(message, history, systemPrompt=SYSTEM_PROMPT, temperature=0.9, maxTokens=512, topP=0.9, repPenalty=1.1):
     # Updating the settings for the generation