Update app.py

app.py

@@ -1,3 +1,4 @@
+{"from": "human", "value": "who are you"}
 import os
 from threading import Thread
 from typing import Iterator
@@ -24,7 +25,7 @@ if not torch.cuda.is_available():
 
 
 if torch.cuda.is_available():
-    model_id = "Nexusflow/Starling-LM-7B-beta"
+    model_id = "mahiatlinux/MasherAI-v6-7B"
     model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", load_in_4bit=True)
     tokenizer = AutoTokenizer.from_pretrained(model_id)
     tokenizer.use_default_system_prompt = False
@@ -43,10 +44,10 @@ def generate(
 ) -> Iterator[str]:
     conversation = []
     if system_prompt:
-        conversation.append({"role": "system", "content": system_prompt})
+        conversation.append({"from": "human", "value": "You are an AI assistant."})
     for user, assistant in chat_history:
-        conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
-    conversation.append({"role": "user", "content": message})
+        conversation.extend([{"from": "human", "value": user}, {"from": "gpt", "value": assistant}])
+    conversation.append({"from": "human", "value": message})
 
     input_ids = tokenizer.apply_chat_template(conversation, return_tensors="pt", add_generation_prompt=True)
     if input_ids.shape[1] > MAX_INPUT_TOKEN_LENGTH: