Update app.py

app.py

@@ -54,26 +54,14 @@ latex_delimiters_set = [{
 @spaces.GPU()
 def predict(message, history, system_prompt, temperature, max_new_tokens, top_k, repetition_penalty, top_p):
     # Format history with a given chat template
-    if CHAT_TEMPLATE == "Auto":
-        stop_tokens = [tokenizer.eos_token_id]
-        instruction = system_prompt + "\n\n"
-        for user, assistant in history:
-            instruction += f"User: {user}\nAssistant: {assistant}\n"
-        instruction += f"User: {message}\nAssistant:"
-    elif CHAT_TEMPLATE == "ChatML":
-        stop_tokens = ["<|endoftext|>", "<|im_end|>"]
-        instruction = '<|im_start|>system\n' + system_prompt + '\n<|im_end|>\n'
-        for user, assistant in history:
-            instruction += f'<|im_start|>user\n{user}\n<|im_end|>\n<|im_start|>assistant\n{assistant}\n<|im_end|>\n'
-        instruction += f'<|im_start|>user\n{message}\n<|im_end|>\n<|im_start|>assistant\n'
-    elif CHAT_TEMPLATE == "Mistral Instruct":
-        stop_tokens = ["</s>", "[INST]", "[INST] ", "<s>", "[/INST]", "[/INST] "]
-        instruction = f'<s>[INST] {system_prompt}\n'
-        for user, assistant in history:
-            instruction += f'{user} [/INST] {assistant}</s>[INST]'
-        instruction += f' {message} [/INST]'
-    else:
-        raise Exception("Incorrect chat template, select 'Auto', 'ChatML' or 'Mistral Instruct'")
+    
+    
+    stop_tokens = ["<|endoftext|>", "<|im_end|>","|im_end|"]
+    instruction = '<|im_start|>system\n' + system_prompt + '\n<|im_end|>\n'
+    for user, assistant in history:
+        instruction += f'<|im_start|>user\n{user}\n<|im_end|>\n<|im_start|>assistant\n{assistant}\n<|im_end|>\n'
+    instruction += f'<|im_start|>user\n{message}\n<|im_end|>\n<|im_start|>assistant\n'
+    
     print(instruction)
     
     streamer = TextIteratorStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)