back to this prompt

app.py

@@ -25,7 +25,7 @@ examples = [
 
 # LLM Settings
 # Initial
-system_prompt = 'You are a helpful assistant that speaks English and 日本語. Reply in the same language as the user.'
+system_prompt = 'You are a helpful, bilingual assistant. Reply in same language as the user.'
 default_prompt = system_prompt
 
 tokenizer = AutoTokenizer.from_pretrained(model_name)
@@ -35,20 +35,16 @@ model = AutoModelForCausalLM.from_pretrained(
     device_map="auto",
     # load_in_8bit=True,
     load_in_4bit=True,
-    quantization_config = BitsAndBytesConfig(
-        load_in_4bit=True,
-        bnb_4bit_quant_type='nf4',
-        bnb_4bit_use_double_quant=True,
-        bnb_4bit_compute_dtype=torch.bfloat16
-    ),
 )
 
 def chat(message, history, system_prompt):
-    print('---')
-    pprint(history)
     if not system_prompt:
         system_prompt = default_prompt
 
+    print('---')
+    print('Prompt:', system_prompt)
+    pprint(history)
+
     # Let's just rebuild every time it's easier
     chat_history = [{"role": "system", "content": system_prompt}]
     for h in history: