Use prompt structure from prompts directory

app.py

@@ -2,6 +2,7 @@ import gradio as gr
 from transformers import AutoTokenizer, AutoModelForCausalLM
 import torch
 import spaces
+from prompts.base_instruction import basic_instruction
 
 # Load the model and tokenizer
 tokenizer = AutoTokenizer.from_pretrained("braindeck/text2text", trust_remote_code=True, subfolder="checkpoints/model")
@@ -12,7 +13,7 @@ def generate_response(prompt):
     """
     Generates a response from the model.
     """
-    chat = [{"role": "user", "content": prompt}]
+    chat = basic_instruction(prompt, "braindeck/text2text")
     inputs = tokenizer.apply_chat_template(chat, tokenize=True, add_generation_prompt=True, return_tensors="pt").to(model.device)
     input_length = inputs.shape[1]
     outputs = model.generate(inputs, max_new_tokens=512, do_sample=False)
@@ -49,4 +50,4 @@ with gr.Blocks() as demo:
     )
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()