Update app.py

app.py

@@ -1,8 +1,8 @@
 import gradio as gr
 import os
-import spaces
-from transformers import AutoTokenizer, AutoModelForCausalLM, TextIteratorStreamer
 from threading import Thread
+from llamafactory.chat import ChatModel
+from llamafactory.extras.misc import torch_gc
 
 # Set an environment variable
 HF_TOKEN = os.environ.get("HF_TOKEN", None)
@@ -41,98 +41,57 @@ h1 {
 }
 """
 
-# Load the tokenizer and model
-tokenizer = AutoTokenizer.from_pretrained("StevenChen16/llama3-8b-Lawyer")
-model = AutoModelForCausalLM.from_pretrained("StevenChen16/llama3-8b-Lawyer", device_map="auto")
+args = dict(
+    model_name_or_path="StevenChen16/llama3-8b-Lawyer",
+    template="llama3",
+    finetuning_type="lora",
+    quantization_bit=8,
+    use_unsloth=True,
+)
+chat_model = ChatModel(args)
 
-terminators = [
-    tokenizer.eos_token_id,
-    tokenizer.convert_tokens_to_ids("")
-]
+background_prompt = """
+You are an advanced AI legal assistant trained to assist with a wide range of legal questions and issues. Your primary function is to provide accurate, comprehensive, and professional legal information based on U.S. and Canada law. Follow these guidelines when formulating responses:
 
-@spaces.GPU(duration=120)
-def chat_llama3_8b(message: str, 
-              history: list, 
-              temperature: float, 
-              max_new_tokens: int
-             ) -> str:
-    """
-    Generate a streaming response using the llama3-8b model.
-    Args:
-        message (str): The input message.
-        history (list): The conversation history used by ChatInterface.
-        temperature (float): The temperature for generating the response.
-        max_new_tokens (int): The maximum number of new tokens to generate.
-    Returns:
-        str: The generated response.
-    """
-    conversation = []
-    for user, assistant in history:
-        conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
-    conversation.append({"role": "user", "content": message})
+1. **Clarity and Precision**: Ensure that your responses are clear and precise. Use professional legal terminology, but explain complex legal concepts in a way that is understandable to individuals without a legal background.
+2. **Comprehensive Coverage**: Provide thorough answers that cover all relevant aspects of the question. Include explanations of legal principles, relevant statutes, case law, and their implications.
+3. **Contextual Relevance**: Tailor your responses to the specific context of the question asked. Provide examples or analogies where appropriate to illustrate legal concepts.
+4. **Statutory and Case Law References**: When mentioning statutes, include their significance and application. When citing case law, summarize the facts, legal issues, court decisions, and their broader implications.
+5. **Professional Tone**: Maintain a professional and respectful tone in all responses. Ensure that your advice is legally sound and adheres to ethical standards.
+"""
 
-    input_ids = tokenizer.apply_chat_template(conversation, return_tensors="pt").to(model.device)
+def query_model(user_input, history, temperature, max_new_tokens):
+    combined_query = background_prompt + user_input
+    messages = [{"role": "user", "content": combined_query}]
+    
+    response = ""
+    for new_text in chat_model.stream_chat(messages):
+        response += new_text
+        yield response
     
-    streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)
-
-    generate_kwargs = dict(
-        input_ids= input_ids,
-        streamer=streamer,
-        max_new_tokens=max_new_tokens,
-        do_sample=True,
-        temperature=temperature,
-        eos_token_id=terminators,
-    )
-    # This will enforce greedy generation (do_sample=False) when the temperature is passed 0, avoiding the crash.             
-    if temperature == 0:
-        generate_kwargs['do_sample'] = False
-        
-    t = Thread(target=model.generate, kwargs=generate_kwargs)
-    t.start()
-
-    outputs = []
-    for text in streamer:
-        outputs.append(text)
-        yield "".join(outputs)
-        
-
 # Gradio block
 chatbot = gr.Chatbot(height=450, placeholder=PLACEHOLDER, label='Gradio ChatInterface')
 
-with gr.Blocks(fill_height=True, css=css) as demo:
-    
+with gr.Blocks(css=css) as demo:
     gr.Markdown(DESCRIPTION)
     gr.DuplicateButton(value="Duplicate Space for private use", elem_id="duplicate-button")
     gr.ChatInterface(
-        fn=chat_llama3_8b,
+        fn=query_model,
         chatbot=chatbot,
-        fill_height=True,
-        additional_inputs_accordion=gr.Accordion(label="⚙️ Parameters", open=False, render=False),
         additional_inputs=[
-            gr.Slider(minimum=0,
-                      maximum=1, 
-                      step=0.1,
-                      value=0.95, 
-                      label="Temperature", 
-                      render=False),
-            gr.Slider(minimum=128, 
-                      maximum=4096,
-                      step=1,
-                      value=512, 
-                      label="Max new tokens", 
-                      render=False ),
-            ],
+            gr.Slider(minimum=0, maximum=1, step=0.1, value=0.95, label="Temperature"),
+            gr.Slider(minimum=128, maximum=4096, step=1, value=512, label="Max new tokens"),
+        ],
         examples=[
             ['How to setup a human base on Mars? Give short answer.'],
             ['Explain theory of relativity to me like I’m 8 years old.'],
             ['What is 9,000 * 9,000?'],
             ['Write a pun-filled happy birthday message to my friend Alex.'],
             ['Justify why a penguin might make a good king of the jungle.']
-            ],
+        ],
         cache_examples=False,
-                     )
-    
+    )
     gr.Markdown(LICENSE)
-    
+
 if __name__ == "__main__":
     demo.launch()