Update app.py

app.py

@@ -1,62 +1,129 @@
+from threading import Thread
+
 import gradio as gr
-from huggingface_hub import InferenceClient
+import spaces
+from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
 
+
+TITLE = "<h1><center>Chat with Llama3-8B-Chinese-Chat-v2.1</center></h1>"
+
+DESCRIPTION = "<h3><center>Visit <a href='https://huggingface.co/shenzhi-wang/Llama3-8B-Chinese-Chat' target='_blank'>our model page</a> for details.</center></h3>"
+
+DEFAULT_SYSTEM = "You are Llama-3, developed by an independent team. You are a helpful assistant."
+
+TOOL_EXAMPLE = '''You have access to the following tools:
+```python
+def generate_password(length: int, include_symbols: Optional[bool]):
+ """
+ Generate a random password.
+ Args:
+ length (int): The length of the password
+ include_symbols (Optional[bool]): Include symbols in the password
+ """
+ pass
+```
+Write "Action:" followed by a list of actions in JSON that you want to call, e.g.
+Action:
+```json
+[
+ {
+ "name": "tool name (one of [generate_password])",
+ "arguments": "the input to the tool"
+ }
+]
+```
+'''
+
+CSS = """
+.duplicate-button {
+ margin: auto !important;
+ color: white !important;
+ background: black !important;
+ border-radius: 100vh !important;
+}
 """
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
 
-def respond(
- message,
- history: list[tuple[str, str]],
- system_message,
- max_tokens,
- temperature,
- top_p,
-):
- messages = [{"role": "system", "content": system_message}]
+tokenizer = AutoTokenizer.from_pretrained("SageLiao/llama3-LlamaFactory-demo-v3")
+model = AutoModelForCausalLM.from_pretrained("SageLiao/llama3-LlamaFactory-demo-v3", device_map="auto")
 
- for val in history:
- if val[0]:
- messages.append({"role": "user", "content": val[0]})
- if val[1]:
- messages.append({"role": "assistant", "content": val[1]})
+system = "You are the expert for customer service in Amazon.com.Please answer every question as a polite and professional agent for Amazon Company and be as detailed as possible. Please answer the customer's question with the language they used.如果使用者使用中文提問，請用繁體中文與台灣用語回答"
+@spaces.GPU
+def stream_chat(message: str, history: list, system: str, temperature: float, max_new_tokens: int):
+ conversation = [{"role": "system", "content": system}]
+ for prompt, answer in history:
+ conversation.extend([{"role": "user", "content": prompt}, {"role": "assistant", "content": answer}])
 
- messages.append({"role": "user", "content": message})
+ conversation.append({"role": "user", "content": message})
 
- response = ""
+ input_ids = tokenizer.apply_chat_template(conversation, add_generation_prompt=True, return_tensors="pt").to(
+ model.device
+ )
+ streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)
 
- for message in client.chat_completion(
- messages,
- max_tokens=max_tokens,
- stream=True,
+ generate_kwargs = dict(
+ input_ids=input_ids,
+ streamer=streamer,
+ max_new_tokens=max_new_tokens,
  temperature=temperature,
- top_p=top_p,
- ):
- token = message.choices[0].delta.content
+ do_sample=True,
+ )
+ if temperature == 0:
+ generate_kwargs["do_sample"] = False
 
- response += token
- yield response
+ t = Thread(target=model.generate, kwargs=generate_kwargs)
+ t.start()
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
- respond,
- additional_inputs=[
- gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
- gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
- gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
- gr.Slider(
- minimum=0.1,
- maximum=1.0,
- value=0.95,
- step=0.05,
- label="Top-p (nucleus sampling)",
- ),
- ],
-)
+ output = ""
+ for new_token in streamer:
+ output += new_token
+ yield output
+
+
+chatbot = gr.Chatbot(height=450)
+
+with gr.Blocks(css=CSS) as demo:
+ gr.HTML(TITLE)
+ gr.HTML(DESCRIPTION)
+ gr.DuplicateButton(value="Duplicate Space for private use", elem_classes="duplicate-button")
+ gr.ChatInterface(
+ fn=stream_chat,
+ chatbot=chatbot,
+ fill_height=True,
+ additional_inputs_accordion=gr.Accordion(label="⚙️ Parameters", open=False, render=False),
+ additional_inputs=[
+ gr.Text(
+ value="",
+ label="System",
+ render=False,
+ ),
+ gr.Slider(
+ minimum=0,
+ maximum=1,
+ step=0.1,
+ value=0.8,
+ label="Temperature",
+ render=False,
+ ),
+ gr.Slider(
+ minimum=128,
+ maximum=4096,
+ step=1,
+ value=1024,
+ label="Max new tokens",
+ render=False,
+ ),
+ ],
+ examples=[
+ ["我的蓝牙耳机坏了，我该去看牙科还是耳鼻喉科？", ""],
+ ["7年前，妈妈年龄是儿子的6倍，儿子今年12岁，妈妈今年多少岁？", ""],
+ ["我的笔记本找不到了。", "扮演诸葛亮和我对话。"],
+ ["我想要一个新的密码，长度为8位，包含特殊符号。", TOOL_EXAMPLE],
+ ["How are you today?", "You are Taylor Swift, use beautiful lyrics to answer questions."],
+ ["用C++实现KMP算法，并加上中文注释", ""],
+ ],
+ cache_examples=False,
+ )
 
 
 if __name__ == "__main__":