app.py

from huggingface_hub import InferenceClient
import gradio as gr

client = InferenceClient(
    "mistralai/Mistral-7B-Instruct-v0.1"
)


def format_prompt(message, history):
  prompt = "<s>"
  for user_prompt, bot_response in history:
    prompt += f"[INST] {user_prompt} [/INST]"
    prompt += f" {bot_response}</s> "
  prompt += f"[INST] {message} [/INST]"
  return prompt

def generate(
    prompt, history, temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0,
):
    temperature = float(temperature)
    if temperature < 1e-2:
        temperature = 1e-2
    top_p = float(top_p)

    generate_kwargs = dict(
        temperature=temperature,
        max_new_tokens=max_new_tokens,
        top_p=top_p,
        repetition_penalty=repetition_penalty,
        do_sample=True,
        seed=42,
    )

    formatted_prompt = format_prompt(prompt, history)

    stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
    output = ""

    for response in stream:
        output += response.token.text
        yield output
    return output


additional_inputs=[
    gr.Slider(
        label="Temperature",
        value=0.9,
        minimum=0.0,
        maximum=1.0,
        step=0.05,
        interactive=True,
        info="Higher values produce more diverse outputs",
    ),
    gr.Slider(
        label="Max new tokens",
        value=256,
        minimum=0,
        maximum=1048,
        step=64,
        interactive=True,
        info="The maximum numbers of new tokens",
    ),
    gr.Slider(
        label="Top-p (nucleus sampling)",
        value=0.90,
        minimum=0.0,
        maximum=1,
        step=0.05,
        interactive=True,
        info="Higher values sample more low-probability tokens",
    ),
    gr.Slider(
        label="Repetition penalty",
        value=1.2,
        minimum=1.0,
        maximum=2.0,
        step=0.05,
        interactive=True,
        info="Penalize repeated tokens",
    )
]

css = """
  #mkd {
    height: 200px; 
    overflow: auto; 
    border: 1px solid #ccc; 
  }
"""

with gr.Blocks(css=css) as demo:
  
    gr.ChatInterface(
        generate,
        additional_inputs=additional_inputs,
        examples=[
        ["Create a ten-point markdown outline with emojis about: Decreased α-ketoglutarate dehydrogenase activity in astrocytes"],
        ["Create a ten-point markdown outline with emojis about: Lewy body dementia"],
        ["Create a ten-point markdown outline with emojis about: Delusional disorder"],
        ["Create a ten-point markdown outline with emojis about: Galantamine"],
        ["Create a ten-point markdown outline with emojis about: Neural crest"],
        ["Create a ten-point markdown outline with emojis about: Progressive multifocal encephalopathy (PML)"],
        ["Create a ten-point markdown outline with emojis about: CT head"],
        ["Create a ten-point markdown outline with emojis about: β-Galactocerebrosidase"],
        ["Create a ten-point markdown outline with emojis about: Dopamine"],
        ["Create a ten-point markdown outline with emojis about: G protein-coupled receptors"],
        ["Create a ten-point markdown outline with emojis about: CT scan of the head without contrast"],
        ["Create a ten-point markdown outline with emojis about: Pyogenic brain abscess"],
        ["Create a ten-point markdown outline with emojis about: Pneumocystitis jiroveci"]
        ]
    )
    gr.HTML("""<h2>🤖 Mistral Chat - Gradio 🤖</h2>
        In this demo, you can chat with <a href='https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1'>Mistral-7B-Instruct</a> model. 💬
        Learn more about the model <a href='https://huggingface.co/docs/transformers/main/model_doc/mistral'>here</a>. 📚
        <h2>🛠 Model Features 🛠</h2>
        <ul>
          <li>🪟 Sliding Window Attention with 128K tokens span</li>
          <li>🚀 GQA for faster inference</li>
          <li>📝 Byte-fallback BPE tokenizer</li>
        </ul>
        <h3>📜 License 📜  Released under Apache 2.0 License</h3>
        <h3>📦 Usage 📦</h3>
        <ul>
          <li>📚 Available on Huggingface Hub</li>
          <li>🐍 Python code snippets for easy setup</li>
          <li>📈 Expected speedups with Flash Attention 2</li>
        </ul>
    """)
demo.queue().launch(debug=True)