app.py

# Inference

import gradio as gr
from huggingface_hub import InferenceClient

model = "meta-llama/Llama-3.2-3B-Instruct"
client = InferenceClient(model)

def fn(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    messages = [{"role": "system", "content": system_message}]

    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            #messages.append({"role": "assistant", "content": val[1]})
            messages.append({"role": "bot", "content": val[1]})

    messages.append({"role": "user", "content": message})

    response = ""

    for message in client.chat_completion(
        messages,
        max_tokens = max_tokens,
        temperature = temperature,
        top_p = top_p,
        stream = True,
    ):
        token = message.choices[0].delta.content

        response += token
        yield response

app = gr.ChatInterface(
    fn = fn,
    additional_inputs = [
        gr.Textbox(value="You are a helpful assistant.", label="System Message"),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max 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"),
    ],
    title = "Meta Llama",
    description = model,
)

if __name__ == "__main__":
    app.launch()

"""
# Pipeline

import gradio as gr
from transformers import pipeline

pipe = pipeline(model = "meta-llama/Llama-3.2-3B-Instruct")

def fn(input):
    output = pipe(
        input,
        max_new_tokens = 2048
    )
    return output[0]["generated_text"]#[len(input):]

app = gr.Interface(
    fn = fn,
    inputs = [gr.Textbox(label = "Input")],
    outputs = [gr.Textbox(label = "Output")],
    title = "Meta Llama",
    description = "Pipeline",
    examples = [
        ["Hello, World."]
    ]
).launch()
"""