app.py

model_name = "gemma2:27b"

import os 

os.system("sudo apt install lshw")
os.system("curl https://ollama.ai/install.sh | sh")

import nest_asyncio
nest_asyncio.apply()

import os
import asyncio

# Run Async Ollama
# Taken from: https://stackoverflow.com/questions/77697302/how-to-run-ollama-in-google-colab
# NB: You may need to set these depending and get cuda working depending which backend you are running.
# Set environment variable for NVIDIA library
# Set environment variables for CUDA
os.environ['PATH'] += ':/usr/local/cuda/bin'
# Set LD_LIBRARY_PATH to include both /usr/lib64-nvidia and CUDA lib directories
os.environ['LD_LIBRARY_PATH'] = '/usr/lib64-nvidia:/usr/local/cuda/lib64'

async def run_process(cmd):
    print('>>> starting', *cmd)
    process = await asyncio.create_subprocess_exec(
        *cmd,
        stdout=asyncio.subprocess.PIPE,
        stderr=asyncio.subprocess.PIPE
    )

    # define an async pipe function
    async def pipe(lines):
        async for line in lines:
            print(line.decode().strip())

        await asyncio.gather(
            pipe(process.stdout),
            pipe(process.stderr),
        )

    # call it
    await asyncio.gather(pipe(process.stdout), pipe(process.stderr))

import asyncio
import threading

async def start_ollama_serve():
    await run_process(['ollama', 'serve'])

def run_async_in_thread(loop, coro):
    asyncio.set_event_loop(loop)
    loop.run_until_complete(coro)
    loop.close()

# Create a new event loop that will run in a new thread
new_loop = asyncio.new_event_loop()

# Start ollama serve in a separate thread so the cell won't block execution
thread = threading.Thread(target=run_async_in_thread, args=(new_loop, start_ollama_serve()))
thread.start()

# Load up model

os.system(f"ollama pull {model_name}")


import copy
import gradio as gr
import spaces
from llama_index.llms.ollama import Ollama
import llama_index
from llama_index.core.llms import ChatMessage


HF_TOKEN = os.environ.get("HF_TOKEN", None)
MODEL_ID = "google/gemma-2-27b-it"
MODEL_NAME = MODEL_ID.split("/")[-1]

os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"


gemma2 = Ollama(model=model_name, request_timeout=30.0)


TITLE = "<h1><center>Chatbox</center></h1>"

DESCRIPTION = f"""
<h3>MODEL: <a href="https://hf.co/{MODELS}">{MODEL_NAME}</a></h3>
<center>
<p>Gemma is the large language model built by Google.
<br>
Feel free to test without log.
</p>
</center>
"""

CSS = """
.duplicate-button {
    margin: auto !important;
    color: white !important;
    background: black !important;
    border-radius: 100vh !important;
}
h3 {
    text-align: center;
}
"""


@spaces.GPU(duration=90)
def stream_chat(message: str, history: list, temperature: float, context_window: int, top_p: float, top_k: int, penalty: float):
    print(f'message is - {message}')
    print(f'history is - {history}')
    conversation = []
    for prompt, answer in history:
        conversation.extend([
            ChatMessage(
            role="user", content=prompt
            ),
            ChatMessage(role="assistant", content=answer),
        ])
    messages = [ChatMessage(role="user", content=message)]

    print(f"Conversation is -\n{conversation}")

    resp = gemma2.stream_chat(
        message = messages,
        chat_history = conversation,
        top_p=top_p,
        top_k=top_k,
        repeat_penalty=penalty,
        context_window=context_window,            
    )
        

    for r in resp:
        yield r.delta


chatbot = gr.Chatbot(height=600)

with gr.Blocks(css=CSS, theme="soft") 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.Slider(
                minimum=0,
                maximum=1,
                step=0.1,
                value=0.8,
                label="Temperature",
                render=False,
            ),
            gr.Slider(
                minimum=128,
                maximum=2048,
                step=1,
                value=1024,
                label="Context window",
                render=False,
            ),
            gr.Slider(
                minimum=0.0,
                maximum=1.0,
                step=0.1,
                value=0.8,
                label="top_p",
                render=False,
            ),
            gr.Slider(
                minimum=1,
                maximum=20,
                step=1,
                value=20,
                label="top_k",
                render=False,
            ),
            gr.Slider(
                minimum=0.0,
                maximum=2.0,
                step=0.1,
                value=1.0,
                label="Repetition penalty",
                render=False,
            ),
        ],
        examples=[
            ["Help me study vocabulary: write a sentence for me to fill in the blank, and I'll try to pick the correct option."],
            ["What are 5 creative things I could do with my kids' art? I don't want to throw them away, but it's also so much clutter."],
            ["Tell me a random fun fact about the Roman Empire."],
            ["Show me a code snippet of a website's sticky header in CSS and JavaScript."],
        ],
        cache_examples=False,
    )


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