app.py

import os
from threading import Thread
from typing import Iterator

import gradio as gr
import spaces
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
import subprocess

subprocess.run(
    "pip install flash-attn --no-build-isolation",
    env={"FLASH_ATTENTION_SKIP_CUDA_BUILD": "TRUE"},
    shell=True,
)


DESCRIPTION = """\
# Lexora 3B ITA 💬 🇮🇹
"""

# Updated CSS to ensure full height and proper markdown rendering
CUSTOM_CSS = """
.gradio-container {
    height: 100vh !important;
    max-height: 100vh !important;
    padding: 0 !important;
    background-color: #0f1117;
}
.contain {
    height: 100vh !important;
    max-height: 100vh !important;
    display: flex;
    flex-direction: column;
}
.main-container {
    flex-grow: 1;
    height: calc(100vh - 100px) !important;
    overflow: hidden !important;
}
.chat-container {
    height: 100% !important;
    overflow: hidden !important;
    display: flex;
    flex-direction: column;
}
.chat-messages {
    flex-grow: 1;
    overflow-y: auto !important;
    padding: 1rem;
}
.message-wrap {
    height: auto !important;
    max-height: none !important;
}
.message {
    padding: 1rem !important;
    margin: 0.5rem 0 !important;
    border-radius: 0.5rem !important;
}
.user-message {
    background-color: #2b2d31 !important;
}
.bot-message {
    background-color: #1e1f23 !important;
}
.examples-container {
    margin-top: auto;
}
/* Markdown styling */
.bot-message p {
    margin-bottom: 0.5rem;
}
.bot-message h1, .bot-message h2, .bot-message h3, 
.bot-message h4, .bot-message h5, .bot-message h6 {
    margin-top: 1rem;
    margin-bottom: 0.5rem;
}
.bot-message code {
    background-color: #2d2d2d;
    padding: 0.2rem 0.4rem;
    border-radius: 0.2rem;
    font-family: monospace;
}
.bot-message pre {
    background-color: #2d2d2d;
    padding: 1rem;
    border-radius: 0.5rem;
    overflow-x: auto;
    margin: 1rem 0;
}
.bot-message pre code {
    background-color: transparent;
    padding: 0;
    border-radius: 0;
}
.bot-message ul, .bot-message ol {
    padding-left: 1.5rem;
    margin-bottom: 0.5rem;
}
.bot-message blockquote {
    border-left: 3px solid #4a4a4a;
    padding-left: 1rem;
    margin: 0.5rem 0;
    color: #a0a0a0;
}
.bot-message table {
    border-collapse: collapse;
    width: 100%;
    margin: 1rem 0;
}
.bot-message th, .bot-message td {
    border: 1px solid #4a4a4a;
    padding: 0.5rem;
    text-align: left;
}
.bot-message th {
    background-color: #2a2a2a;
}
.bot-message img {
    max-width: 100%;
    height: auto;
    margin: 1rem 0;
}
.bot-message a {
    color: #3291ff;
    text-decoration: none;
}
.bot-message a:hover {
    text-decoration: underline;
}
"""

MAX_MAX_NEW_TOKENS = 2048
DEFAULT_MAX_NEW_TOKENS = 1024
MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))

device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")

model_id = "DeepMount00/Lexora-Lite-3B"
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True,)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    device_map="auto",
    torch_dtype=torch.bfloat16,
    attn_implementation="flash_attention_2",
    trust_remote_code=True,
)
model.config.sliding_window = 4096
model.eval()


@spaces.GPU(duration=90)
def generate(
    message: str,
    chat_history: list[tuple[str, str]],
    system_message: str = "",
    max_new_tokens: int = 1024,
    temperature: float = 0.001,
    top_p: float = 1.0,
    top_k: int = 50,
    repetition_penalty: float = 1.0,
) -> Iterator[str]:
    conversation = [{"role": "system", "content": system_message}]
    for user, assistant in chat_history:
        conversation.extend(
            [
                {"role": "user", "content": user},
                {"role": "assistant", "content": assistant},
            ]
        )
    conversation.append({"role": "user", "content": message})

    input_ids = tokenizer.apply_chat_template(conversation, add_generation_prompt=True, return_tensors="pt")
    if input_ids.shape[1] > MAX_INPUT_TOKEN_LENGTH:
        input_ids = input_ids[:, -MAX_INPUT_TOKEN_LENGTH:]
        gr.Warning(f"Trimmed input from conversation as it was longer than {MAX_INPUT_TOKEN_LENGTH} tokens.")
    input_ids = input_ids.to(model.device)

    streamer = TextIteratorStreamer(tokenizer, timeout=20.0, 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,
        top_p=top_p,
        top_k=top_k,
        temperature=temperature,
        num_beams=1,
        repetition_penalty=repetition_penalty,
    )
    t = Thread(target=model.generate, kwargs=generate_kwargs)
    t.start()

    outputs = []
    for text in streamer:
        outputs.append(text)
        yield "".join(outputs)


chat_interface = gr.ChatInterface(
    fn=generate,
    additional_inputs=[
        gr.Textbox(
            value="",
            label="System message",
            render=False,
        ),        
        gr.Slider(
            label="Max new tokens",
            minimum=1,
            maximum=MAX_MAX_NEW_TOKENS,
            step=1,
            value=DEFAULT_MAX_NEW_TOKENS,
        ),
        gr.Slider(
            label="Temperature",
            minimum=0,
            maximum=4.0,
            step=0.1,
            value=0.001,
        ),
        gr.Slider(
            label="Top-p (nucleus sampling)",
            minimum=0.05,
            maximum=1.0,
            step=0.05,
            value=1.0,
        ),
        gr.Slider(
            label="Top-k",
            minimum=1,
            maximum=1000,
            step=1,
            value=50,
        ),
        gr.Slider(
            label="Repetition penalty",
            minimum=1.0,
            maximum=2.0,
            step=0.05,
            value=1.0,
        ),
    ],
    stop_btn=None,
    examples=[
        ["Ciao! Come stai?"],
        ["Puoi scrivere una lista markdown?"],
        ["Scrivi un esempio di codice Python"],
    ],
    cache_examples=False,
    render_markdown=True,  # Enable Markdown rendering
)

with gr.Blocks(css=CUSTOM_CSS, fill_height=True, theme=gr.themes.Base()) as demo:
    with gr.Column(elem_classes="contain"):
        gr.Markdown(DESCRIPTION)
        gr.DuplicateButton(value="Duplicate Space for private use", elem_id="duplicate-button")
        with gr.Column(elem_classes="main-container"):
            chat_interface.render()

if __name__ == "__main__":
    demo.queue(max_size=20).launch()