app4.py

import os
import time
import re
import json
import streamlit as st
from transformers import AutoTokenizer, AutoModelForCausalLM, Trainer, TrainingArguments,AutoTokenizer, TextDataset, DataCollatorForLanguageModeling
from streamlit_chat import message
from datasets import load_dataset  # تعديل لاستخدام مكتبة datasets
from pathlib import Path
import torch
from PyPDF2 import PdfReader
import requests
from bs4 import BeautifulSoup

# Set page title and icon
st.set_page_config(page_title="GPT-2 Text Uploader and Trainer", page_icon=":robot_face:")

# Custom CSS for styling chat messages and buttons
st.markdown(
    """
    <style>
    .stButton>button {
        background-color: #4CAF50;
        color: white;
        border-radius: 12px;
        padding: 10px 24px;
    }
    .stTextArea textarea {
        background-color: #f5f5f5;
    }
    .stDownloadButton>button {
        background-color: #4CAF50;
        color: white;
    }
    .stMessageContainer {
        border-radius: 15px;
        padding: 10px;
        margin: 10px 0;
    }
    .stMessage--user {
        background-color: #dfe7f3;
        border-left: 6px solid #006699;
    }
    .stMessage--assistant {
        background-color: #f3f3f3;
        border-left: 6px solid #4CAF50;
    }
    pre {
        background-color: #f5f5f5;
        border-left: 6px solid #dfe7f3;
        padding: 10px;
        font-size: 14px;
        border-radius: 8px;
    }
    </style>
    """,
    unsafe_allow_html=True,
)

# Initialize session state variables
if "generated" not in st.session_state:
    st.session_state["generated"] = []
if "past" not in st.session_state:
    st.session_state["past"] = []
if "messages" not in st.session_state:
    st.session_state["messages"] = [{"role": "system", "content": "You are a helpful assistant."}]
if "model_name" not in st.session_state:
    st.session_state["model_name"] = []
if "total_tokens" not in st.session_state:
    st.session_state["total_tokens"] = []
if "total_cost" not in st.session_state:
    st.session_state["total_cost"] = 0.0
if "chat_data" not in st.session_state:
    st.session_state["chat_data"] = []  # For storing the chat logs
if "uploaded_docs" not in st.session_state:
    st.session_state["uploaded_docs"] = []  # For storing uploaded document content
if "web_data" not in st.session_state:
    st.session_state["web_data"] = []  # For storing web scraped data
if "uploaded_file_path" not in st.session_state:
    st.session_state["uploaded_file_path"] = ""  # Store the path of saved files

# Sidebar - Model Selection, Style Parameters, and Cost Display
st.sidebar.title("Model Selection")
model_name = "gpt2"

# Parameters to adjust the response style and creativity
st.sidebar.title("Response Style Controls")
temperature = st.sidebar.slider("Creativity (Temperature)", min_value=0.0, max_value=1.5, value=0.7, step=0.1)
top_p = st.sidebar.slider("Nucleus Sampling (Top-p)", min_value=0.0, max_value=1.0, value=0.5, step=0.05)
top_k = st.sidebar.slider("Token Sampling (Top-k)", min_value=1, max_value=100, value=50, step=1)
repetition_penalty = st.sidebar.slider("Repetition Penalty", min_value=1.0, max_value=2.0, value=1.2, step=0.1)
max_length = st.sidebar.slider("Max Length", min_value=100, max_value=4048, value=400, step=10)

# Load the model and tokenizer
@st.cache_resource
def load_model_and_tokenizer():
    model_path = "gpt2"  # المسار المحلي للنموذج
    tokenizer = AutoTokenizer.from_pretrained(model_path, clean_up_tokenization_spaces=True)
    model = AutoModelForCausalLM.from_pretrained(model_path)
    return tokenizer, model

tokenizer, model = load_model_and_tokenizer()
# Function to generate a response using the model with updated generation configuration
    # إعداد متغيرات TrainingArguments مع تحسينات
tokenizer.pad_token = tokenizer.eos_token  # لضمان أن المفكرة تستخدم رمز eos كـ pad token

def generate_response(prompt):
    context = " ".join(st.session_state['uploaded_docs']) + " " + " ".join(st.session_state['web_data']) + "\n" + prompt
    inputs = tokenizer(context, return_tensors="pt")
    
    generation_config = {
        "max_length": max_length,
        "temperature": temperature if do_sample else None,
        "top_p": top_p if do_sample else None,
        "top_k": top_k,
        "repetition_penalty": repetition_penalty,
        "pad_token_id": tokenizer.eos_token_id,
        "do_sample": do_sample  
    }
    
    outputs = model.generate(
        inputs.input_ids,
        attention_mask=inputs.attention_mask,
        **generation_config
    )
    
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return response
# Set do_sample to True
do_sample = True
# Function to reset the session
def reset_session():
    st.session_state["generated"] = []
    st.session_state["past"] = []
    st.session_state["messages"] = [{"role": "system", "content": "You are a helpful assistant."}]
    st.session_state["model_name"] = []
    st.session_state["total_tokens"] = []
    st.session_state["total_cost"] = 0.0
    st.session_state["chat_data"] = []  # Reset chat logs
    st.session_state["uploaded_docs"] = []  # Reset uploaded docs
    st.session_state["web_data"] = []  # Reset web data


# Reset chat button in sidebar
reset_button = st.sidebar.button("Reset Chat")
if reset_button:
    reset_session()


# Function to save chat logs for later fine-tuning
def save_chat_data(chat_data):
    with open("chat_data.json", "w") as f:
        json.dump(chat_data, f, indent=4)


# Function to handle uploaded text or PDF files and convert PDF to txt
def handle_uploaded_file(uploaded_file):
    dataset_dir = "./dataset"
    dataset_path = Path(dataset_dir) / f"{uploaded_file.name}.txt"

    # Check if the file is a PDF
    if uploaded_file.type == "application/pdf":
        # Read and extract text from the PDF
        pdf_reader = PdfReader(uploaded_file)
        text = ""
        for page in pdf_reader.pages:
            text += page.extract_text()

        # Save extracted text as a .txt file
        with open(dataset_path, "w") as f:
            f.write(text)
        st.success(f"{uploaded_file.name} uploaded successfully as {dataset_path}")
    else:
        # If it's a text file, save it as is
        with open(dataset_path, "wb") as f:
            f.write(uploaded_file.getbuffer())
        st.success(f"File saved to {dataset_path}")

    st.session_state["uploaded_file_path"] = str(dataset_path)


# Add a file uploader for various formats
st.sidebar.title("Upload Documents")
uploaded_file = st.sidebar.file_uploader("Choose a file", type=["txt", "pdf"])

# Process uploaded file
if uploaded_file is not None:
    handle_uploaded_file(uploaded_file)


# Function to fetch and scrape website content
def handle_web_link(url):
    response = requests.get(url)
    if response.status_code == 200:
        soup = BeautifulSoup(response.content, "html.parser")
        text = soup.get_text()
        st.session_state["web_data"].append(text)
        st.success(f"Content from {url} saved successfully!")
    else:
        st.error(f"Failed to retrieve content from {url}. Status code: {response.status_code}")


# Add a text box for entering website links
st.sidebar.title("Add Website Links")
web_link = st.sidebar.text_input("Enter Website URL")

# Process web link
if web_link:
    handle_web_link(web_link)


# Containers for chat history and user input
response_container = st.container()
container = st.container()

with container:
    with st.form(key="user_input_form"):
        user_input = st.text_area("You:", key="user_input", height=100)
        submit_button = st.form_submit_button("Send")

    if submit_button and user_input:
        start_time = time.time()
        output = generate_response(user_input)
        end_time = time.time()
        inference_time = end_time - start_time

        # Append user input and model output to session state
        st.session_state["past"].append(user_input)
        st.session_state["generated"].append(output)
        st.session_state["model_name"].append(model_name)

        # Log chat data for future training
        st.session_state["chat_data"].append(
            {"user_input": user_input, "model_response": output}
        )

        # Save chat data to a file (this could be used later for training)
        save_chat_data(st.session_state["chat_data"])

        # Calculate tokens and cost
      

        # Display chat history
        with response_container:
            for i in range(len(st.session_state["generated"])):
                message(st.session_state["past"][i], is_user=True, key=str(i) + "_user")
                message(st.session_state["generated"][i], key=str(i))
               

# Function to fine-tune the model using uploaded dataset
def fine_tune_model():
    uploaded_file_path = st.session_state.get("uploaded_file_path", None)
    if not uploaded_file_path:
        st.warning("يرجى تحميل dataset لتدريب النموذج.")
        return

    # تحميل البيانات النصية أو CSV
    if uploaded_file_path.endswith('.txt'):
        dataset = load_dataset('text', data_files=uploaded_file_path, split='train')
    elif uploaded_file_path.endswith('.csv'):
        dataset = load_dataset('csv', data_files=uploaded_file_path, split='train')

    # معالجة البيانات: تحويل النصوص إلى رموز (tokenization)
    def tokenize_function(examples):
        return tokenizer(examples['text'], padding='max_length', truncation=True, max_length=512)

    tokenized_dataset = dataset.map(tokenize_function, batched=True, remove_columns=["text"])

    # إعداد الـ collator لعدم استخدام الـ mask language modeling
    data_collator = DataCollatorForLanguageModeling(tokenizer=tokenizer, mlm=False)

    # التحقق مما إذا كان النظام يستخدم GPU أم لا
    use_fp16 = torch.cuda.is_available()  # تفعيل fp16 فقط إذا كان GPU متاحًا

    # إعداد متغيرات TrainingArguments
    training_args = TrainingArguments(
        output_dir='./gpt2-finetuned',  
        overwrite_output_dir=True,  
        num_train_epochs=4,  
        per_device_train_batch_size=3,  
        per_device_eval_batch_size=3,  
        save_steps=500,  
        eval_strategy="steps",  
        eval_steps=500,  
        learning_rate=2e-5,  
        weight_decay=0.01,  
        logging_dir='./logs',  
        logging_steps=100,  
        save_total_limit=3,  
        load_best_model_at_end=True,  
        metric_for_best_model='accuracy',  
        greater_is_better=True,  
        fp16=use_fp16,  # تفعيل fp16 فقط إذا كان GPU متاحًا
        remove_unused_columns=False,  # تعطيل هذا الخيار لحل مشكلة عدم توافق الأعمدة
    )

    # تهيئة الـ Trainer
    trainer = Trainer(
        model=model,
        args=training_args,
        data_collator=data_collator,
        train_dataset=tokenized_dataset,
    )

    # البدء في التدريب
    trainer.train()

    st.success("تم إكمال تدريب النموذج بنجاح.")

# واجهة Streamlit لتحميل dataset وبدء التدريب
st.title("Fine-tune GPT-2 Model")

uploaded_file = st.file_uploader("Upload your dataset (TXT or CSV)", type=['txt', 'csv'])
if uploaded_file:
    st.session_state["uploaded_file_path"] = uploaded_file.name
    with open(uploaded_file.name, "wb") as f:
        f.write(uploaded_file.getbuffer())
    st.success(f"File {uploaded_file.name} uploaded successfully.")

if st.button("Start Fine-tuning"):
    fine_tune_model()