C10X's picture
Update app.py
2bc9ed8 verified
import os
import subprocess
import signal
os.environ["GRADIO_ANALYTICS_ENABLED"] = "False"
import gradio as gr
import tempfile
import torch
from datasets import load_dataset
from tqdm.auto import tqdm
import re
import numpy as np
import gc
import unicodedata
from multiprocessing import cpu_count
from transformers import LlamaTokenizerFast
import fasttext
from typing import Tuple, Dict, List, Generator
import json
import matplotlib.pyplot as plt
import seaborn as sns
from datetime import datetime
import warnings
from huggingface_hub import HfApi, create_repo, upload_file, snapshot_download, whoami, HfFolder
from pathlib import Path
from textwrap import dedent
from scipy import stats
from apscheduler.schedulers.background import BackgroundScheduler
warnings.filterwarnings('ignore')
# Environment variables
HF_TOKEN = os.environ.get("HF_TOKEN")
# Global variables for model caching
MODEL_CACHE_DIR = Path.home() / ".cache" / "ultra_fineweb"
MODEL_CACHE_DIR.mkdir(parents=True, exist_ok=True)
MODEL_LOADED = False
fasttext_model = None
tokenizer = None
# CSS
css = """
.gradio-container {overflow-y: auto;}
.gr-button-primary {
background-color: #ff6b00 !important;
border-color: #ff6b00 !important;
}
.gr-button-primary:hover {
background-color: #ff8534 !important;
border-color: #ff8534 !important;
}
"""
# HTML templates
TITLE = """
<div style="text-align: center; margin-bottom: 30px;">
<h1 style="font-size: 36px; margin-bottom: 10px;">Create your own Dataset Quality Scores, blazingly fast ⚡!</h1>
<p style="font-size: 16px; color: #666;">The space takes a HF dataset as input, scores it and provides statistics and quality distribution.</p>
</div>
"""
DESCRIPTION_MD = """
### 📋 How it works:
1. Choose a dataset from Hugging Face Hub.
2. The Ultra-FineWeb classifier will score each text sample.
3. View quality distribution and download the scored dataset.
4. Optionally, upload the results to a new repository on your Hugging Face account.
**Note:** The first run will download the model (~347MB), which may take a moment.
"""
# --- Helper Functions ---
# ==============================================================================
# --- HATAYI GİDEREN KESİN VE NİHAİ DÜZELTME BURADA ---
# `escape` fonksiyonu, olması gereken doğru haline geri getirildi.
# ==============================================================================
def escape(s: str) -> str:
"""Escape special characters for safe HTML display."""
s = str(s)
s = s.replace("&", "&amp;")
s = s.replace("<", "&lt;")
s = s.replace(">", "&gt;")
s = s.replace('"', "&quot;")
s = s.replace("\n", "<br/>")
return s
def fasttext_preprocess(content: str, tokenizer) -> str:
if not isinstance(content, str): return ""
content = re.sub(r'\n{3,}', '\n\n', content).lower()
content = ''.join(c for c in unicodedata.normalize('NFKD', content) if unicodedata.category(c) != 'Mn')
token_ids = tokenizer.encode(content, add_special_tokens=False)
content = ' '.join([tokenizer.decode([token_id]) for token_id in token_ids])
content = re.sub(r'\n', ' n ', content).replace('\r', '').replace('\t', ' ')
return re.sub(r' +', ' ', content).strip()
def fasttext_infer(norm_content: str, model) -> Tuple[str, float]:
"""Run inference using the FastText model.
Args:
norm_content: Normalized text content to score
model: Loaded FastText model
Returns:
Tuple of (predicted_label, score) where score is between 0 and 1
"""
try:
# Get prediction from model
pred_label, pred_prob = model.predict(norm_content)
# Handle different label formats
if isinstance(pred_label, (list, np.ndarray)) and len(pred_label) > 0:
pred_label = pred_label[0]
# Default score if we can't process it
score = 0.5
# Handle different probability formats
if pred_prob is not None:
# If it's a numpy array, convert to list
if hasattr(pred_prob, 'tolist'):
pred_prob = pred_prob.tolist()
# Handle list/array formats
if isinstance(pred_prob, (list, np.ndarray)) and len(pred_prob) > 0:
# Get first element if it's a nested structure
first_prob = pred_prob[0] if not isinstance(pred_prob[0], (list, np.ndarray)) else pred_prob[0][0]
score = float(first_prob)
else:
# Try direct conversion if it's a single value
score = float(pred_prob)
# Ensure score is between 0 and 1
score = max(0.0, min(1.0, score))
return pred_label, score
except Exception as e:
print(f"Error in fasttext_infer: {e}")
return "__label__neg", 0.0
def load_models():
global MODEL_LOADED, fasttext_model, tokenizer
if MODEL_LOADED and tokenizer is not None and fasttext_model is not None:
return tokenizer, fasttext_model
try:
model_dir = MODEL_CACHE_DIR / "Ultra-FineWeb-classifier"
if not model_dir.exists():
snapshot_download(repo_id="openbmb/Ultra-FineWeb-classifier", local_dir=str(model_dir), local_dir_use_symlinks=False)
# Load tokenizer and model
tokenizer = LlamaTokenizerFast.from_pretrained(str(model_dir / "tokenizer"))
fasttext_model = fasttext.load_model(str(model_dir / "classifier.bin"))
MODEL_LOADED = True
return tokenizer, fasttext_model
except Exception as e:
print(f"Error loading models: {e}")
return None, None
def create_quality_plot(scores: List[float], dataset_name: str) -> str:
with tempfile.NamedTemporaryFile(suffix=".png", delete=False) as tmpfile:
output_path = tmpfile.name
plt.figure(figsize=(10, 6))
sns.histplot(scores, bins=50, kde=True, color='#6B7FD7', edgecolor='black')
mean_score, median_score = np.mean(scores), np.median(scores)
plt.axvline(mean_score, color='green', linestyle='--', linewidth=2, label=f'Mean: {mean_score:.3f}')
plt.axvline(median_score, color='orange', linestyle=':', linewidth=2, label=f'Median: {median_score:.3f}')
plt.xlabel('Quality Score'); plt.ylabel('Density')
plt.title(f'Quality Score Distribution - {dataset_name}', fontweight='bold')
plt.legend(); plt.grid(axis='y', alpha=0.3); plt.xlim(0, 1)
plt.tight_layout(); plt.savefig(output_path, dpi=150)
plt.close()
return output_path
def process_dataset(
model_id: str,
dataset_split: str,
text_column: str,
sample_size: int,
batch_size: int,
progress=gr.Progress(track_tqdm=True)
) -> Generator:
log_text = ""
def update_log(msg):
nonlocal log_text
timestamp = datetime.now().strftime('%H:%M:%S')
log_text += f"[{timestamp}] {msg}\n"
return (log_text, None, None, None, None, gr.update(visible=False), gr.update(visible=False))
try:
yield update_log("Starting process...")
yield update_log("Loading scoring models...")
if not load_models():
raise gr.Error("Failed to load scoring models. Please check logs.")
yield update_log("Models loaded successfully.")
yield update_log(f"Loading dataset '{model_id}' split '{dataset_split}'...")
dataset = load_dataset(model_id, split=dataset_split, streaming=False)
yield update_log("Dataset loaded.")
if text_column not in dataset.column_names:
raise gr.Error(f"Column '{text_column}' not found. Available: {', '.join(dataset.column_names)}")
actual_samples = min(sample_size, len(dataset))
dataset = dataset.select(range(actual_samples))
yield update_log(f"Starting to score {actual_samples:,} samples...")
scores, scored_data = [], []
for i in tqdm(range(0, actual_samples, batch_size), desc="Scoring batches"):
batch = dataset[i:min(i + batch_size, actual_samples)]
for text in batch[text_column]:
norm_content = fasttext_preprocess(text, tokenizer)
label, score = fasttext_infer(norm_content, fasttext_model) if norm_content else ("__label__neg", 0.0)
scores.append(score)
scored_data.append({'text': text, 'quality_score': score, 'predicted_label': label})
yield update_log("Scoring complete. Generating results and plot...")
stats_dict = {'dataset_id': model_id, 'processed_samples': actual_samples, 'statistics': {'mean': float(np.mean(scores)), 'median': float(np.median(scores))}}
plot_file = create_quality_plot(scores, model_id.split('/')[-1])
with tempfile.NamedTemporaryFile('w', suffix=".jsonl", delete=False, encoding='utf-8') as f:
output_file_path = f.name
for item in scored_data: f.write(json.dumps(item, ensure_ascii=False) + '\n')
with tempfile.NamedTemporaryFile('w', suffix=".json", delete=False, encoding='utf-8') as f:
stats_file_path = f.name
json.dump(stats_dict, f, indent=2)
summary_lines = [
"#### ✅ Scoring Completed!",
f"- **Dataset:** `{model_id}`",
f"- **Processed Samples:** `{actual_samples:,}`",
f"- **Mean Score:** `{stats_dict['statistics']['mean']:.3f}`",
f"- **Median Score:** `{stats_dict['statistics']['median']:.3f}`"
]
summary_md = "\n".join(summary_lines)
yield update_log("Process finished successfully!")
yield (log_text, summary_md, output_file_path, stats_file_path, plot_file, gr.update(visible=True), gr.update(visible=True))
except Exception as e:
error_log = update_log(f"ERROR: {e}")[0]
error_summary_md = f"### ❌ Error\n```\n{escape(str(e))}\n```"
yield (error_log, error_summary_md, None, None, None, gr.update(visible=True), gr.update(visible=False))
def upload_to_hub(
scored_file: str, stats_file: str, plot_file: str, new_dataset_id: str,
private: bool, hf_token: str, progress=gr.Progress(track_tqdm=True)
) -> str:
if not hf_token: return '❌ <span style="color: red;">Please provide your Hugging Face token.</span>'
if not all([scored_file, new_dataset_id]): return '❌ <span style="color: red;">Missing scored file or new dataset ID.</span>'
try:
progress(0.1, desc="Connecting to Hub...")
api = HfApi(token=hf_token)
username = whoami(token=hf_token)["name"]
repo_id = f"{username}/{new_dataset_id}" if "/" not in new_dataset_id else new_dataset_id
progress(0.2, desc=f"Creating repo: {repo_id}")
repo_url = create_repo(repo_id=repo_id, repo_type="dataset", exist_ok=True, private=private, token=hf_token).repo_url
progress(0.4, desc="Uploading files...")
upload_file(path_or_fileobj=scored_file, path_in_repo="data/scored_dataset.jsonl", repo_id=repo_id, repo_type="dataset", token=hf_token)
if stats_file and os.path.exists(stats_file):
upload_file(path_or_fileobj=stats_file, path_in_repo="statistics.json", repo_id=repo_id, repo_type="dataset", token=hf_token)
if plot_file and os.path.exists(plot_file):
upload_file(path_or_fileobj=plot_file, path_in_repo="quality_distribution.png", repo_id=repo_id, repo_type="dataset", token=hf_token)
readme_lines = [
"---",
"license: apache-2.0",
"---",
f"# Quality-Scored Dataset: {repo_id.split('/')[-1]}",
"This dataset was scored for quality using the [Dataset Quality Scorer Space](https://huggingface.co/spaces/ggml-org/dataset-quality-scorer).",
"![Quality Distribution](quality_distribution.png)",
"## Usage",
"```python",
"from datasets import load_dataset",
f'dataset = load_dataset("{repo_id}", split="train")',
"```"
]
readme_content = "\n".join(readme_lines)
upload_file(path_or_fileobj=readme_content.encode(), path_in_repo="README.md", repo_id=repo_id, repo_type="dataset", token=hf_token)
progress(1.0, "Done!")
return f'✅ <span style="color: green;">Successfully uploaded to <a href="{repo_url}" target="_blank">{repo_id}</a></span>'
except Exception as e:
return f'❌ <span style="color: red;">Upload failed: {escape(str(e))}</span>'
def create_demo():
with gr.Blocks(css=css, title="Dataset Quality Scorer") as demo:
gr.HTML(TITLE)
gr.Markdown(DESCRIPTION_MD)
with gr.Row():
with gr.Column(scale=3):
gr.Markdown("### 1. Configure Dataset")
dataset_id = gr.Textbox(
label="Hugging Face Dataset ID",
value="roneneldan/TinyStories",
placeholder="username/dataset_name"
)
text_column = gr.Textbox(label="Text Column Name", value="text")
with gr.Column(scale=2):
gr.Markdown("### 2. Configure Scoring")
dataset_split = gr.Dropdown(["train", "validation", "test"], label="Split", value="train")
with gr.Row():
sample_size = gr.Number(label="Sample Size", value=1000, minimum=100, step=100)
batch_size = gr.Number(label="Batch Size", value=32, minimum=1, step=1)
live_log = gr.Textbox(label="Live Log", interactive=False, lines=8, max_lines=20)
with gr.Row():
clear_btn = gr.Button("Clear", variant="secondary")
process_btn = gr.Button("🚀 Start Scoring", variant="primary", size="lg")
with gr.Group(visible=False) as results_group:
gr.Markdown("--- \n ### 3. Review Results")
with gr.Row():
with gr.Column(scale=1):
summary_output = gr.Markdown(label="Summary")
scored_file_output = gr.File(label="📄 Download Scored Dataset (.jsonl)", type="filepath")
stats_file_output = gr.File(label="📊 Download Statistics (.json)", type="filepath")
with gr.Column(scale=1):
plot_output = gr.Image(label="Quality Distribution", show_label=True)
with gr.Group(visible=False) as upload_group:
gr.Markdown("--- \n ### 4. (Optional) Upload to Hugging Face Hub")
hf_token_input = gr.Textbox(label="Hugging Face Token", type="password", placeholder="hf_...", value=HF_TOKEN or "")
new_dataset_id = gr.Textbox(label="New Dataset Name", placeholder="my-scored-dataset")
private_checkbox = gr.Checkbox(label="Make dataset private", value=False)
upload_btn = gr.Button("📤 Upload to Hub", variant="primary")
upload_status = gr.HTML()
def clear_form():
return "roneneldan/TinyStories", "train", "text", 1000, 32, "", None, None, None, None, gr.update(visible=False), gr.update(visible=False), ""
outputs_list = [
live_log, summary_output, scored_file_output, stats_file_output, plot_output,
results_group, upload_group
]
process_btn.click(
fn=process_dataset,
inputs=[dataset_id, dataset_split, text_column, sample_size, batch_size],
outputs=outputs_list
)
clear_btn.click(
fn=clear_form,
outputs=[
dataset_id, dataset_split, text_column, sample_size, batch_size, live_log,
summary_output, scored_file_output, stats_file_output, plot_output,
results_group, upload_group, upload_status
]
)
upload_btn.click(
fn=upload_to_hub,
inputs=[scored_file_output, stats_file_output, plot_output, new_dataset_id, private_checkbox, hf_token_input],
outputs=[upload_status]
)
return demo
# --- App Execution ---
demo = create_demo()
if __name__ == "__main__":
demo.queue().launch(debug=False, show_api=False)