Ruspamns Big

🚀 NeuroSpaceX/ruSpamNS_big

This is a binary classification model for spam detection in Russian Telegram messages, based on the DeepPavlov/distilrubert-small-cased-conversational model.

🚀 Quick Start

The ruSpamNS_big model is designed for binary classification of spam in Russian Telegram messages. It can effectively identify spam messages in various Telegram scenarios, including chats, channels, and groups.

• Supported Platforms: Telegram (Chats, Channels, Groups)
• Base Model: Based on ruBERT, fine - tuned on Russian spam detection datasets
• Number of Samples: 4 million samples from Russian Telegram messages
• Author: @NeuroSpaceX
• Telegram Bot: @ruSpamNS_bot

✨ Features

• High - Performance Spam Detection: Capable of accurately identifying spam messages in Russian Telegram conversations.
• Fine - Tuned Model: Fine - tuned on a large amount of Russian Telegram data to improve detection accuracy.

💻 Usage Examples

Basic Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

# Replace with your actual Hugging Face token
hf_token = "hf_–≤–∞—à_–ø–µ—Ä—Å–æ–Ω–∞–ª—å–Ω—ã–π_—Ç–æ–∫–µ–Ω"

# Model name
model_name = "NeuroSpaceX/ruSpamNS_big"
tokenizer = AutoTokenizer.from_pretrained(model_name, token=hf_token)
model = AutoModelForSequenceClassification.from_pretrained(
    model_name, 
    token=hf_token, 
    attn_implementation="eager"
)
model.eval()

# Input text
text = "–ü—Ä–∏–≤–µ—Ç, –Ω—É–∂–Ω—ã —Å–æ—Ç—Ä—É–¥–Ω–∏–∫–∏ –Ω–∞ —É–¥–∞–ª–µ–Ω–∫—É, 300 –¥–æ–ª–ª–∞—Ä–æ–≤ –≤ –Ω–µ–¥–µ–ª—é"

# Tokenize the input text
inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)

# Make prediction
with torch.no_grad():
    logits = model(**inputs).logits
    spam_score = torch.sigmoid(logits).item()
    is_spam = spam_score >= 0.5

print(f"Is spam: {is_spam} (Spam score: {spam_score:.2%})")

📚 Documentation

The model is fine - tuned on the NeuroSpaceX/v17 dataset, which contains a large number of Russian Telegram messages. It can be used to detect spam in various Telegram scenarios, such as:

• Chat messages
• Channel posts
• Group messages

📄 License

This model is licensed under the CC - BY - NC - ND - 3.0 license.

• Attribution: You must give appropriate credit, provide a link to the license, and indicate if changes were made.
• Non - Commercial: You may not use the material for commercial purposes.
• No Derivatives: If you remix, transform, or build upon the material, you may not distribute the modified material.
• ShareAlike: If you distribute the material, you must license it under the same terms as the original.

© NeuroSpaceX 2025

⚠️ Important Note

To use this model, you need to obtain permission from @NeuroSpaceX. You need to provide information such as your full name and usage intent, and agree to the terms of use and attribution requirements.

💡 Usage Tip

When using this model, make sure your input text is in Russian and from the Telegram environment for better performance.