Open-source Russian dialogue classification model replica_classification - accurately classify dialogues into 6 types

Home

Replica Classification

Developed by TeraSpace

Russian dialogue classification model based on xlm-roberta-base, capable of categorizing user input into 6 types

Text Classification

Transformers

OtherOpen Source License:MIT #Multilingual Dialogue Classification #Intent Recognition #XLM-RoBERTa Base

Downloads 263

Release Time : 6/12/2023

Model Overview

This model is used to classify user input statements, identifying whether they belong to dialogue, questions, requests for help, user information, model information, or instructional requests

Model Features

Multi-category Classification

Accurately classifies user input into 6 different dialogue types

Built on Powerful Pre-trained Model

Constructed based on xlm-roberta-base with strong text comprehension capabilities

Russian Language Optimization

Specially optimized for Russian dialogue scenarios

Model Capabilities

Dialogue Intent Recognition

Text Classification

Natural Language Understanding

Use Cases

Dialogue Systems

Chatbot Routing

Routes conversations to different processing modules based on user input type

Improves response accuracy and user experience in dialogue systems

Customer Service Systems

Automatically identifies whether customer inquiries are consultative or requests for assistance

Optimizes customer service workflows and improves efficiency

User Analysis

User Profiling

Understands user characteristics by analyzing dialogue types

Provides data support for personalized services

🚀 TeraSpace/replica_classification

This project is based on [xlm - roberta - base](https://huggingface.co/xlm - roberta - base). It can classify different types of user utterances into six categories.

🚀 Quick Start

This model is based on [xlm - roberta - base](https://huggingface.co/xlm - roberta - base) and can classify user utterances into the following six categories:

dialog - Responds to dialogic utterances. For example, "привет".
trouble - Responds to utterances where the user tells about their problems. For example, "у меня болит зуб, мне проткнули колесо".
question - Responds to non - dialog - related questions. For example, "когда родился пушкин" or "когда я стану миллионером".
about_user - Responds when the user talks about themselves. For example, "меня зовут андрей".
about_model - Responds to questions about the assistant's personality. For example, "как тебя зовут, ты кто такая".
instruct - Responds to questions whose answers are instructions. For example, "как установить windows, как приготовить борщ".

💻 Usage Examples

Basic Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

del_symbs = ["?","!",".",","]
classes = ["dialog","trouble","question","about_user","about_model","instruct"]

device = torch.device("cuda")
model_name = 'TeraSpace/replica_classification'
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name, num_labels = len(classes)).to(device)

while True:
    text = input("=>").lower()
    for del_symb in del_symbs:
        text = text.replace(del_symb,"")
        
    inputs = tokenizer(text, truncation=True, max_length=512, padding='max_length',
                            return_tensors='pt').to(device)
    with torch.no_grad():
        logits = model(**inputs).logits
        probas = list(torch.sigmoid(logits)[0].cpu().detach().numpy())
        
    out = classes[probas.index(max(probas))]
    print(out)

📄 License

This project is licensed under the MIT license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご