tanaos-topic-classification-v1: A small but performant topic classification model

This model was created by Tanaos with the Artifex Python library.

This is a topic classification model based on FacebookAI/roberta-base and fine-tuned on a synthetic dataset to classify text into one of 15 different intent categories:

Topic	Description
politics	elections, policies, scandals, ideology.
health	physical health, mental health, fitness, diets, medical advice.
technology	gadgets, software, AI, cybersecurity.
entertainment	movies, TV shows, music, celebrities, streaming platforms.
money_finance	investing, budgeting, crypto, real estate.
relationships_dating	romance, breakups, marriage, family drama.
education_learning	schools, universities, self-study, online courses.,
work_careers	job hunting, workplace culture, remote work, career advice.
science	research, space, climate, biology, physics, chemistry and the scientific method.
society_culture	identity, inequality, norms, language, and society.
gaming	video games, esports, hardware, mods, and gaming culture.
lifestyle_hobbies	travel, food, fashion, DIY, productivity systems.
sports	teams, athletes, events, scores, and sports culture.
automotive	cars, motorcycles, reviews, maintenance, and industry news.
other	miscellaneous topics not covered by the other categories.

How to Use

Via the Artifex library (pip install artifex)

from artifex import Artifex

topic_classification = Artifex().topic_classification

print(topic_classification("What do you think about the latest AI advancements?"))
# >>> [{'label': 'technology', 'score': 0.9910}]

Via the Transformers library

from transformers import pipeline

clf = pipeline("text-classification", model="tanaos/tanaos-topic-classification-v1")

print(clf("What do you think about the latest AI advancements?"))
# >>> [{'label': 'technology', 'score': 0.9910}]

Model Description

• Base model: FacebookAI/roberta-base
• Task: Text classification (topic classification)
• Languages: English
• Fine-tuning data: A synthetic, custom dataset of 10,000 utterances, each belonging to one of 15 different topic categories.

Training Details

This model was trained using the Artifex Python library

pip install artifex

by providing the following instructions and generating 10,000 synthetic training samples:

from artifex import Artifex


topic_classification = Artifex().topic_classification

topic_classification.train(
    domain="general",
    classes={
        "politics": "elections, policies, scandals, ideology",
        "health": "physical health, mental health, fitness, diets, medical advice.",
        "technology": "gadgets, software, AI, cybersecurity.",
        "entertainment": "movies, TV shows, music, celebrities, streaming platforms.",
        "money_finance": "investing, budgeting, crypto, real estate.",
        "relationships_dating": "romance, breakups, marriage, family drama.",
        "education_learning": "schools, universities, self-study, online courses.",
        "work_careers": "job hunting, workplace culture, remote work, career advice.",
        "science": "research, space, climate, biology, physics, chemistry and the scientific method.",
        "society_culture": "identity, inequality, norms, language, and society.",
        "gaming": "video games, esports, hardware, mods, and gaming culture.",
        "lifestyle_hobbies": "travel, food, fashion, DIY, productivity systems.",
        "sports": "teams, athletes, events, scores, and sports culture.",
        "automotive": "cars, motorcycles, reviews, maintenance, and industry news.",
        "other": "miscellaneous topics not covered by the other categories."
    },
    num_samples=10000
)

Intended Uses

This model is intended to:

• Classify conversations, reviews, articles, or any text into one of the predefined topic categories.
• Be used in applications such as chatbots, content categorization, and sentiment analysis.
• Serve as a lightweight alternative for topic classification tasks.

Not intended for:

• Use cases requiring extremely high accuracy or domain-specific knowledge without further fine-tuning.