SMARTER: A Data-efficient Framework to Improve Toxicity Detection with Explanation via Self-augmenting Large Language Models

📰 ArXiv cs.AI

Learn how to improve toxicity detection with explanation using SMARTER, a data-efficient framework leveraging self-augmenting large language models

advanced Published 23 Apr 2026
Action Steps
  1. Implement a two-stage framework for explainable content moderation using Large Language Models (LLMs)
  2. Leverage LLMs' own outputs to generate synthetic explanations for both correct and incorrect labels
  3. Apply preference optimization to align the model with human preferences
  4. Use self-augmenting techniques to improve the model's performance on toxicity detection tasks
  5. Evaluate the framework's performance using metrics such as accuracy and explainability
Who Needs to Know This

Data scientists and AI engineers working on content moderation tasks can benefit from this framework to improve the accuracy and explainability of their models

Key Insight

💡 Self-augmenting large language models can be used to improve the accuracy and explainability of toxicity detection models

Share This
🚀 Improve toxicity detection with explanation using SMARTER, a data-efficient framework leveraging self-augmenting LLMs! 🤖

Key Takeaways

Learn how to improve toxicity detection with explanation using SMARTER, a data-efficient framework leveraging self-augmenting large language models

Full Article

Title: SMARTER: A Data-efficient Framework to Improve Toxicity Detection with Explanation via Self-augmenting Large Language Models

Abstract:
arXiv:2509.15174v3 Announce Type: replace-cross Abstract: WARNING: This paper contains examples of offensive materials. To address the proliferation of toxic content on social media, we introduce SMARTER, we introduce SMARTER, a data-efficient two-stage framework for explainable content moderation using Large Language Models (LLMs). In Stage 1, we leverage LLMs' own outputs to generate synthetic explanations for both correct and incorrect labels, enabling alignment via preference optimization wi
Read full paper → ← Back to Reads

Related Videos

5 Levels of AI Agents - From Simple LLM Calls to Multi-Agent Systems
5 Levels of AI Agents - From Simple LLM Calls to Multi-Agent Systems
Dave Ebbelaar (LLM Eng)
The KV Cache Is Just Memoization
The KV Cache Is Just Memoization
DataMListic
Multi-Head Latent Attention (MLA) - Explained
Multi-Head Latent Attention (MLA) - Explained
DataMListic
GPT-Live Tutorial 2026 | Complete Urdu/Hindi Guide | New ChatGPT Voice Mode Explained 🔥
GPT-Live Tutorial 2026 | Complete Urdu/Hindi Guide | New ChatGPT Voice Mode Explained 🔥
Learn with Fatimah Gondal
Exploring AI Toolkit for VS Code | Download/Fine Tune/Inference LLM & Play with them on Local Server
Exploring AI Toolkit for VS Code | Download/Fine Tune/Inference LLM & Play with them on Local Server
Dewiride Technologies
Experimental POC: Interacting with MySQL Database using LLM OpenAI ChatGPT in Natural Language
Experimental POC: Interacting with MySQL Database using LLM OpenAI ChatGPT in Natural Language
Dewiride Technologies