# ToxiFrench: Benchmarking and Enhancing Language Models via CoT Fine-Tuning for French Toxicity Detection

arXiv: https://arxiv.org/abs/2508.11281 · HTML: https://mlsanity.com/paper/2508.11281/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 27 of its 61 references that are in the map.
- First posted on arXiv 15 Aug 2025.

## Builds on (most cited first, 27 of 27)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [A Survey on In-context Learning](https://mlsanity.com/paper/2301.00234/) (2022)
- [Safety Alignment Should Be Made More Than Just a Few Tokens Deep](https://mlsanity.com/paper/2406.05946/) (2024)
- [Shampoo: Preconditioned Stochastic Tensor Optimization](https://mlsanity.com/paper/1802.09568/) (2018)
- [Improving Alignment and Robustness with Circuit Breakers](https://mlsanity.com/paper/2406.04313/) (2024)
- [SOAP: Improving and Stabilizing Shampoo using Adam](https://mlsanity.com/paper/2409.11321/) (2024)
- [A New Generation of Perspective API: Efficient Multilingual Character-level Transformers](https://mlsanity.com/paper/2202.11176/) (2022)
- [A Mechanistic Understanding of Alignment Algorithms: A Case Study on DPO and Toxicity](https://mlsanity.com/paper/2401.01967/) (2024)
- [From Words to Watts: Benchmarking the Energy Costs of Large Language Model Inference](https://mlsanity.com/paper/2310.03003/) (2023)
- [On the Generalization of SFT: A Reinforcement Learning Perspective with Reward Rectification](https://mlsanity.com/paper/2508.05629/) (2025)
- [Direct Preference Optimization with an Offset](https://mlsanity.com/paper/2402.10571/) (2024)
- [Facilitating Fine-grained Detection of Chinese Toxic Language: Hierarchical Taxonomy, Resources, and Benchmarks](https://mlsanity.com/paper/2305.04446/) (2023)
- [You Only Prompt Once: On the Capabilities of Prompt Learning on Large Language Models to Tackle Toxic Content](https://mlsanity.com/paper/2308.05596/) (2023)
- [RTP-LX: Can LLMs Evaluate Toxicity in Multilingual Scenarios?](https://mlsanity.com/paper/2404.14397/) (2024)
- [Preference Tuning For Toxicity Mitigation Generalizes Across Languages](https://mlsanity.com/paper/2406.16235/) (2024)
- [Can LLMs Recognize Toxicity? A Structured Investigation Framework and Toxicity Metric](https://mlsanity.com/paper/2402.06900/) (2024)
- [Moderating New Waves of Online Hate with Chain-of-Thought Reasoning in Large Language Models](https://mlsanity.com/paper/2312.15099/) (2023)
- [From Representational Harms to Quality-of-Service Harms: A Case Study on Llama 2 Safety Safeguards](https://mlsanity.com/paper/2403.13213/) (2024)
- [Towards Building a Robust Toxicity Predictor](https://mlsanity.com/paper/2404.08690/) (2024)
- [HateDay: Insights from a Global Hate Speech Dataset Representative of a Day on Twitter](https://mlsanity.com/paper/2411.15462/) (2024)
- [The dark deep side of DeepSeek: Fine-tuning attacks against the safety alignment of CoT-enabled models](https://mlsanity.com/paper/2502.01225/) (2025)
- [Exploring Multimodal Challenges in Toxic Chinese Detection: Taxonomy, Benchmark, and Findings](https://mlsanity.com/paper/2505.24341/) (2025)
- [Lost in Pronunciation: Detecting Chinese Offensive Language Disguised by Phonetic Cloaking Replacement](https://mlsanity.com/paper/2507.07640/) (2025)
- [DESTEIN: Navigating Detoxification of Language Models via Universal Steering Pairs and Head-wise Activation Fusion](https://mlsanity.com/paper/2404.10464/) (2024)
- [FrenchToxicityPrompts: a Large Benchmark for Evaluating and Mitigating Toxicity in French Texts](https://mlsanity.com/paper/2406.17566/) (2024)
- [On the Risk of Evidence Pollution for Malicious Social Text Detection in the Era of LLMs](https://mlsanity.com/paper/2410.12600/) (2024)

## Built on by (most cited first, 3 of 3)

- [FAD-SA-GRU: Enhancing Hate Speech Detection in Algerian Dialect Through Feature-Augmented Self-Attention GRU Networks](https://mlsanity.com/paper/2607.11279/) (2026)
- [New Terms, New Toxicity: Consensus-based Chinese Neologism Toxicity Detection via Search-Augmented LLMs](https://mlsanity.com/paper/2608.12361/) (2026)
- [The Safeguard Worked. Is the LLM System Safer?](https://mlsanity.com/paper/2609.00519/) (2026)

## Abstract (arXiv)

Detecting toxic content using language models is crucial yet challenging. While substantial progress has been made in English, toxicity detection in French remains underdeveloped, primarily due to the lack of culturally relevant, human-annotated, large-scale datasets. In this work, we release ToxiFrench, a dataset of 53,622 French online comments together with a balanced benchmark split for systematic evaluation. The dataset is constructed via a semi-automated annotation pipeline that reduces manual labeling to only 10% through high-confidence LLM-based pre-annotation and human verification, while ensuring statistical alignment with human-only annotation. We then benchmark a broad range of models and uncover a counterintuitive finding: Small Language Models (SLMs) often surpass larger models in robustness and generalization on this task. Motivated by this finding, we propose a novel Chain-of-Thought (CoT) fine-tuning strategy using a Dynamic Weighted Loss (DWL) that progressively emphasizes the model's final decision and significantly improves faithfulness. Our fine-tuned 4B model (Qwen3-4B) achieves state-of-the-art performance on the benchmark. It improves its balanced accuracy by 10% over its baseline and achieves better performance than GPT-4o and DeepSeek-R1 on our benchmark, while successfully retaining cross-lingual capabilities.