PSK at SemEval-2026 Task 9: Multilingual Polarization Detection Using Ensemble Gemma Models with Synthetic Data Augmentation
On the map: Synthetic data and self-improvement for LLMs → LLM synthetic data generation
Abstract (arXiv)
We present our system for SemEval-2026 Task 9: Multilingual Polarization Detection, a binary classification task spanning 22 languages. Our approach fine-tunes separate Gemma~3 models (12B and 27B parameters) per language using Low-Rank Adaptation (LoRA), augmented with synthetic data generated by a large language model (LLM). We employ three synthetic data strategies (direct generation, paraphrasing, and contrastive pair creation) using GPT-4o-mini, with a multi-stage quality filtering pipeline including embedding-based deduplication. We find that per-language threshold tuning on the development set yields 2 to 4% F1 improvements without retraining. We also use weighted ensembles of 12B and 27B model predictions with per-language strategy selection. Our final system achieves a mean macro-F1 of 0.811 across all 22 languages, ranking 2nd overall of the participating teams, with 1st place finishes in 3 languages and top-3 in 8 languages. We also find that alternative architectures (XLM-RoBERTa, Qwen3) that showed strong development set performance suffered 30 to 50% F1 drops on the test set, highlighting the importance of generalization.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in LLM synthetic data generation, one of 13 directions in Synthetic data and self-improvement for LLMs. Builds on 3 of its 13 references that are in the map. First posted on arXiv 6 May 2026.
Builds on
3 of 13 references.
References · 3
- 2019Sentence-BERT: Sentence Embeddings using Siamese BERT-Networkscited by 9,049
- 2015Improving Neural Machine Translation Models with Monolingual Datacited by 784
- 2021Improving Counterfactual Generation for Fair Hate Speech Detectioncited by 10
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2026BITS Pilani at SemEval-2026 Task 9: Structured Supervised Fine-Tuning with DPO Refinement for Polarization Detection
- 2026Better as Generators Than Classifiers: Leveraging LLMs and Synthetic Data for Low-Resource Multilingual Classification
- 2023Self-Evolved Diverse Data Sampling for Efficient Instruction Tuning
- 2025EuroLLM-9B: Technical Report
- 2026Bootstrapping Embeddings for Low Resource Languages
- 2024SpecFuse: Ensembling Large Language Models via Next-Segment Prediction
- 2025Artificial Conversations, Real Results: Fostering Language Detection with Synthetic Data
- 2024FuxiTranyu: A Multilingual Large Language Model Trained with Balanced Data