Polarization Detection on SemEval Task 9 2026 (test)
88.92Macro-F1language-adaptive modeling framework
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| language-adaptive modeling frameworkLanguage=Telugu (tel)2026.04 | 88.92 | 90.53 | -0.0161 | 7 | |
| language-adaptive modeling frameworkLanguage=Burmese (mya)2026.04 | 88.74 | 89.13 | -0.0039 | 2 | |
| language-adaptive modeling frameworkLanguage=Bengali (ben)2026.04 | 84.46 | 86.25 | -0.0179 | 7 | |
| language-adaptive modeling frameworkLanguage=Persian (fas)2026.04 | 83.08 | 83.48 | -0.004 | 3 | |
| language-adaptive modeling frameworkLanguage=Arabic (arb)2026.04 | 82.94 | 84.88 | -0.0194 | 12 | |
| language-adaptive modeling frameworkLanguage=Hausa (hau)2026.04 | 81.68 | 83.36 | -0.0168 | 6 | |
| Best of Sub 1 + Sub 4Submission ID=52026.05 | 81.2 | — | — | — | |
| Ensemble + thresholdSubmission ID=42026.05 | 81.1 | — | — | — | |
| language-adaptive modeling frameworkLanguage=Polish (pol)2026.04 | 80.61 | 84.31 | -0.037 | 13 | |
| language-adaptive modeling frameworkLanguage=Hindi (hin)2026.04 | 80.32 | 82.36 | -0.0204 | 17 | |
| Best Gemma-12B per langSubmission ID=12026.05 | 79.7 | — | — | — | |
| language-adaptive modeling frameworkLanguage=English (eng)2026.04 | 79.58 | 82.52 | -0.0294 | 18 | |
| language-adaptive modeling frameworkLanguage=Odia (ori)2026.04 | 79.03 | 82.55 | -0.0352 | 14 | |
| language-adaptive modeling frameworkLanguage=Swahili (swa)2026.04 | 78.23 | 81.13 | -0.029 | 15 | |
| language-adaptive modeling frameworkLanguage=Amharic (amh)2026.04 | 77.44 | 80.02 | -0.0258 | 10 | |
| language-adaptive modeling frameworkLanguage=Spanish (spa)2026.04 | 76.32 | 80.3 | -0.0398 | 22 | |
| Hybrid (XLM-R, Qwen3)Submission ID=22026.05 | 66.5 | — | — | — |