Emoji-Affect Decodability (Bangla): leaderboard
Metric: Macro-F1 (%; emoji-affect decodability: 5-fold cross-validated l2-regularised logistic regression on sublinear TF-IDF features of each model's emoji graphemes, predicting the source emotion label of the 4,206 single-label Bangla sentences over the five emotions shared by the three corpora; one generation per sentence at temperature 0.7; majority-class floor 8.7). Source: arxiv.org. Saturation forecast: Around 2029. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.0 Flash | 65.7 |
| 2 | Grok 4 Fast | 65.5 |
| 3 | Gemma 3 27B (IT) | 65.3 |
| 4 | Mistral Large 3 | 63.9 |
| 5 | Qwen 3 VL 235B A22B | 59.7 |
| 6 | Claude 3 Haiku | 58.4 |
| 7 | GPT-4.1 Nano | 56.2 |
| 8 | DeepSeek V3.2 | 56 |
Interactive version: theaggregate.ai/benchmark?slug=emoji-affect-decodability-bangla · How It Works · Data refreshed daily, snapshot 2026-09-26.