LA Leaderboard - BHTCv2 Basque: leaderboard
Topic classification of Basque news texts (BHTCv2, from BasqueGLUE). From the LA Leaderboard, which evaluates models on Spanish, Catalan, Basque and Galician tasks with the lm-evaluation-harness; F1 (%), higher is better. Each model result file is a separate run.
Metric: F1 (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 69 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | salamandra-7B | 46.28 | #1446 |
| 2 | Gemma 2 9B (IT) | 39.64 | #795 |
| 3 | Llama 3.1 8B Instruct | 38.24 | #1017 |
| 4 | gemma-3-4B-pt Base | 38.24 | #1110 |
| 5 | Gemma 2 9B | 37.27 | #924 |
| 6 | Llama 3.1 8B | 37 | #1130 |
| 7 | salamandra-7B-instruct | 35.49 | #1392 |
| 8 | salamandra-2B | 32.47 | #1520 |
| 9 | Llama 3.2 3B Instruct | 30.96 | #1327 |
| 10 | Llama 3.2 3B | 29.45 | #1376 |
| 11 | aya-expanse-8B | 28.48 | #914 |
| 12 | Qwen3-8B-Base | 28.32 | #626 |
| 13 | salamandra-2B-instruct | 26.65 | |
| 14 | Gemma 3 4B (IT) | 26.27 | #997 |
| 15 | occiglot-7B-eu5-instruct | 26 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=la-leaderboard-bhtcv2-basque · How It Works · Data refreshed daily, snapshot 2026-11-05.