Sparse-Llama-3.1-8B-2of4: benchmark results
Provider: Neural Magic. Released 2024-11-20. Access: Open.
Aggregate Rating 1362 ± 33, rank #7393 of 10916 rated models, from 47 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EuroEval Spanish NLU - MLQA ES | 66.04 | Character F1 (%) | 86.8 |
| EuroEval French NLU - FQuAD | 71.95 | Character F1 (%) | 85 |
| EuroEval Portuguese NLU - MultiWikiQA PT | 72.47 | Character F1 (%) | 70 |
| EuroEval Italian NLU - SQuAD IT | 69.88 | Character F1 (%) | 68.1 |
| EuroEval Portuguese NLU - SST-2 PT | 79.91 | Matthews correlation coefficient (%) | 65.3 |
| EuroEval Faroese NLU - FoQA | 62.46 | Character F1 (%) | 59.7 |
| EuroEval Dutch NLU - DBRD | 88 | Matthews correlation coefficient (%) | 56.1 |
| EuroEval French NLU - Allocine | 92.74 | Matthews correlation coefficient (%) | 55.9 |
| EuroEval English Knowledge | 71.68 | Knowledge Average Score (%) | 46.7 |
| EuroEval Spanish NLU - ScaLA ES | 10.69 | Matthews correlation coefficient (%) | 46.5 |
| EuroEval Spanish | 37.05 | Average Score (%) | 45.5 |
| EuroEval Spanish NLU | 40.78 | NLU Average Score (%) | 45.4 |
Interactive version: theaggregate.ai/model?slug=sparse-llama-3-1-8b-2of4 · How It Works · Data refreshed daily, snapshot 2026-11-08.