Benchmark
LuxIT base SST-2 F1-macro
Multilingual benchmarks compare model performance across languages. Higher scores are better.
- Models and systems measured
- 14
- Scores recorded
- 14
- Sources
- 1
- IQ weight
- No weight published
Score distribution
- 0.483
- Mean
- 0.565
- Median
- 0.245
- Std dev
- 0.175
- 10th percentile
- 0.783
- 90th percentile
- 0.015
- Min
Results
The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.
| Model or system | Score |
|---|---|
| Apertus-8B-Instruct-2509table: 10, column: SST-2, system: apertus-8b· LuxIT (arXiv:2510.24434v4) data | 0.706 |
| EuroLLM-1.7B-Instructtable: 10, column: SST-2, system: eurollm-1.7b· LuxIT (arXiv:2510.24434v4) data | 0.267 |
| Gemma-3-12B-ITtable: 10, column: SST-2, system: gemma-3-12b· LuxIT (arXiv:2510.24434v4) data | 0.783 |
| Gemma-3-1B-ITtable: 10, column: SST-2, system: gemma-3-1b· LuxIT (arXiv:2510.24434v4) data | 0.636 |
| GLM-4-9B-0414table: 10, column: SST-2, system: glm-4-9b· LuxIT (arXiv:2510.24434v4) data | 0.823 |
| Llama-3.1-8B-Instructtable: 10, column: SST-2, system: llama-3.1-8b· LuxIT (arXiv:2510.24434v4) data | 0.711 |
| Llama-3.2-1B-Instructtable: 10, column: SST-2, system: llama-3.2-1b· LuxIT (arXiv:2510.24434v4) data | 0.175 |
| Ministral-3-3B-Instruct-2512table: 10, column: SST-2, system: ministral-3-3b· LuxIT (arXiv:2510.24434v4) data | 0.545 |
| Mistral-7B-Instruct-v0.3table: 10, column: SST-2, system: mistral-7b· LuxIT (arXiv:2510.24434v4) data | 0.652 |
| Olmo-3-7B-Instructtable: 10, column: SST-2, system: olmo-3-7b· LuxIT (arXiv:2510.24434v4) data | 0.3 |
| Phi-4table: 10, column: SST-2, system: phi-4· LuxIT (arXiv:2510.24434v4) data | 0.015 |
| Qwen2.5-0.5B-Instructtable: 10, column: SST-2, system: qwen2.5-0.5b· LuxIT (arXiv:2510.24434v4) data | 0.297 |
| Qwen2.5-1.5B-Instructtable: 10, column: SST-2, system: qwen2.5-1.5b· LuxIT (arXiv:2510.24434v4) data | 0.565 |
| Qwen2.5-7B-Instructtable: 10, column: SST-2, system: qwen2.5-7b· LuxIT (arXiv:2510.24434v4) data | 0.285 |
Data credits: Julian Valline, Cedric Lothritz, Siwen Guo and Jordi Cabot. LuxIT: A Luxembourgish Instruction Tuning Dataset from Monolingual Seed Data. arXiv:2510.24434v4. CC BY 4.0 article numeric/protocol extract; restructured, printed values unchanged. Task data, private fine-tunes, exams and weights excluded. (licence)