The evidence
Evaluation logs
We publish the raw measurement logs behind our figures on the scoreboard, so you can check them yourself.
Location
Where are the logs?
They are on GitHub, in the folder evaluatie. It holds the raw output of EuroEval 17.6.0 for all 10 tasks. You will find more there than in the table on the scoreboard, such as the 29.13 the model scores on conll-nl including MISC.
Which tasks do they cover?
- dbrdSentimentMCCdirect
- squad-nlReading comprehensionEMdirect
- conll-nlNamed entity recognitionmicro-F1 (no-MISC)direct
- scala-nlLinguistic acceptabilityMCCdirect
- mmlu-nlKnowledgeMCCdirect
- hellaswag-nlCommonsense reasoningMCCdirect
- wikilingua-nlSummarisationChrF3++metric change
- duidelijke-taalPlain languageMETEORno publication
- valeu-nlEuropean valuesEuropeanValuesno publication
- mbbq-nlStereotypingbias-corrected accuracyno publication
Updated: 17 September 2026
Openness
Can you use the logs?
- Evaluation logs: Apache 2.0, public, raw jsonl
- Evaluation configuration: Apache 2.0, public, pinned to 17.6.0