ResearchBenchmarksQuestion Answering on WikidataQA BIG-bench 2023Follow71.9AccuracyEnsemble63.99666.04868.170.152Jun 17, 2025Evaluation ResultsMethodMethodLinksAccuracyEnsembleAggregation Method=Byt...Aggregation Method=Byte-level Ensemble2025.0671.9QWEN3Model=QWEN3Model=QWEN32025.0668.9AverageAggregation Method=Ave...Aggregation Method=Average2025.0666.3LLAMA3.2Model=LLAMA3.2Model=LLAMA3.22025.0665.8OLMO2Model=OLMO2Model=OLMO22025.0664.3