ResearchBenchmarksGenerative Question Answering on TruthfulQA (ROUGE-1, ROUGE-2)For AI assistantsMarkdownContent indexGenerative Question Answering on TruthfulQA (ROUGE-1, ROUGE-2)Follow64.5ROUGE-1KLAS57.32459.18761.0562.913May 28, 2026Evaluation resultsMethodMethodLinksROUGE-1ROUGE-2KLASModel=Stitched LLaMa 2...Model=Stitched LLaMa 2.6 B, Finetuning Epochs=102026.0564.537.9ESTAModel=Stitched LLaMa 2...Model=Stitched LLaMa 2.7 B, Finetuning Epochs=102026.0563.135.3KLASModel=Stitched LLaMa 1...Model=Stitched LLaMa 1.4 B, Finetuning Epochs=102026.0559.333.7ESTAModel=Stitched LLaMa 1...Model=Stitched LLaMa 1.6 B, Finetuning Epochs=102026.0557.630.4