List your company

Question Answering on SQuAD 2.0 (Factual Accuracy & Relevance)

1.66Factual AccuracyEmbGen
1.26481.36741.471.5726May 19, 2026

Evaluation results

MethodLinks
EmbGenSize=20M2026.051.662.062.842.9928.8
InstructLabSize=5M2026.051.652.142.92.9828
InstructLabSize=20M2026.051.652.092.94328
EmbGenSize=5M2026.051.62.062.88323.2
Knowledge-InstructSize=20M2026.051.42.022.682.986.4
Knowledge-InstructSize=5M2026.051.391.972.562.983.6
EntiGraphSize=20M2026.051.372.042.242.914
EntiGraph Squad 5MSize=5M2026.051.291.952.352.93.6
Baseline: llama 8B InstructSize=5M2026.051.281.992.32.890.8
Baseline: llama 8B InstructSize=20M2026.051.281.992.32.890.8