List your company

Question Answering on SQuAD 2.0 (Exclusive Catch Rate)

9.7Exclusive Catch Rate (@10%)Mistral 7B-I
4.8126.0817.358.619Apr 27, 2026

Evaluation results

MethodLinks
Mistral 7B-IProbe training=WikiText, Decoding=Greedy, Prompt=Minimal task prompt, Task-specific training=None2026.049.711
Phi-3 Mini-IProbe training=WikiText, Decoding=Greedy, Prompt=Minimal task prompt, Task-specific training=None2026.045.86.6
QWEN 2.5 7B-IProbe training=WikiText, Decoding=Greedy, Prompt=Minimal task prompt, Task-specific training=None2026.04510.9