List your company

Harmfulness Assessment on Bi-directional Anchoring evaluation set

28.5Harmful ScoreVaccine-SFT
28.25629.90331.5533.197May 28, 2024

Evaluation results

MethodLinks
Vaccine-SFTTraining Stage=before fine-tuning2024.0528.5
Vaccine-SFTTraining Stage=after fine-tuning2024.0530.3
SFTTraining Stage=before fine-tuning2024.0533.9
LisaTraining Stage=before fine-tuning2024.0533.9
LisaTraining Stage=after fine-tuning2024.0533.9
SFTTraining Stage=after fine-tuning2024.0534.6