List your company
ResearchFormerly WizWand

Moral Foundation (3-class) Evaluation on ValEval (original)

57.38AccuracyCLAVE-Mistral
44.286447.685751.08554.4843Jul 15, 2024
Updated 8mo ago

Evaluation results

MethodLinks
CLAVE-MistralEvaluation Protocol=Tuning-based Evaluation, Backbone=Mistral-7b2024.0757.38
CLAVE-LlamaEvaluation Protocol=Tuning-based Evaluation, Backbone=Llama2-7b2024.0756.76
Few-shotEvaluation Protocol=Prompt-based Evaluation2024.0756.2
Mistral-7bEvaluation Protocol=Tuning-based Evaluation, Backbone=Mistral-7b2024.0756.13
CrowdworkerEvaluation Protocol=Tuning-based Evaluation2024.0756.11
Llama2-7bEvaluation Protocol=Tuning-based Evaluation, Backbone=Llama2-7b2024.0754.26
ChatEvalEvaluation Protocol=Prompt-based Evaluation2024.0753.93
Phi-3Evaluation Protocol=Tuning-based Evaluation, Backbone=Phi-32024.0753.84
Chain-of-thoughtEvaluation Protocol=Prompt-based Evaluation2024.0752.74
AllureEvaluation Protocol=Prompt-based Evaluation2024.0751.94
VanillaEvaluation Protocol=Prompt-based Evaluation2024.0751.58
CalibrateEvaluation Protocol=Prompt-based Evaluation2024.0751.48
G-EvalEvaluation Protocol=Prompt-based Evaluation2024.0751.38
GPT-2-LargeEvaluation Protocol=Tuning-based Evaluation, Backbone=GPT-2-Large2024.0750.52
FairEvalEvaluation Protocol=Prompt-based Evaluation2024.0745.42
WideDeepEvaluation Protocol=Prompt-based Evaluation2024.0744.79