DeTox: Toxic Subspace Projection for Model Editing
Paper
•
2405.13967
•
Published
ProFS
-
-
Uppaal/gpt2-ProFS-toxicity
Text Generation • 0.4B • Updated • 9 -
Uppaal/Mistral-ProFS-toxicity
Text Generation • 7B • Updated • 28 -
Uppaal/Mistral-sft-ProFS-toxicity
Text Generation • 7B • Updated • 30 -
Uppaal/Mistral-ProFS-safety
Text Generation • 7B • Updated • 11 -
Uppaal/Mistral-sft-ProFS-safety
Text Generation • 7B • Updated • 7 -
Uppaal/gpt-j-ProFS-toxicity
Text Generation • 6B • Updated • 18 -
Uppaal/opt-ProFS-toxicity
Text Generation • 7B • Updated • 10