Released models trained by Selective DPO.
-
glorgao/SelectiveDPO-Llama3-8B-SFT-UFBinarized
Text Generation • 8B • Updated • 11 • 1 -
glorgao/SelectiveDPO-Qwen2.5-7B-SFT-UFBinarized
Text Generation • 7B • Updated • 14 • 1 -
glorgao/SelectiveDPO-Mistral-7B-SFT-UFBinarized
Text Generation • 7B • Updated • 15 -
Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples
Paper • 2502.09650 • Published
