<html xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40"><head><meta http-equiv=Content-Type content="text/html; charset=utf-8"><meta name=Generator content="Microsoft Word 15 (filtered medium)"><style><!--
/* Font Definitions */
@font-face
{font-family:"Cambria Math";
panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
{font-family:DengXian;
panose-1:2 1 6 0 3 1 1 1 1 1;}
@font-face
{font-family:DengXian;
panose-1:2 1 6 0 3 1 1 1 1 1;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0cm;
text-align:justify;
text-justify:inter-ideograph;
font-size:10.5pt;
font-family:DengXian;}
.MsoChpDefault
{mso-style-type:export-only;}
/* Page Definitions */
@page WordSection1
{size:612.0pt 792.0pt;
margin:72.0pt 90.0pt 72.0pt 90.0pt;}
div.WordSection1
{page:WordSection1;}
--></style></head><body lang=ZH-CN style='word-wrap:break-word'><div class=WordSection1><p class=MsoNormal><span lang=EN-US style='font-family:"Arial",sans-serif;color:black'>Hi Adapters,<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='font-family:"Arial",sans-serif;color:black'><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US style='font-family:"Arial",sans-serif;color:black'>Natural language reflects our private lives and identities, making its privacy concerns as broad as those of real life. People start to pay more attention to privacy issues after ChatGPT emerges these days.<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='font-family:"Arial",sans-serif;color:black'><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US style='font-family:"Arial",sans-serif;color:black'>Today I’m going to talk about some privacy risks of language model for the training data, and introduce some techniques to preserve privacy. According to the paper “What Does it Mean for a Language Model to Preserve Privacy?”, I’ll also explain why it’s still challenging to preserve privacy in language models since there is a mismatch between the narrow assumptions made by popular data protection techniques, and the broadness of natural language and of privacy as a social norm.<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='font-family:"Arial",sans-serif;color:black'><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US style='font-family:"Arial",sans-serif;color:black'>Hope you find this talk interesting!<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='font-family:"Arial",sans-serif;color:black'><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US style='font-family:"Arial",sans-serif;color:black'>Best regards,<o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US style='font-family:"Arial",sans-serif;color:black'>Jingchun<o:p></o:p></span></p></div></body></html>