shakedown.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
A community for live music fans with roots in the jam scene. Shakedown Social is run by a team of volunteers (led by @clifff and @sethadam1) and funded by donations.

Administered by:

Server stats:

282
active users

#nlp

3 posts3 participants0 posts today
John :au: :hb:<p>My <a href="https://fairdinkum.one/tags/schadenfreude" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>schadenfreude</span></a> side wants the <a href="https://fairdinkum.one/tags/NationalParty" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NationalParty</span></a> to demand to have the <a href="https://fairdinkum.one/tags/LNP" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LNP</span></a> referred to as the <a href="https://fairdinkum.one/tags/NLP" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NLP</span></a> &amp; claim the right to be the major party in the <a href="https://fairdinkum.one/tags/COALition" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>COALition</span></a>. 🤗 </p><p><a href="https://fairdinkum.one/tags/auspol" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>auspol</span></a></p>
Harald Sack<p>Next step in our NLP timeline is Claude Elwood Shannon, who already laid the foundations for statistical language modeling by recognising the relevance of n-grams to model properties of language and predicting the likelihood of word sequences.</p><p>C.E. Shannon ""A Mathematical Theory of Communication" (1948) <a href="https://web.archive.org/web/19980715013250/http://cm.bell-labs.com/cm/ms/what/shannonday/shannon1948.pdf" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">web.archive.org/web/1998071501</span><span class="invisible">3250/http://cm.bell-labs.com/cm/ms/what/shannonday/shannon1948.pdf</span></a></p><p><a href="https://sigmoid.social/tags/ise2025" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ise2025</span></a> <a href="https://sigmoid.social/tags/nlp" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>nlp</span></a> <a href="https://sigmoid.social/tags/lecture" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>lecture</span></a> <a href="https://sigmoid.social/tags/languagemodel" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>languagemodel</span></a> <a href="https://sigmoid.social/tags/informationtheory" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>informationtheory</span></a> <a href="https://sigmoid.social/tags/historyofscience" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>historyofscience</span></a> <span class="h-card" translate="no"><a href="https://sigmoid.social/@enorouzi" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>enorouzi</span></a></span> <span class="h-card" translate="no"><a href="https://fedihum.org/@tabea" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>tabea</span></a></span> <span class="h-card" translate="no"><a href="https://fedihum.org/@sourisnumerique" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>sourisnumerique</span></a></span> <span class="h-card" translate="no"><a href="https://wisskomm.social/@fiz_karlsruhe" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>fiz_karlsruhe</span></a></span> <span class="h-card" translate="no"><a href="https://sigmoid.social/@fizise" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>fizise</span></a></span></p>
Miguel Afonso Caetano<p>"Asking scientists to identify a paradigm shift, especially in real time, can be tricky. After all, truly ground-shifting updates in knowledge may take decades to unfold. But you don’t necessarily have to invoke the P-word to acknowledge that one field in particular — natural language processing, or NLP — has changed. A lot.</p><p>The goal of natural language processing is right there on the tin: making the unruliness of human language (the “natural” part) tractable by computers (the “processing” part). A blend of engineering and science that dates back to the 1940s, NLP gave Stephen Hawking a voice, Siri a brain and social media companies another way to target us with ads. It was also ground zero for the emergence of large language models — a technology that NLP helped to invent but whose explosive growth and transformative power still managed to take many people in the field entirely by surprise.</p><p>To put it another way: In 2019, Quanta reported on a then-groundbreaking NLP system called BERT without once using the phrase “large language model.” A mere five and a half years later, LLMs are everywhere, igniting discovery, disruption and debate in whatever scientific community they touch. But the one they touched first — for better, worse and everything in between — was natural language processing. What did that impact feel like to the people experiencing it firsthand?</p><p>Quanta interviewed 19 current and former NLP researchers to tell that story. From experts to students, tenured academics to startup founders, they describe a series of moments — dawning realizations, elated encounters and at least one “existential crisis” — that changed their world. And ours."</p><p><a href="https://www.quantamagazine.org/when-chatgpt-broke-an-entire-field-an-oral-history-20250430/" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="ellipsis">quantamagazine.org/when-chatgp</span><span class="invisible">t-broke-an-entire-field-an-oral-history-20250430/</span></a></p><p><a href="https://tldr.nettime.org/tags/AI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AI</span></a> <a href="https://tldr.nettime.org/tags/GenerativeAI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>GenerativeAI</span></a> <a href="https://tldr.nettime.org/tags/ChatGPT" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ChatGPT</span></a> <a href="https://tldr.nettime.org/tags/NLP" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NLP</span></a> <a href="https://tldr.nettime.org/tags/OralHistory" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>OralHistory</span></a> <a href="https://tldr.nettime.org/tags/LLMs" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LLMs</span></a> <a href="https://tldr.nettime.org/tags/Chatbots" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Chatbots</span></a></p>
Erika Varis Doggett<p>At <a href="https://mas.to/tags/NAACL" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NAACL</span></a> this week and I’m delighted to see the name change to “Nations of the Americas” as well as the special theme for this year of multi- and cross-culturalism in <a href="https://mas.to/tags/NLP" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NLP</span></a>. </p><p><a href="https://mas.to/tags/NLProc" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NLProc</span></a> <a href="https://mas.to/tags/AI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AI</span></a> <a href="https://mas.to/tags/LLMs" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LLMs</span></a> <a href="https://mas.to/tags/LanguageModels" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LanguageModels</span></a> <a href="https://mas.to/tags/CompLing" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>CompLing</span></a> <a href="https://mas.to/tags/ComputationalLinguistics" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ComputationalLinguistics</span></a></p>
AmyFou 🕊️<p>Friends and colleagues - please enjoy this talk by Rolando Coto Solano about the goals and methods behind our 'Advancing Indigenous Language Technologies' initiative (<a href="https://ailt.arizona.edu" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="">ailt.arizona.edu</span><span class="invisible"></span></a>)</p><p><a href="https://lingo.lol/tags/language" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>language</span></a> <a href="https://lingo.lol/tags/technology" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>technology</span></a> <a href="https://lingo.lol/tags/languageRevitalization" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>languageRevitalization</span></a> <a href="https://lingo.lol/tags/languageDocumentation" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>languageDocumentation</span></a> <a href="https://lingo.lol/tags/NLP" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NLP</span></a> </p><p><span class="h-card" translate="no"><a href="https://a.gup.pe/u/linguistics" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>linguistics</span></a></span> </p><p><a href="https://www.youtube.com/live/owC7nO3ls8Y?si=klmffzfdXyNoMpMU" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="ellipsis">youtube.com/live/owC7nO3ls8Y?s</span><span class="invisible">i=klmffzfdXyNoMpMU</span></a></p>
David J. Atkinson<p><span class="h-card" translate="no"><a href="https://sciencemastodon.com/@brianvastag" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>brianvastag</span></a></span> Which is ironic since I’ve heard it before. The circle is coming around. I wouldn’t be mad at your friends. They are in good company.</p><p>The app “Eliza” was created by Joseph Weizenbaum c. 1967. He was a critic of early <a href="https://c.im/tags/AI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AI</span></a> and wanted to show how easily it could be faked. The Eliza app was scripted for various scenarios (sound familiar yet?). The most famous one simulated a <a href="https://c.im/tags/psychotherapist" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>psychotherapist</span></a>. People tried it and got hooked. Weizenbaum was proved correct - intelligence was really easy to fake. Some people protested vehemently when told the experiment was over, saying it was the best therapy they ever had! Maybe so. What surprised everyone was how people reacted to Eliza. Weizenbaum pointed out that Eliza had no knowledge and it didn’t understand anything people said to it. Eliza composed its replies based entirely on <a href="https://c.im/tags/scripts" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>scripts</span></a> and syntactic rules. Nobody really cared. And thus began the great schism in AI research, particularly natural language processing aka <a href="https://c.im/tags/NLP" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NLP</span></a>. The syntactics people went one way, producing <a href="https://c.im/tags/chatbots" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>chatbots</span></a>, and today’s <a href="https://c.im/tags/LLMs" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LLMs</span></a> like <a href="https://c.im/tags/chatgpt" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>chatgpt</span></a> , and the semantics people (later, myself included) went another, producing many automated knowledge-based problem-solving techniques that today are embedded in thousands of applications.</p>
Benjamin Han<p>Are you passionate about language technologies? About translating creative ideas among languages? Be a machine learning engineer for Apple Services Localization Engineering team to build models and algorithms to power our service offerings at scale!</p><p>Please apply directly via the link:</p><p><a href="https://jobs.apple.com/en-us/details/200599851/sr-applied-ml-engineer-apple-services-localization-engineering?team=SFTWR" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">jobs.apple.com/en-us/details/2</span><span class="invisible">00599851/sr-applied-ml-engineer-apple-services-localization-engineering?team=SFTWR</span></a></p><p><a href="https://sigmoid.social/tags/jobs" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>jobs</span></a> <a href="https://sigmoid.social/tags/ML" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ML</span></a> <a href="https://sigmoid.social/tags/NLP" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NLP</span></a> <a href="https://sigmoid.social/tags/NLProc" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NLProc</span></a> <a href="https://sigmoid.social/tags/MT" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>MT</span></a> <a href="https://sigmoid.social/tags/MachineTranslation" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>MachineTranslation</span></a></p>
Harald Sack<p>What does it mean "to know" something? Have you ever thought about it? We tried to make our students think about it in this week's first <a href="https://sigmoid.social/tags/ise2025" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ise2025</span></a> lecture.</p><p><a href="https://sigmoid.social/tags/kit200" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>kit200</span></a> <a href="https://sigmoid.social/tags/lecture" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>lecture</span></a> <a href="https://sigmoid.social/tags/knowledge" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>knowledge</span></a> <a href="https://sigmoid.social/tags/philosophy" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>philosophy</span></a> <a href="https://sigmoid.social/tags/knowledgerepresentation" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>knowledgerepresentation</span></a> <a href="https://sigmoid.social/tags/understanding" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>understanding</span></a> <a href="https://sigmoid.social/tags/semweb" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>semweb</span></a> <a href="https://sigmoid.social/tags/knowledgegraph" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>knowledgegraph</span></a> <a href="https://sigmoid.social/tags/nlp" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>nlp</span></a> <span class="h-card" translate="no"><a href="https://wisskomm.social/@fiz_karlsruhe" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>fiz_karlsruhe</span></a></span> <span class="h-card" translate="no"><a href="https://sigmoid.social/@fizise" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>fizise</span></a></span> <span class="h-card" translate="no"><a href="https://sigmoid.social/@enorouzi" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>enorouzi</span></a></span> <span class="h-card" translate="no"><a href="https://fedihum.org/@sourisnumerique" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>sourisnumerique</span></a></span></p>
Harald Sack<p>One of the central topics discussed in today's first ISE 2025 lecture is "Knowledge". How can we define knowledge? How does it differ from data, information, or wisdom? How does the process of "understanding" work? Welcome to "The Art of Understanding", which is the title of this lecture...</p><p><a href="https://sigmoid.social/tags/ise2025" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ise2025</span></a> <a href="https://sigmoid.social/tags/semweb" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>semweb</span></a> <a href="https://sigmoid.social/tags/semanticweb" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>semanticweb</span></a> <a href="https://sigmoid.social/tags/AI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AI</span></a> <a href="https://sigmoid.social/tags/nlp" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>nlp</span></a> <a href="https://sigmoid.social/tags/philosophy" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>philosophy</span></a> <a href="https://sigmoid.social/tags/lecture" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>lecture</span></a> <span class="h-card" translate="no"><a href="https://sigmoid.social/@fizise" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>fizise</span></a></span> <span class="h-card" translate="no"><a href="https://wisskomm.social/@fiz_karlsruhe" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>fiz_karlsruhe</span></a></span> <span class="h-card" translate="no"><a href="https://fedihum.org/@tabea" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>tabea</span></a></span> <span class="h-card" translate="no"><a href="https://sigmoid.social/@enorouzi" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>enorouzi</span></a></span> <span class="h-card" translate="no"><a href="https://fedihum.org/@sourisnumerique" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>sourisnumerique</span></a></span></p>
petersuber<p>Update. In the fields of <a href="https://fediscience.org/tags/NLP" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NLP</span></a> and <a href="https://fediscience.org/tags/LIS" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LIS</span></a>, "papers with different <a href="https://fediscience.org/tags/gender" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>gender</span></a> compositions achieve varying numbers of citations, with mixed-gender collaborations gradually obtaining higher average citation counts compared to same-gender collaborations."<br><a href="https://doi.org/10.1016/j.joi.2025.101662" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">doi.org/10.1016/j.joi.2025.101</span><span class="invisible">662</span></a></p>
h-pom<p><a href="https://social.vivaldi.net/tags/film" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>film</span></a> <a href="https://social.vivaldi.net/tags/filmcamera" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>filmcamera</span></a> <a href="https://social.vivaldi.net/tags/filmphotography" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>filmphotography</span></a> <a href="https://social.vivaldi.net/tags/snapshot" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>snapshot</span></a> <a href="https://social.vivaldi.net/tags/photography" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>photography</span></a> <a href="https://social.vivaldi.net/tags/streetphotography" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>streetphotography</span></a> <a href="https://social.vivaldi.net/tags/slr" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>slr</span></a> <a href="https://social.vivaldi.net/tags/kodak" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>kodak</span></a> <a href="https://social.vivaldi.net/tags/portra" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>portra</span></a> <a href="https://social.vivaldi.net/tags/portra400" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>portra400</span></a> <a href="https://social.vivaldi.net/tags/mediumformat" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>mediumformat</span></a> <a href="https://social.vivaldi.net/tags/nlp" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>nlp</span></a> <a href="https://social.vivaldi.net/tags/japan" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>japan</span></a> <a href="https://social.vivaldi.net/tags/schneider" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>schneider</span></a> <a href="https://social.vivaldi.net/tags/aposymmar" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>aposymmar</span></a> <a href="https://social.vivaldi.net/tags/makro" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>makro</span></a> <a href="https://social.vivaldi.net/tags/90mmf4" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>90mmf4</span></a></p>
Kaa :linux: 🧶<p>J'ai fait une intervention au siège du CNRS mercredi sur les biais et plus généralement les pbs d'évaluation des LLMs. <br>Pour celles et ceux que ça intéresse, les diapos sont ici :<br><a href="https://members.loria.fr/KFort/files/fichiers_cours/KarenFort_LLMEvaluation.pdf" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">members.loria.fr/KFort/files/f</span><span class="invisible">ichiers_cours/KarenFort_LLMEvaluation.pdf</span></a><br><a href="https://social.sciences.re/tags/LLM" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LLM</span></a> <a href="https://social.sciences.re/tags/evaluation" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>evaluation</span></a> <a href="https://social.sciences.re/tags/IA" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>IA</span></a> <a href="https://social.sciences.re/tags/nlp" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>nlp</span></a> <a href="https://social.sciences.re/tags/tal" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>tal</span></a></p>
Fanny<p>Je profite du fil de <span class="h-card" translate="no"><a href="https://social.sciences.re/@kfort" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>kfort</span></a></span> et des échanges qui en ont découlé pour présenter un article accepté en Findings à NAACL (co-écrit avec <span class="h-card" translate="no"><a href="https://social.sciences.re/@kfort" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>kfort</span></a></span>, Aurélie Névéol, Nicolas Hiebel et Olivier Ferret), déjà dispo sur hal (<a href="https://inria.hal.science/hal-04938811" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="">inria.hal.science/hal-04938811</span><span class="invisible"></span></a>) :</p><p>De plus en plus de facs de médecine songent à faire plancher les étudiant·es sur des cas cliniques générés par des modèles de langue (LLMs). Pourtant, on sait que ces LLMs sont biaisés, et que les biais des modèles peuvent créer/amplifier les biais d'humains (<a href="https://www.nature.com/articles/s41598-023-42384-8" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="ellipsis">nature.com/articles/s41598-023</span><span class="invisible">-42384-8</span></a>).</p><p>Notre étude prouve, grâce à un corpus de 21 000 cas concernant 10 pathologies et générés par 7 LLMs affinés (fine-tunés), que :</p><p>- Les modèles génèrent par défaut des patients (et non de patientes)<br>- La sur-génération d'hommes n'est pas liée aux prévalences médicales réelles (les proportions réelles de femmes sont sous-estimées par les modèles) <br>- Les biais sont parfois si forts que le genre donné dans l'invite (prompt) est contredit (voir image ci-dessous)<br>- Les femmes et les personnes trans sont plus à risque d'être impactées par ces biais, qui peuvent se traduire de manière très concrète : erreurs de diagnostics, errance médicale, traitements inadaptés, tabou, mégenrage, essentialisme biologique</p><p><a href="https://social.sciences.re/tags/llm" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>llm</span></a> <a href="https://social.sciences.re/tags/nlp" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>nlp</span></a> <a href="https://social.sciences.re/tags/ai" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ai</span></a> <a href="https://social.sciences.re/tags/tal" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>tal</span></a> <a href="https://social.sciences.re/tags/naacl" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>naacl</span></a> <a href="https://social.sciences.re/tags/bias" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>bias</span></a> <a href="https://social.sciences.re/tags/sexism" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>sexism</span></a></p>
h-pom<p><a href="https://social.vivaldi.net/tags/film" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>film</span></a> <a href="https://social.vivaldi.net/tags/filmcamera" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>filmcamera</span></a> <a href="https://social.vivaldi.net/tags/filmphotography" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>filmphotography</span></a> <a href="https://social.vivaldi.net/tags/snapshot" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>snapshot</span></a> <a href="https://social.vivaldi.net/tags/photography" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>photography</span></a> <a href="https://social.vivaldi.net/tags/streetphotography" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>streetphotography</span></a> <a href="https://social.vivaldi.net/tags/slr" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>slr</span></a> <a href="https://social.vivaldi.net/tags/kodak" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>kodak</span></a> <a href="https://social.vivaldi.net/tags/portra" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>portra</span></a> <a href="https://social.vivaldi.net/tags/portra400" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>portra400</span></a> <a href="https://social.vivaldi.net/tags/mediumformat" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>mediumformat</span></a> <a href="https://social.vivaldi.net/tags/nlp" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>nlp</span></a> <a href="https://social.vivaldi.net/tags/japan" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>japan</span></a> <a href="https://social.vivaldi.net/tags/schneider" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>schneider</span></a> <a href="https://social.vivaldi.net/tags/aposymmar" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>aposymmar</span></a> <a href="https://social.vivaldi.net/tags/makro" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>makro</span></a> <a href="https://social.vivaldi.net/tags/90mmf4" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>90mmf4</span></a></p>
Compsci Weekly<p>computing semantic similarity of English words</p><p><a href="https://fasttext.cc/docs/en/english-vectors.html" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">fasttext.cc/docs/en/english-ve</span><span class="invisible">ctors.html</span></a></p><p>Discussions: <a href="https://discu.eu/q/https://fasttext.cc/docs/en/english-vectors.html" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">discu.eu/q/https://fasttext.cc</span><span class="invisible">/docs/en/english-vectors.html</span></a></p><p><a href="https://mastodon.social/tags/compsci" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>compsci</span></a> <a href="https://mastodon.social/tags/nlp" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>nlp</span></a></p>
Cedric<p>New blog post on the Vulnerability-Lookup blog:</p><p>LLMs + Vulnerability-Lookup: What We’re Testing and Where We’re Headed</p><p><a href="https://www.vulnerability-lookup.org/2025/02/26/exploring-llm-in-vulnerability-lookup/" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="ellipsis">vulnerability-lookup.org/2025/</span><span class="invisible">02/26/exploring-llm-in-vulnerability-lookup/</span></a></p><p><a href="https://social.circl.lu/tags/NLP" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NLP</span></a> <a href="https://social.circl.lu/tags/LLM" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LLM</span></a> <a href="https://social.circl.lu/tags/Vulnerability" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Vulnerability</span></a> <a href="https://social.circl.lu/tags/vulnerabilitylookup" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>vulnerabilitylookup</span></a> <a href="https://social.circl.lu/tags/opensource" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>opensource</span></a> <a href="https://social.circl.lu/tags/cybersecurity" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>cybersecurity</span></a></p>
Michael Franke<p>Position opening: <a href="https://fediscience.org/tags/CompLing" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>CompLing</span></a>, <a href="https://fediscience.org/tags/NLP" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NLP</span></a>, <a href="https://fediscience.org/tags/CogSci" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>CogSci</span></a></p><p>Junior Group Leader / PostDoc <br>(4y, German E14 salary scale) @ Uni Tübingen</p><p>- no teaching obligations<br>- PhD co-supervision<br>- dynamic interdisciplinary team in one of Germany’s top research universities</p><p><a href="https://tinyurl.com/jhxb4wu2" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="">tinyurl.com/jhxb4wu2</span><span class="invisible"></span></a></p>
Kaa :linux: 🧶<p>Comme j'ai dû lire la note qui fait le buzz "La Fièvre parlementaire : ce monde où<br>l’on catche !" (<a href="https://www.cepremap.fr/depot/2025/01/2025-01-Fievre-Parlementaire.pdf" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="ellipsis">cepremap.fr/depot/2025/01/2025</span><span class="invisible">-01-Fievre-Parlementaire.pdf</span></a>) pour en évaluer le sérieux scientifique et qu'au final ça ne servira pas, voici mes retours de taliste (n'hésitez-pas à ajouter vos propres remarques) : </p><p>- Ce ne sont ni des talistes, ni des linguistes, il y a donc des différences disciplinaires, à la fois sur le fond et sur la forme, ce n'est donc pas facile à lire</p><p>- Je vois 3 endroits où ils disent avoir utilisé des technos de TAL :<br> - 1 - pour le filtrage du début (classification binaire) : émotionnel / rationnel<br> - 2 - pour l'analyse de sentiments (émotions)<br> - 3 - pour l'indicateur de polarisation<br>Il est mentionné aussi un "degré de conflictualité des discours", mais je ne vois pas à quoi cela correspond je dois dire.</p><p>- Les technos utilisées ne sont pas claires : pour 1, on sait que c'est (Chat)GPT 4o mini, mais pour le reste, on ne sait pas exactement. </p><p>- Les paramètres des expés ne sont pas explicites : combien de fois ont-ils fait passer les prompts (sachant que ces technos ne sont pas déterministes, il faut faire tourner plusieurs fois et faire des moyennes, par ex) ? quelle température pour les expés (plus c bas, plus c déterministe, mais jamais totalement) ?</p><p>Si on part sur du ChatGPT 4o mini pour la tâche 2 (3 étant des embeddings, c'est forcément autre chose), alors la question se pose des performances sur la tâche (ici, analyse de sentiments). Si on regarde sur le site d'OpenAI (<a href="https://openai.com/index/gpt-4o-mini-advancing-cost-efficient-intelligence/" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">openai.com/index/gpt-4o-mini-a</span><span class="invisible">dvancing-cost-efficient-intelligence/</span></a>), on peut lire qu'il a été testé sur différents benchmarks et on y trouve les résultats. Disons qu'on les croit. <br>Quels sont ces benchmarks ? Eh bien on les retrouve ici un peu décrits : <a href="https://wielded.com/blog/gpt-4o-benchmark-detailed-comparison-with-claude-and-gemini" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">wielded.com/blog/gpt-4o-benchm</span><span class="invisible">ark-detailed-comparison-with-claude-and-gemini</span></a>.<br>Où l'on peut voir nulle part l'analyse de sentiments.<br>Encore moins en français.</p><p>Donc, l'**article présente toute une série d'expériences (sans les détailler suffisamment) réalisées à partir d'un outil dont on ne connaît pas les performances sur la tâche réalisée, encore moins pour la langue en question**. 1/n</p><p><a href="https://social.sciences.re/tags/tal" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>tal</span></a> <a href="https://social.sciences.re/tags/nlp" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>nlp</span></a> <a href="https://social.sciences.re/tags/politique" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>politique</span></a> <a href="https://social.sciences.re/tags/chatgpt" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>chatgpt</span></a> <br><span class="h-card" translate="no"><a href="https://mastodon.zaclys.com/@MarCandea" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>MarCandea</span></a></span> <span class="h-card" translate="no"><a href="https://social.sciences.re/@tract_linguistes" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>tract_linguistes</span></a></span> <span class="h-card" translate="no"><a href="https://wikis.world/@Poslovitch" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>Poslovitch</span></a></span> <span class="h-card" translate="no"><a href="https://mediapart.social/@mediapart" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>mediapart</span></a></span> <a href="https://social.sciences.re/tags/media" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>media</span></a> (voir le pouet de <span class="h-card" translate="no"><a href="https://mastodon.zaclys.com/@MarCandea" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>MarCandea</span></a></span> ci-dessous)</p>
pettter<p>I might as well do another <a href="https://social.accum.se/tags/introduction" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>introduction</span></a> specifically for the <a href="https://social.accum.se/tags/academic" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>academic</span></a> side of this here fediverse:</p><p>Coming from <a href="https://social.accum.se/tags/theoreticalCS" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>theoreticalCS</span></a> (with applications in <a href="https://social.accum.se/tags/NLP" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NLP</span></a>) to doing <a href="https://social.accum.se/tags/digitalhumanities" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>digitalhumanities</span></a> (computational <a href="https://social.accum.se/tags/musicology" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>musicology</span></a>), I've now landed in <a href="https://social.accum.se/tags/ResponsibleAI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ResponsibleAI</span></a>. Specifically, I'm interested in exploring <a href="https://social.accum.se/tags/AntiCapitalistAI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AntiCapitalistAI</span></a>, both sharpening existing critiques of current AI practise by confronting capital and exploring inherent politics of technologies, and finding better ones for a socialist world.</p>
ApaulD<p><a href="https://aus.social/tags/australia" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>australia</span></a> (neo)Liberal Party <a href="https://aus.social/tags/nlp" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>nlp</span></a> talks popularism but votes for the .1% <a href="https://aus.social/tags/auspol" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>auspol</span></a></p>