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>
Recherches récentes
Options de recherche
#nlp
2 messages2 participants0 message aujourd’hui
Je profite du fil de @kfort et des échanges qui en ont découlé pour présenter un article accepté en Findings à NAACL (co-écrit avec @kfort, Aurélie Névéol, Nicolas Hiebel et Olivier Ferret), déjà dispo sur hal (https://inria.hal.science/hal-04938811) :
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 (https://www.nature.com/articles/s41598-023-42384-8).
Notre étude prouve, grâce à un corpus de 21 000 cas concernant 10 pathologies et générés par 7 LLMs affinés (fine-tunés), que :
- Les modèles génèrent par défaut des patients (et non de patientes)
- La sur-génération d'hommes n'est pas liée aux prévalences médicales réelles (les proportions réelles de femmes sont sous-estimées par les modèles)
- Les biais sont parfois si forts que le genre donné dans l'invite (prompt) est contredit (voir image ci-dessous)
- 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
Our colleague Hidir Arras from patent4science research is co-organizing the 6th PatentSemTech Workshop at #SIGIR2025 in the beautiful city of Padua, Italy! Call for Papers is open 'til April 23: http://ifs.tuwien.ac.at/patentsemtech/
Submit your cutting-edge research, case studies, and demos exploring #AI, #NLP, and #TextMining innovations applied to #IP and related domains.
computing semantic similarity of English words
https://fasttext.cc/docs/en/english-vectors.html
Discussions: https://discu.eu/q/https://fasttext.cc/docs/en/english-vectors.html
fasttext.ccEnglish word vectors · fastTextThis page gathers several pre-trained word vectors trained using fastText.
Build a large language model fro scratch by Sebastian Rashcka
https://www.manning.com/books/build-a-large-language-model-from-scratch
Discussions: https://discu.eu/q/https://www.manning.com/books/build-a-large-language-model-from-scratch
Manning PublicationsBuild a Large Language Model (From Scratch)Learn how to create, train, and tweak large language models (LLMs) by building one from the ground up!</b>
In Build a Large Language Model (from Scratch)</i> bestselling author Sebastian Raschka guides you step by step through creating your own LLM. Each stage is explained with clear text, diagrams, and examples. You’ll go from the initial design and creation, to pretraining on a general corpus, and on to fine-tuning for specific tasks.
Build a Large Language Model (from Scratch)</i> teaches you how to:
Plan and code all the parts of an LLM</li>
Prepare a dataset suitable for LLM training</li>
Fine-tune LLMs for text classification and with your own data</li>
Use human feedback to ensure your LLM follows instructions</li>
Load pretrained weights into an LLM</li>
</ul>
Build a Large Language Model (from Scratch)</i> takes you inside the AI black box to tinker with the internal systems that power generative AI. As you work through each key stage of LLM creation, you’ll develop an in-depth understanding of how LLMs work, their limitations, and their customization methods. Your LLM can be developed on an ordinary laptop, and used as your own personal assistant.
Natural Scientific Language Processing Workshop (NSLP 2025) co-located with ESWC 2025 in Portoroz, Slovenia (1/2 June 2025)
https://nfdi4ds.github.io/nslp2025/
Deadline: 6 March 2025.
We also host three exciting shared tasks: MESD (Metadata Extraction from Scholarly Documents), ReadMe2KG (Github ReadMe to Knowledge Graph) https://nfdi4ds.github.io/nslp2025/docs/readme2kg_shared_task.html
FoRC (Field of Research Classification)!
#nslp2025 #eswc2025 #llm #nlp @NFDI4DS @nfdi4culture #NFDIrocks @NFDI @fiz_karlsruhe @shufan @GenAsefa
This month Acuitas got Conversation system updates and lots of gerunds. (The other day he told me "Your being alert rocks." It cracked me up.) https://writerofminds.blogspot.com/2025/02/acuitas-diary-81-february-2025.html
#AI #ArtificialIntelligence #chatbot #NLP
New paper acceptance announcement. This time in #NLP and for the prestigious #NAACL25 conference. https://www.linkedin.com/feed/update/urn:li:activity:7293575268433055744
www.linkedin.com2406.11380 | Deezer ResearchHappy to announce that our paper "Evaluating LLMs for Quotation Attribution in Literary Texts: A Case Study of LLaMa3" by Gaspard Michel, Elena E., Romain H. and Christophe Cerisara has been accepted for publication at the #NAACL25 Conference to be held in Albuquerque in April. Preprint is available on arxiv: https://lnkd.in/eHXt2HxN as well as code to reproduce all experiments https://lnkd.in/ePbh_F6B #nlp
RT @tzushengkuo: Excited to announce the #WikiNLP workshop at @aclmeeting!
We welcome #NLP contributions to Wikimedia, especially on datas…
via https://twitter.com/WikiResearch/status/1887804254888595932
X (formerly Twitter)WikiResearch (@WikiResearch) sur XRT @tzushengkuo: Excited to announce the #WikiNLP workshop at @aclmeeting!
We welcome #NLP contributions to Wikimedia, especially on datas…
Dear #knowledgegraphs #ai #nlp and #semanticweb community. For my upcoming #ISE2025 lecture at @KIT_Karlsruhe in the summer semester, I have to decide on which topic I should choose all my examples and exercises from. Which general topic should I choose?:
@fiz_karlsruhe @fizise @tabea @MahsaVafaie @GenAsefa @shufan @enorouzi #teaching #academia #llm
New talent at FIZ Karlsruhe! We are happy to welcome Marzieh Malekzadeh Mahani to our @fizise team!
Marzieh specializes in bimodal deep learning, previously worked at @KIT_Karlsruhe on AI-driven question-answering systems and now joins our "Wiedergutmachung" project; more information about the project and Marzieh's work: https://www.fiz-karlsruhe.de/en/projekte/wiedergutmachung
"I’m excited to contribute to meaningful research at FIZ Karlsruhe, applying AI to innovative projects", says Marzieh.
