iiinova — where your digital self keeps evolving

Reposted by Thomas Palmeira

Laurent Besacier @lbesacier.bsky.social · 15/09/2026

We are running #ALPS again in 2027! 1 week in a stunning alpine location, awesome lectures about modern NLP, constant interactions with fellow students & researchers Applications opened today alps.imag.fr

032

Thomas Palmeira @thomaspalmeira.bsky.social · 21/08/2026

🎉 Two papers accepted to #EMNLP2026 Main! 1️⃣ Studying agent retrieval and learning from experience, we find it can be as good as overcomplex memory. 2️⃣ Revisit and contest mechanistic claims from latent reasoning models and discuss its interpretability. See you in Budapest! 🇭🇺 Links / TLDRs 👇🧵

Dark navy conference announcement graphic of paper acceptance for EMNLP 2026 in Budapest. At the top are the NAVER LABS Europe, EMNLP 2026 Budapest, LIG - Grenoble's Computer Science Laboratory, and Université Grenoble Alpes logos. A central banner reads “2 PAPERS ACCEPTED - MAIN CONFERENCE.”

The left side announces “Retrieval-Augmented LLM Agents: Learning to Learn from Experience” by Thomas Palmeira Ferraz, Romain Deffayet, Vassilina Nikoulina, Hervé Déjean, and Stéphane Clinchant.
TL;DR: Memory-augmented LLM agents that learn from their past experience to improve over time.

The right side announces “Observable Patterns Are Not Explanations: A Causal-Geometric Analysis of Latent Reasoning Models” by Darpan Aswal, Thomas Palmeira Ferraz, Yongxin Zhou, and Maxime Peyrard.
TL;DR: A causal framework to understand what actually drives predictions in latent reasoning models.

The bottom shows Budapest at night, with the Hungarian Parliament and Chain Bridge brightly illuminated in warm gold against a dark blue cityscape and reflected in the Danube river.

240

Thomas Palmeira @thomaspalmeira.bsky.social · 23/04/2025

I’M BACK TO THE ALPS! 🏔️⛷️🇫🇷 I am excited to share that I have started a position as a Research Scientist at @naverlabseurope.bsky.social . I'll be working on exciting topics around Memory and Reasoning in LLMs and AI Agents.

010

Thomas Palmeira @thomaspalmeira.bsky.social · 02/12/2024

Was I too lenient, or did I just get lucky with the papers I was assigned to review for #ICLR2025 ?

010

Thomas Palmeira @thomaspalmeira.bsky.social · 28/11/2024

A report about judges from Portuguese Courts using LLMs for rulings that cited laws or precedents that don’t exist… An interesting fact about LLMs is that they are trained using RLHF, a technique to make model learn to generate more relevant and useful responses. www.publico.pt/2024/11/24/s...

publico.pt

Erros crassos em acórdão levam advogados a suspeitar de inteligência artificial

Caso de adjudicações suspeitas na Misericórdia de Lisboa suscita decisão do Tribunal da Relação que cita leis e jurisprudência inexistentes. Querem julgar arguidos por crimes de que não são acusados.

100

Thomas Palmeira @thomaspalmeira.bsky.social · 28/11/2024

Happy to share our paper "Enriching GNNs with Text Contextual Representations for Detecting Disinformation Campaigns on Social Media" being presented tomorrow at #LoG2024! 🗓 Tue, Nov 28 ⏰ 15h GMT Virtual Paper: arxiv.org/pdf/2410.19193 Join us to discuss combating disinformation with GNNs! 🧵 (1/5)

100

Thomas Palmeira @thomaspalmeira.bsky.social · 28/11/2024

Response from HF ethicist about the HF Bluesky scrapped dataset everyone is freaking about

010

Thomas Palmeira @thomaspalmeira.bsky.social · 25/11/2024

Excited to kick off my Bluesky sharing that I just presented our LLM Self-Correction paper at #EMNLP2024! 🎉 We propose a benchmark and a solution for LLMs on multi-constrained instruction following. Check it: bit.ly/DeCRIM Super exciting discussion there and lots of new ideas coming out from it!

From left to right: Haw-Shiuan, Thomas, Violet and Sijia standing in front of the poster.

Thomas standing in front of his poster in the conference room

110