Getting ready to present my work (together with my supervisor @PiekVossen ) at COLING 2022. If you are interested in the evaluation of open-domain dialogue, come to room 104 at 14:45 pm, join us online, or read the paper here: aclanthology.org/2022.ccgpk-1.3…

Our @CLTLVU colleague Selene Baez Santamaria together with @PiekVossen and Thomas Baier just won the best paper award at the Workshop on Customized Chat Grounding Persona and Knowledge. Hurray! COLING 2022

Arrived at #ICIP2022 in Bordeaux! Check out our poster on “Humans disagree with the #IoU for measuring #objectdetector localization error”, tomorrow from 10:00 to 12:30, zone 5 😉

Thanks! Khalid Al-Khatib next time in Groningen? ;) Talking about granularity: any overlap with a paper in this same session by Mattes Ruckdeschel from Leibniz-Institut für Medienforschung? Check the proceedings here: aclanthology.org/volumes/2022.a…

What can we learn about how named entities are represented in language models by substituting them for different ones and observing how predictions change? Come see my poster this Thursday December 8th @ BlackboxNLP 2022 in Abu Dhabi (and online).

Moonlit (moonlit.ai) just released the interview they had with me! Check it out if you want to get some idea of what my research on #interpretability and (legal) #argmining is about! medium.com/@moonlitai.leg…

So far today: I've been able to virtually attend co-occurring NLLP Workshop and #BlackboxNLP and listen to some inspiring talks. 🧙 Ito hard skills, I learned how to mute/unmute individual zoom chrome tabs

Some smooth work by Gabriele Sarti and colleagues from the #InDeep project in releasing a tool for interpreting sequence generation models. Keep up the development! @InseqDev

Shout-out to Hosein Mohebbi et al. from our #InDeep consortium for their awesome work "Quantifying Context Mixing in Transformers", introducing Value Zeroing as a new promising post-hoc interpretability approach for NLP! 🎉 Paper: arxiv.org/abs/2301.12971

Mapping Token-level Attributions to Interpretable Spans - presented by Jonathan Kamp

Talk at GroNLP by the invitation of Khalid Al-Khatib with my wonderful colleagues Jonathan Kamp and @myrthereuver Leiden Computer Science HHAI Conference - 🦋 @hhaiconference.bsky.social Text Mining and Retrieval Leiden @CLTLVU Interactive Intelligence Group

What a way of visiting Groningen again after so many years: by meeting GroNLP and presenting our research!

Just wrapped up our presentation at #MMNLG! Big thanks to everyone for the thought-provoking questions, and kudos to @bertugatt for the seamless organisation. Can't wait to dive deeper into this work! #SIGDIALxINLG2023

Looking forward to this! We will be giving a tutorial on the current state of Transformer interpretability methods at EACL! 🇲🇹

Welcome to #EMNLP2023 ! Here to welcome you this morning as a volunteer :)

Interested in post-hoc explanation methods? Swing by our #EMNLP2023 poster today at 10:30 to chat about their (dis)agreement and the implications of selecting the top-k most important tokens! (arxiv.org/abs/2310.05619)

Did you know that token-level differences between feature attribution methods are smoothed out if we compare them on the syntactic span level? I'll be at LREC COLING 2024 (Turin) next month to present our paper on the linguistic preferences of such methods :) arxiv.org/abs/2403.19424

Also at LREC COLING 2024 and interested in #interpretability in #nlp / #xai? I'm happy to tell you something about post-hoc attribution methods and their linguistic preferences 🔻 Poster session >> from 17:30 onwards Proceedings: aclanthology.org/2024.lrec-main…

⚠️ Citations from prompting or NLI seem plausible, but may not faithfully reflect LLM reasoning. 🏝️ MIRAGE detects context dependence in generations via model internals, producing granular and faithful RAG citations. 🚀 Demo: huggingface.co/spaces/gsarti/… Fun collab w/ Jirui Qi @EMNLP25 ✈️,