Joris Baan (@jsbaan) header
Joris Baan (@jsbaan) profile picture

PhD candidate @AmsterdamNLP @MaiNLPLab @NLPnorth · Previously intern @allen_ai @ai2_aristo & ML engineer @DPGMediaAI

Amsterdam, The Netherlands
Joined January 2019
Photo by Joris Baan (@jsbaan)Photo by Joris Baan (@jsbaan)Photo by Joris Baan (@jsbaan)Photo by Joris Baan (@jsbaan)Photo by Joris Baan (@jsbaan)Photo by Joris Baan (@jsbaan)Photo by Joris Baan (@jsbaan)Photo by Joris Baan (@jsbaan)Photo by Joris Baan (@jsbaan)Photo by Joris Baan (@jsbaan)
⭐ WEEKLY ROUNDUP ⭐ This lineup of articles tackle the nuts and bolts of building AI apps and leveraging their potential for well-defined goals, while avoiding common pain points. 🖊️ by @onlyMarcelM, @jsbaan, @ChengzhiZhao, @carolinabento, @murtazaali_5253 + more towardsdatascience.com/build…
2
4
2,362
How can you go about designing and building an AI chat app? @jsbaan invites all tinkerers to follow along part 1 of his hands-on guide, focusing on microservice architectures and local development. towardsdatascience.com/desig…
1
10
2,347
Some highlights on recent work from my group @AmsterdamNLP to be presented at #EMNLP2023 this week, with wonderful students and collaborators. Themes include language generation, computational psycholinguistics, multimodality and more! (I'm in Singapore if you want to chat) 🧵👇
1
8
44
6,577
I'm excited to share that I was elected VP-elect for the ACL (Association for Computational Linguistics) - This is a great honor and I expect also a challenge. I will try to do my best in the years to come. Looking fwd to join the board w/ @Yuki_arase aclweb.org/portal/content/ac…
25
15
3
418
33,875
Joris Baan retweeted
📣Hi #NLProc, we are looking for new colleagues! 📣 Assistant/Associate Professorships in NLP in Copenhagen 🇩🇰: tinyurl.com/5a7cfyp4 Come join us! @ITUkbh #NLProc
10
2
18
15,802
You work in #NLProc and are interested in submitting your work on uncertainty or related topics? Consider submitting to the First Workshop on Uncertainty-Aware Natural Language Processing, co-located with #EACL2024 in Malta! Check our website and CfP here: uncertainlp.github.io/
1
5
1
20
6,676
Conformal prediction has recently sparked a lot of interest, but what if coverage is not your main concern? What if the data is not i.i.d.? Check our new preprint on “Non-Exchangeable Conformal Risk Control” with @chryssaZrv @dnnslmr @andre_t_martins: arxiv.org/abs/2310.01262. 1/N
2
16
3
74
27,303
This past week was terrible. In addition, I faced some academic rejections (3 just this past week). I figured it's a good time to discuss rejections in academia. I created a 502 website describing my recent failures: yanaiela.github.io/index502.…
19
21
1
237
31,011
Looking forward to this! We will be giving a tutorial on the current state of Transformer interpretability methods at EACL! 🇲🇹
Replying to @aclmeeting
EACL 2024 (2/3) @eaclmeeting - Transformer-specific Interpretability. Hosein Mohebbi, Jaap Jumelet, Michael Hanna, Afra Alishahi and Willem Zuidema. - Item Response Theory for Natural Language Processing. João Sedoc, John P. Lalor, Pedro Rodriguez and Jose Hernandez-Orallo.
4
38
3,840
✨Recruiting a funded PhD student interested in natural language processing, cognitive science, and psycholinguistics, keen to research adaptive language generation with me @ABDNCompSci, co-supervised by @EhudReiter. Deadline 4th November. More details: tinyurl.com/adaptNLG 💫
1
23
52
7,054
The Aristo team at AI2 is looking for interns for Summer 2024. The deadline for applying is Nov. 15th. If collaborating with top mentors on researching and building machines that can reason and learn sounds exciting to you, apply at: boards.greenhouse.io/thealle…
2
20
3
79
21,626
Joris Baan retweeted
✨How do representations evolve in Encoder-Decoder models? We introduce DecoderLens, an adaptation of the LogitLens that applies the full decoder to each encoder layer, allowing us to generate rich explanations for intermediate representations! arxiv.org/abs/2310.03686 👇Thread
1
7
3
58
9,232
A short thread on my two #EMNLP2023 submissions, which have been accepted to the main conference 🎉🧵 One on probing uncertainty in neural text generators; the other on measuring utterance predictability
1
7
47
7,730
Come intern with us @ai2_mosaic next summer!! I am recruiting interns interested in 📋evaluating generative models 🔥robustness 🔎model interpretability 🥸 privacy If this interests you, please apply by **November 6th**! Retweets/shares appreciated 🥰
The Mosaic team at AI2 is looking for interns for Summer 2024. The deadline for applying is November 6th. If building and researching machines with commonsense intelligence while working with awesome mentors sounds exciting to you, apply at boards.greenhouse.io/thealle…
50
1
267
58,683
Fantastic uncertainty quantification posters and where to find them at #ICML2023 @icmlconf, chronologically sorted 👇🧵 (spoiler: Wednesday will be wild)
2
11
2
57
17,849
How informative are the confidence intervals of the uncertainty quantification methods we use? Do they cover the ground truth values? Can we do better? In our paper “Conformalizing Machine Translation Evaluation” with @andre_t_martins we address this exact issue. [1/6]
3
25
1
99
13,013
Happy to share my #position paper to appear at @aclmeeting 2023. I argue that #language and #vision #NLProc models should master #semantic #underspecification, a key feature of human language that boosts #communication efficiency. 🧵coming soon! arxiv.org/abs/2306.05240
1
5
2
60
8,472
I'm excited to share a new paper with @coulispi and @A_Variengien: arxiv.org/abs/2305.00586! We dive into how GPT-2 implements one simple ability: greater-than. We find the circuit responsible: it has interpretable components, and generalizes to (limited) new scenarios.
1
8
7
57
11,394
Ambiguity is an intrinsic feature of language; as LMs are increasingly deployed to interface with human communication, handling ambiguity is critical🧐. So, we collect 🎉AmbiEnt, a benchmark with diverse ambiguities, challenging even GPT-4 🤖. arxiv.org/abs/2304.14399 🧵
7
100
9
442
106,247