4 papers (co-)authored by UKP colleagues have been accepted in the Findings section at EMNLP2023, the 2023 Conference on Empirical Methods in Natural Language Processing. Congratulations to everybody involved! 💐
The conference will take place in Singapore from December 6-10, 2023. (1/🧵)
The conference will take place in Singapore from December 6-10, 2023. (1/🧵)
October 12, 2023 at 10:03 AM
4 papers (co-)authored by UKP colleagues have been accepted in the Findings section at EMNLP2023, the 2023 Conference on Empirical Methods in Natural Language Processing. Congratulations to everybody involved! 💐
The conference will take place in Singapore from December 6-10, 2023. (1/🧵)
The conference will take place in Singapore from December 6-10, 2023. (1/🧵)
In our paper, we use causal discovery algorithms to reveal intricate relationships among research entities in #NLProc: Tasks, Datasets, Methods, and Metrics.
With causal inference, we evaluate the strength of these relationships. (2/🧵) #EMNLP2023
With causal inference, we evaluate the strength of these relationships. (2/🧵) #EMNLP2023
December 7, 2023 at 9:14 AM
In our paper, we use causal discovery algorithms to reveal intricate relationships among research entities in #NLProc: Tasks, Datasets, Methods, and Metrics.
With causal inference, we evaluate the strength of these relationships. (2/🧵) #EMNLP2023
With causal inference, we evaluate the strength of these relationships. (2/🧵) #EMNLP2023
»Romanization-based Large-scale Adaptation of Multilingual Language Models« by Sukannya Purkayastha, Sebastian Ruder, Jonas Pfeiffer, Iryna Gurevych and Ivan Vulić
➡️ arxiv.org/abs/2304.08865
(3/🧵) #EMNLP2023
➡️ arxiv.org/abs/2304.08865
(3/🧵) #EMNLP2023
October 12, 2023 at 10:06 AM
»Romanization-based Large-scale Adaptation of Multilingual Language Models« by Sukannya Purkayastha, Sebastian Ruder, Jonas Pfeiffer, Iryna Gurevych and Ivan Vulić
➡️ arxiv.org/abs/2304.08865
(3/🧵) #EMNLP2023
➡️ arxiv.org/abs/2304.08865
(3/🧵) #EMNLP2023
✨ Excited to share our Chain-of-Questions paper #EMNLP2023: we develop a framework that trains *one T5 model* to robustly answer multistep questions by generating and answering sub-questions. Outperforms ChatGPT on DROP, HotpotQA and their contrast/adversarial sets.
October 10, 2023 at 10:57 PM
✨ Excited to share our Chain-of-Questions paper #EMNLP2023: we develop a framework that trains *one T5 model* to robustly answer multistep questions by generating and answering sub-questions. Outperforms ChatGPT on DROP, HotpotQA and their contrast/adversarial sets.
December 5, 2023 at 6:55 PM
We just submitted to arXiv our paper (w/ @carlosg.bsky.social and Diego Roca): "4 and 7-bit Labeling for Projective and Non-Projective Dependency Trees", which we will be presenting at #EMNLP2023, in Singapore!
arxiv.org/abs/2310.14319
arxiv.org/abs/2310.14319
4 and 7-bit Labeling for Projective and Non-Projective Dependency Trees
We introduce an encoding for parsing as sequence labeling that can represent any projective dependency tree as a sequence of 4-bit labels, one per word. The bits in each word's label represent (1)...
arxiv.org
October 24, 2023 at 3:25 PM
We just submitted to arXiv our paper (w/ @carlosg.bsky.social and Diego Roca): "4 and 7-bit Labeling for Projective and Non-Projective Dependency Trees", which we will be presenting at #EMNLP2023, in Singapore!
arxiv.org/abs/2310.14319
arxiv.org/abs/2310.14319
With this aim, we introduce two data selection strategies to detect representative sentences, both unsupervised & semi-supervised.
For the latter, we propose an annotation schema to obtain relevant training samples. (6/🧵) #EMNLP2023
For the latter, we propose an annotation schema to obtain relevant training samples. (6/🧵) #EMNLP2023
December 8, 2023 at 10:25 AM
With this aim, we introduce two data selection strategies to detect representative sentences, both unsupervised & semi-supervised.
For the latter, we propose an annotation schema to obtain relevant training samples. (6/🧵) #EMNLP2023
For the latter, we propose an annotation schema to obtain relevant training samples. (6/🧵) #EMNLP2023
How to improve the response diversity of dialog systems? 🛠️ Use a diffusion model to build prior distributions with Dior C-VAE!
Discover more about our hierarchical C-VAE in our latest #EMNLP2023 paper. Learn more in this 🧵. (1/8)
➡️ arxiv.org/abs/2305.15025
Discover more about our hierarchical C-VAE in our latest #EMNLP2023 paper. Learn more in this 🧵. (1/8)
➡️ arxiv.org/abs/2305.15025
December 6, 2023 at 12:42 PM
How to improve the response diversity of dialog systems? 🛠️ Use a diffusion model to build prior distributions with Dior C-VAE!
Discover more about our hierarchical C-VAE in our latest #EMNLP2023 paper. Learn more in this 🧵. (1/8)
➡️ arxiv.org/abs/2305.15025
Discover more about our hierarchical C-VAE in our latest #EMNLP2023 paper. Learn more in this 🧵. (1/8)
➡️ arxiv.org/abs/2305.15025
To appear EMNLP2023: simplifying text involves explaining and elaborating concepts. Using QUDs in a question generation -> answering pipeline leads to much better generation of such elaborations!
arxiv.org/abs/2305.10387
w/ @yatingwu.bsky.social Will Sheffield @kmahowald.bsky.social
arxiv.org/abs/2305.10387
w/ @yatingwu.bsky.social Will Sheffield @kmahowald.bsky.social
October 25, 2023 at 11:57 PM
To appear EMNLP2023: simplifying text involves explaining and elaborating concepts. Using QUDs in a question generation -> answering pipeline leads to much better generation of such elaborations!
arxiv.org/abs/2305.10387
w/ @yatingwu.bsky.social Will Sheffield @kmahowald.bsky.social
arxiv.org/abs/2305.10387
w/ @yatingwu.bsky.social Will Sheffield @kmahowald.bsky.social
You can find our paper here:
📃https://arxiv.org/abs/2311.00408
and our code here:
💻https://github.com/UKPLab/AdaSent
Check out the work of our authors Yongxin Huang, Kexin Wang, Sourav Dutta, Raj Nath Patel, Goran Glavaš and Iryna Gurevych! (7/🧵) #EMNLP2023
📃https://arxiv.org/abs/2311.00408
and our code here:
💻https://github.com/UKPLab/AdaSent
Check out the work of our authors Yongxin Huang, Kexin Wang, Sourav Dutta, Raj Nath Patel, Goran Glavaš and Iryna Gurevych! (7/🧵) #EMNLP2023
December 9, 2023 at 10:46 AM
You can find our paper here:
📃https://arxiv.org/abs/2311.00408
and our code here:
💻https://github.com/UKPLab/AdaSent
Check out the work of our authors Yongxin Huang, Kexin Wang, Sourav Dutta, Raj Nath Patel, Goran Glavaš and Iryna Gurevych! (7/🧵) #EMNLP2023
📃https://arxiv.org/abs/2311.00408
and our code here:
💻https://github.com/UKPLab/AdaSent
Check out the work of our authors Yongxin Huang, Kexin Wang, Sourav Dutta, Raj Nath Patel, Goran Glavaš and Iryna Gurevych! (7/🧵) #EMNLP2023
LLMs have the potential to make education personalized and more accessible 🎓 !
Yet, such systems have been hampered by a lack of large high-quality datasets. (2/🧵) #EMNLP2023.
Yet, such systems have been hampered by a lack of large high-quality datasets. (2/🧵) #EMNLP2023.
December 5, 2023 at 11:43 AM
LLMs have the potential to make education personalized and more accessible 🎓 !
Yet, such systems have been hampered by a lack of large high-quality datasets. (2/🧵) #EMNLP2023.
Yet, such systems have been hampered by a lack of large high-quality datasets. (2/🧵) #EMNLP2023.
Some insightful recent works report model consistency across bundles of related instances. But since this naturally increases with accuracy, how should these consistency scores at different accuracies be compared? Our paper for BlackboxNLP@EMNLP2023: arxiv.org/abs/2310.13781
October 31, 2023 at 8:34 PM
Some insightful recent works report model consistency across bundles of related instances. But since this naturally increases with accuracy, how should these consistency scores at different accuracies be compared? Our paper for BlackboxNLP@EMNLP2023: arxiv.org/abs/2310.13781
Grounded on research papers from ACL Anthology between 1979 and 2022, our framework unveils the primary drivers behind research trends in Natural Language Processing. (3/🧵) #EMNLP2023
December 7, 2023 at 9:15 AM
Grounded on research papers from ACL Anthology between 1979 and 2022, our framework unveils the primary drivers behind research trends in Natural Language Processing. (3/🧵) #EMNLP2023
A group photo from the poster presentation of »AmbiFC: Fact-Checking Ambiguous Claims with Evidence«, co-authored by our colleague Max Glockner, Ieva Staliūnaitė, James Thorne, Gisela Vallejo, Andreas Vlachos and Iryna Gurevych. #EMNLP2023
December 11, 2023 at 10:39 AM
A group photo from the poster presentation of »AmbiFC: Fact-Checking Ambiguous Claims with Evidence«, co-authored by our colleague Max Glockner, Ieva Staliūnaitė, James Thorne, Gisela Vallejo, Andreas Vlachos and Iryna Gurevych. #EMNLP2023
In citation text generation, given papers A and B, we generate text that cites paper A in the context of paper B. The task has many flavors. Sentence- or paragraph-level? Abstractive or extractive? One or multiple cited papers? This makes it hard to compare systems across studies. (3/🧵) #EMNLP2023
December 6, 2023 at 1:01 PM
In citation text generation, given papers A and B, we generate text that cites paper A in the context of paper B. The task has many flavors. Sentence- or paragraph-level? Abstractive or extractive? One or multiple cited papers? This makes it hard to compare systems across studies. (3/🧵) #EMNLP2023
妹がEMNLP2023のナップサックでフェスに出かけたのわろ
June 23, 2024 at 12:03 AM
妹がEMNLP2023のナップサックでフェスに出かけたのわろ
Findings from #WMT23
Our Chat4 friend is in the winning group across tasks
Most submissions still use from scratch training
Less constrained (low resource) submissions than before
More test suit submissions!
Low resource results TBD (tech issue)
#EMNLP2023 #WMT #neuralEmpty #LLMs
Our Chat4 friend is in the winning group across tasks
Most submissions still use from scratch training
Less constrained (low resource) submissions than before
More test suit submissions!
Low resource results TBD (tech issue)
#EMNLP2023 #WMT #neuralEmpty #LLMs
December 6, 2023 at 1:41 AM
Findings from #WMT23
Our Chat4 friend is in the winning group across tasks
Most submissions still use from scratch training
Less constrained (low resource) submissions than before
More test suit submissions!
Low resource results TBD (tech issue)
#EMNLP2023 #WMT #neuralEmpty #LLMs
Our Chat4 friend is in the winning group across tasks
Most submissions still use from scratch training
Less constrained (low resource) submissions than before
More test suit submissions!
Low resource results TBD (tech issue)
#EMNLP2023 #WMT #neuralEmpty #LLMs
Happening now! Come to the Law Law Land in Aquarius 2 for 🐤BoF session of 🪶 NLP on legal text ⚖️ #EMNLP2023
March 11, 2025 at 10:35 PM
Happening now! Come to the Law Law Land in Aquarius 2 for 🐤BoF session of 🪶 NLP on legal text ⚖️ #EMNLP2023
Registered for EMNLP2023? Consider chatting with Natalia Flechas Manrique (11am Dec7 gathertown/blackboxnlp). She'll be talking about this work 👇
#PsychSciSky #CogSci #CogPsyc #NLP #AI #neuroskyence #NeuroAI #compneuro #PsychSciSky
#PsychSciSky #CogSci #CogPsyc #NLP #AI #neuroskyence #NeuroAI #compneuro #PsychSciSky
We're presenting, "Enhancing Interpretability using Human Similarity Judgements to Prune Word Embeddings", at the BlackboxNLP workshop in EMNLP2023. We improve prediction of human similarity judgments for co-hyponyms and provide interpretability of pruned embeddings. #MLSky #CogSci Link: t.ly/NTRRS
December 6, 2023 at 6:53 AM
Registered for EMNLP2023? Consider chatting with Natalia Flechas Manrique (11am Dec7 gathertown/blackboxnlp). She'll be talking about this work 👇
#PsychSciSky #CogSci #CogPsyc #NLP #AI #neuroskyence #NeuroAI #compneuro #PsychSciSky
#PsychSciSky #CogSci #CogPsyc #NLP #AI #neuroskyence #NeuroAI #compneuro #PsychSciSky
Honored my paper was accepted to Findings of #EMNLP2023! Many psycholinguistics studies use LLMs to estimate the probability of words in context. But LLMs process statistically derived subword tokens, while human processing doesn't. Does this matter? (w/Philip Resnik) 🧵
arxiv.org/abs/2310.17774
arxiv.org/abs/2310.17774
November 2, 2023 at 10:20 PM
Honored my paper was accepted to Findings of #EMNLP2023! Many psycholinguistics studies use LLMs to estimate the probability of words in context. But LLMs process statistically derived subword tokens, while human processing doesn't. Does this matter? (w/Philip Resnik) 🧵
arxiv.org/abs/2310.17774
arxiv.org/abs/2310.17774
It was observed that the performance of ICL depends heavily on the selection of the exemplars.
Jonathan Tonglet et al. show how this combinatorial optimization problem can be formulated as a Knapsack Integer Linear program and optimized efficiently with deterministic solvers. (2/🧵) #EMNLP2023
Jonathan Tonglet et al. show how this combinatorial optimization problem can be formulated as a Knapsack Integer Linear program and optimized efficiently with deterministic solvers. (2/🧵) #EMNLP2023
November 8, 2023 at 10:20 AM
It was observed that the performance of ICL depends heavily on the selection of the exemplars.
Jonathan Tonglet et al. show how this combinatorial optimization problem can be formulated as a Knapsack Integer Linear program and optimized efficiently with deterministic solvers. (2/🧵) #EMNLP2023
Jonathan Tonglet et al. show how this combinatorial optimization problem can be formulated as a Knapsack Integer Linear program and optimized efficiently with deterministic solvers. (2/🧵) #EMNLP2023
We also illustrate how our semantic retrieval pipeline provides interpretability of the symptom estimation, highlighting the most relevant sentences. (8/🧵) #EMNLP2023
December 8, 2023 at 10:25 AM
We also illustrate how our semantic retrieval pipeline provides interpretability of the symptom estimation, highlighting the most relevant sentences. (8/🧵) #EMNLP2023
Are you interested in word lengths and natural language’s efficiency? If yes, check out our new #EMNLP2023 paper! It has everything you need: drama, suspense, a new derivation of Zipf’s law, an update to Piantadosi et al’s classic word length paper, transformers... 😄
arxiv.org/abs/2312.03897
arxiv.org/abs/2312.03897
December 8, 2023 at 5:46 PM
Are you interested in word lengths and natural language’s efficiency? If yes, check out our new #EMNLP2023 paper! It has everything you need: drama, suspense, a new derivation of Zipf’s law, an update to Piantadosi et al’s classic word length paper, transformers... 😄
arxiv.org/abs/2312.03897
arxiv.org/abs/2312.03897
🔄 Data and Parameter Efficiency: The Magic Behind Adaptation 🧙♂️
How do we adapt mPLMs to romanized and non-romanized corpora of 14 low-resource languages? We explore a plethora of strategies – data and parameter-efficient ones! (5/🧵) #EMNLP2023
How do we adapt mPLMs to romanized and non-romanized corpora of 14 low-resource languages? We explore a plethora of strategies – data and parameter-efficient ones! (5/🧵) #EMNLP2023
December 7, 2023 at 8:30 AM
🔄 Data and Parameter Efficiency: The Magic Behind Adaptation 🧙♂️
How do we adapt mPLMs to romanized and non-romanized corpora of 14 low-resource languages? We explore a plethora of strategies – data and parameter-efficient ones! (5/🧵) #EMNLP2023
How do we adapt mPLMs to romanized and non-romanized corpora of 14 low-resource languages? We explore a plethora of strategies – data and parameter-efficient ones! (5/🧵) #EMNLP2023
In our paper, we propose a framework to generate personalized dialogues by pairing human teachers with a LLM prompted to represent common student errors. (3/🧵) #EMNLP2023
December 5, 2023 at 11:43 AM
In our paper, we propose a framework to generate personalized dialogues by pairing human teachers with a LLM prompted to represent common student errors. (3/🧵) #EMNLP2023