diff --git a/data/xml/2023.emnlp.xml b/data/xml/2023.emnlp.xml index 6c41896817..8bdb72d42d 100644 --- a/data/xml/2023.emnlp.xml +++ b/data/xml/2023.emnlp.xml @@ -14758,7 +14758,7 @@ The experiments were repeated and the tables and figures were updated. Changes a Fabricator: An Open Source Toolkit for Generating Labeled Training Data with Teacher <fixed-case>LLM</fixed-case>s JonasGoldeHumboldt-University of Berlin - PatrickHallerMachine Learning Group - Humboldt University of Berlin + PatrickHallerMachine Learning Group - Humboldt University of Berlin FelixHamborgUniversity of Konstanz JulianRischdeepset AlanAkbikHumboldt University of Berlin diff --git a/data/xml/2024.blackboxnlp.xml b/data/xml/2024.blackboxnlp.xml index 68315055e9..0bebd4b622 100644 --- a/data/xml/2024.blackboxnlp.xml +++ b/data/xml/2024.blackboxnlp.xml @@ -183,7 +183,7 @@ On the alignment of <fixed-case>LM</fixed-case> language generation and human language comprehension Lena SophiaBolligerUniversity of Zurich - PatrickHallerUniversity of Zurich + PatrickHallerUniversity of Zurich Lena AnnJägerUniversity of Zurich and Universität Potsdam 217-231 Previous research on the predictive power (PP) of surprisal and entropy has focused on determining which language models (LMs) generate estimates with the highest PP on reading times, and examining for which populations the PP is strongest. In this study, we leverage eye movement data on texts that were generated using a range of decoding strategies with different LMs. We then extract the transition scores that reflect the models’ production rather than comprehension effort. This allows us to investigate the alignment of LM language production and human language comprehension. Our findings reveal that there are differences in the strength of the alignment between reading behavior and certain LM decoding strategies and that this alignment further reflects different stages of language understanding (early, late, or global processes). Although we find lower PP of transition-based measures compared to surprisal and entropy for most decoding strategies, our results provide valuable insights into which decoding strategies impose less processing effort for readers. Our code is available via https://github.com/DiLi-Lab/LM-human-alignment. diff --git a/data/xml/2024.conll.xml b/data/xml/2024.conll.xml index 3a416daafc..c1a7daf61f 100644 --- a/data/xml/2024.conll.xml +++ b/data/xml/2024.conll.xml @@ -604,7 +604,7 @@ <fixed-case>B</fixed-case>aby<fixed-case>HGRN</fixed-case>: Exploring <fixed-case>RNN</fixed-case>s for Sample-Efficient Language Modeling - PatrickHallerHumboldt Universität Berlin + PatrickHallerHumboldt Universität Berlin JonasGoldeDepartment of Computer Science, Humboldt University Berlin, Humboldt Universität Berlin AlanAkbikHumboldt Universität Berlin 82-94 diff --git a/data/xml/2024.lrec.xml b/data/xml/2024.lrec.xml index 56d1229604..d84c6d3fc4 100644 --- a/data/xml/2024.lrec.xml +++ b/data/xml/2024.lrec.xml @@ -13104,7 +13104,7 @@ <fixed-case>PECC</fixed-case>: Problem Extraction and Coding Challenges - PatrickHaller + PatrickHaller JonasGolde AlanAkbik 12690–12699 diff --git a/data/xml/2024.naacl.xml b/data/xml/2024.naacl.xml index aec119e8ec..aa6ad3460a 100644 --- a/data/xml/2024.naacl.xml +++ b/data/xml/2024.naacl.xml @@ -8082,7 +8082,7 @@ <fixed-case>O</fixed-case>pinion<fixed-case>GPT</fixed-case>: Modelling Explicit Biases in Instruction-Tuned <fixed-case>LLM</fixed-case>s - PatrickHallerHumboldt Universität Berlin + PatrickHallerHumboldt Universität Berlin AnsarAynetdinovDepartment of Computer Science, Humboldt University Berlin, Humboldt Universität Berlin AlanAkbikHumboldt Universität Berlin 78-86 diff --git a/data/xml/2025.acl.xml b/data/xml/2025.acl.xml index b033804434..8c9e533d77 100644 --- a/data/xml/2025.acl.xml +++ b/data/xml/2025.acl.xml @@ -17605,7 +17605,7 @@ Leveraging In-Context Learning for Political Bias Testing of <fixed-case>LLM</fixed-case>s - PatrickHallerUniversity of Zurich + PatrickHallerUniversity of Zurich JannisVamvasUniversity of Zurich RicoSennrichUniversity of Zurich Lena AnnJägerUniversity of Zurich diff --git a/data/xml/2025.babylm.xml b/data/xml/2025.babylm.xml index d27a390e68..1f9ed754ce 100644 --- a/data/xml/2025.babylm.xml +++ b/data/xml/2025.babylm.xml @@ -181,7 +181,7 @@ Sample-Efficient Language Modeling with Linear Attention and Lightweight Enhancements - PatrickHallerHumboldt Universität Berlin + PatrickHallerHumboldt Universität Berlin JonasGoldeDepartment of Computer Science, Humboldt University Berlin, Humboldt Universität Berlin AlanAkbikHumboldt Universität Berlin 175-191 diff --git a/data/xml/2025.l2m2.xml b/data/xml/2025.l2m2.xml index 14a17123c6..03dfd22d47 100644 --- a/data/xml/2025.l2m2.xml +++ b/data/xml/2025.l2m2.xml @@ -52,7 +52,7 @@ From Data to Knowledge: Evaluating How Efficiently Language Models Learn Facts DanielChristoph MaxPlonerHumboldt Universität Berlin - PatrickHallerHumboldt Universität Berlin + PatrickHallerHumboldt Universität Berlin AlanAkbikHumboldt Universität Berlin 29-46 Sample efficiency is a crucial property of language models with practical implications for training efficiency. In real-world text, information follows a long-tailed distribution. Yet, we expect models to learn and recall frequent and infrequent facts. Sample efficient models are better equipped to handle this challenge of learning and retaining rare information without requiring excessive exposure. This study analyzes multiple models of varying architectures and sizes, all trained on the same pre-training data. By annotating relational facts with their frequencies in the training corpus, we examine how model performance varies with fact frequency. Our findings show that most models perform similarly on high-frequency facts but differ notably on low-frequency facts. This analysis provides new insights into the relationship between model architecture, size, and factual learning efficiency. diff --git a/data/xml/2025.naacl.xml b/data/xml/2025.naacl.xml index 8a0feddd02..9cba1a9e70 100644 --- a/data/xml/2025.naacl.xml +++ b/data/xml/2025.naacl.xml @@ -515,7 +515,7 @@ Familiarity: Better Evaluation of Zero-Shot Named Entity Recognition by Quantifying Label Shifts in Synthetic Training Data JonasGolde - PatrickHaller + PatrickHaller MaxPloner FabioBarth NicolaasJedema diff --git a/data/yaml/name_variants.yaml b/data/yaml/name_variants.yaml index 277227befc..a899ce287c 100644 --- a/data/yaml/name_variants.yaml +++ b/data/yaml/name_variants.yaml @@ -4015,6 +4015,11 @@ - canonical: {first: Mark, last: Hall} variants: - {first: Mark Michael, last: Hall} +- canonical: {first: Patrick, last: Haller} + id: patrick-haller + comment: HU Berlin + degree: Humboldt Universität zu Berlin + orcid: 0009-0006-0445-4765 - canonical: {first: Patrick, last: Haller} id: patrick-haller-zurich comment: University of Zurich