Papers with regression analyses
Why Does Surprisal From Larger Transformer-Based Language Models Provide a Poorer Fit to Human Reading Times? (2023.tacl-1)
Copied to clipboard
| Challenge: | Existing studies have shown that larger pre-trained language models with more parameters and lower perplexity are less predictive of human reading times. |
| Approach: | They propose to use a transformer-based model with more parameters and lower perplexity to investigate why these models are less predictive of human reading times. |
| Outcome: | The results show that the larger models with more parameters and lower perplexity are less predictive of human reading times and eye-gaze durations collected during naturalistic reading. |
Willkommens-Merkel, Chaos-Johnson, and Tore-Klose: Modeling the Evaluative Meaning of German Personal Name Compounds (2024.lrec-main)
Copied to clipboard
Annerose Eichel, Tana Deeg, Andre Blessing, Milena Belosevic, Sabine Arndt-Lappe, Sabine Schulte im Walde
| Challenge: | Personal name compounds (PNCs) are compositions that refer to a person, such as Willkommens-Merkel ('Welcome-Meerkel') and a personal name such as Merkel. |
| Approach: | They propose to model 321 personal name compounds and their corresponding full names at discourse level and compare two approaches to assess whether a PNC is more positively or negatively evaluative . they further enrich data with personal, domain-specific, and extra-linguistic information and perform regression analyses revealing that factors including compound and modifier valence, domain, and political party membership influence how a pnc is evaluated. |
| Outcome: | The proposed model shows that the PNCs are perceived as more positively or negatively than their full name and that they are perceived to be more positive or negative. |
Recursive Training Loops in LLMs: How training data properties modulate distribution shift in generated data? (2025.emnlp-main)
Copied to clipboard
| Challenge: | Large language models (LLMs) are increasingly used in the creation of online content, creating feedback loops as future generations of models will be trained on this synthetic data. |
| Approach: | They propose to use large language models to create feedback loops as future models are trained on this data. |
| Outcome: | The proposed model collapse effects are found to be detrimental to the results of recursive training on human datasets. |