Byung-Doh Oh

28 papers A* 2B 1Journal 14Unranked 11
YearRankTypeTitle / Venue / Authors
2026 J jnl
Clozing the Gap: Exploring Why Language Model Surprisal Outperforms Cloze Surprisal.
CoRR
Sathvik Nair, Byung-Doh Oh
2025 — conf
How Well Does First-Token Entropy Approximate Word Entropy as a Psycholinguistic Predictor?
IJCNLP-AACL (Short Papers)
Christian Clark, Byung-Doh Oh, William Schuler
2025 J jnl
How Well Does First-Token Entropy Approximate Word Entropy as a Psycholinguistic Predictor?
CoRR
Christian Clark, Byung-Doh Oh, William Schuler
2025 B conf
Linear Recency Bias During Training Improves Transformers' Fit to Reading Times.
COLING
Christian Clark, Byung-Doh Oh, William Schuler
2025 — conf
The Impact of Token Granularity on the Predictive Power of Language Model Surprisal.
ACL (1)
Byung-Doh Oh, William Schuler
2025 — conf
The Inverse Scaling Effect of Pre-Trained Language Model Surprisal Is Not Due to Data Leakage.
ACL (Findings)
Byung-Doh Oh, Hongao Zhu, William Schuler
2025 J jnl
The Inverse Scaling Effect of Pre-Trained Language Model Surprisal Is Not Due to Data Leakage.
CoRR
Byung-Doh Oh, Hongao Zhu, William Schuler
2025 J jnl
To model human linguistic prediction, make LLMs less superhuman.
CoRR
Byung-Doh Oh, Tal Linzen
2024 — conf
Frequency Explains the Inverse Correlation of Large Language Models' Size, Training Data Amount, and Surprisal's Fit to Reading Times.
EACL (1)
Byung-Doh Oh, Shisen Yue, William Schuler
2024 J jnl
Frequency Explains the Inverse Correlation of Large Language Models' Size, Training Data Amount, and Surprisal's Fit to Reading Times.
CoRR
Byung-Doh Oh, Shisen Yue, William Schuler
2024 A* conf
Leading Whitespaces of Language Models' Subword Vocabulary Pose a Confound for Calculating Word Probabilities.
EMNLP
Byung-Doh Oh, William Schuler
2024 J jnl
Leading Whitespaces of Language Models' Subword Vocabulary Poses a Confound for Calculating Word Probabilities.
CoRR
Byung-Doh Oh, William Schuler
2024 J jnl
Linear Recency Bias During Training Improves Transformers' Fit to Reading Times.
CoRR
Christian Clark, Byung-Doh Oh, William Schuler
2024 J jnl
The Impact of Token Granularity on the Predictive Power of Language Model Surprisal.
CoRR
Byung-Doh Oh, William Schuler
2023 — conf
Token-wise Decomposition of Autoregressive Language Model Hidden States for Analyzing Model Predictions.
ACL (1)
Byung-Doh Oh, William Schuler
2023 J jnl
Token-wise Decomposition of Autoregressive Language Model Hidden States for Analyzing Model Predictions.
CoRR
Byung-Doh Oh, William Schuler
2023 J jnl
Transformer-Based LM Surprisal Predicts Human Reading Times Best with About Two Billion Training Tokens.
CoRR
Byung-Doh Oh, William Schuler
2023 — conf
Transformer-Based Language Model Surprisal Predicts Human Reading Times Best with About Two Billion Training Tokens.
EMNLP (Findings)
Byung-Doh Oh, William Schuler
2023 J jnl
Why Does Surprisal From Larger Transformer-Based Language Models Provide a Poorer Fit to Human Reading Times?
Trans. Assoc. Comput. Linguistics
Byung-Doh Oh, William Schuler
2022 J jnl
Comparison of Structural Parsers and Neural Language Models as Surprisal Estimators.
Frontiers Artif. Intell.
Byung-Doh Oh, Christian Clark, William Schuler
2022 A* conf
Entropy- and Distance-Based Predictors From GPT-2 Attention Patterns Predict Reading Times Over and Above GPT-2 Surprisal.
EMNLP
Byung-Doh Oh, William Schuler
2022 J jnl
Entropy- and Distance-Based Predictors From GPT-2 Attention Patterns Predict Reading Times Over and Above GPT-2 Surprisal.
CoRR
Byung-Doh Oh, William Schuler
2022 J jnl
Why Does Surprisal From Larger Transformer-Based Language Models Provide a Poorer Fit to Human Reading Times?
CoRR
Byung-Doh Oh, William Schuler
2021 — conf
Character-based PCFG Induction for Modeling the Syntactic Acquisition of Morphologically Rich Languages.
EMNLP (Findings)
Lifeng Jin, Byung-Doh Oh, William Schuler
2021 — conf
Contributions of Propositional Content and Syntactic Category Information in Sentence Processing.
CMLS
Byung-Doh Oh, William Schuler
2021 — conf
Coreference-aware Surprisal Predicts Brain Response.
EMNLP (Findings)
Evan Jaffe, Byung-Doh Oh, William Schuler
2021 — conf
Surprisal Estimators for Human Reading Times Need Character Models.
ACL/IJCNLP (1)
Byung-Doh Oh, Christian Clark, William Schuler
2021 — conf
Team Ohio State at CMCL 2021 Shared Task: Fine-Tuned RoBERTa for Eye-Tracking Data Prediction.
CMLS
Byung-Doh Oh