--- title: "Yejin Choi — papers, page 3 · Stanford University" description: "Yejin Choi, Stanford University: 92 papers on Elamur Research (2005–2026), h-index 84 — affiliations, topics, co-authors and who cites them…" canonical: "https://research.elamur.ai/free/en/authors/yejin-choi/page/3/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/authors/yejin-choi/page/3/) [home](https://md.research.elamur.ai/free/en/) / [authors](https://md.research.elamur.ai/free/en/authors/) / [Yejin Choi](https://md.research.elamur.ai/free/en/authors/yejin-choi/) / page 3 # Yejin Choi · page 3 92 papers in our library · 2005–2026 · page 3 of 4 [The profile, affiliations and co-authors →](https://md.research.elamur.ai/free/en/authors/yejin-choi/) ## Papers The most cited first (OpenAlex and our library). [← prev](https://md.research.elamur.ai/free/en/authors/yejin-choi/page/2/) [1](https://md.research.elamur.ai/free/en/authors/yejin-choi/) [2](https://md.research.elamur.ai/free/en/authors/yejin-choi/page/2/) **3** [4](https://md.research.elamur.ai/free/en/authors/yejin-choi/page/4/) [next →](https://md.research.elamur.ai/free/en/authors/yejin-choi/page/4/) 51. **[NeuroLogic A*esque Decoding: Constrained Text Generation with Lookahead Heuristics](https://md.research.elamur.ai/free/en/papers/lu-2022-neurologic-esque-decoding-constrained-text/)** *(2022 · 140 citations)* Left-to-right decoders such as beam search and top-k sampling pick each token based only on the prefix generated so far. 52. **[GO FIGURE: A Meta Evaluation of Factuality in Summarization](https://md.research.elamur.ai/free/en/papers/gabriel-2021-go-figure-meta-evaluation-factuality/)** *(2021 · 130 citations)* Neural summarizers often produce fluent text that misstates or invents facts. Many new automatic metrics aim to detect such errors, but there has been no systematic way to check whether these metrics actually work. GO FIGURE is a meta-evaluation framework that tests factuality metrics against five requirements… 53. **[Do Androids Laugh at Electric Sheep? Humor “Understanding” Benchmarks from The New Yorker Caption Contest](https://md.research.elamur.ai/free/en/papers/hessel-2023-do-androids-laugh-electric-sheep/)** *(2023 · 112 citations)* The paper asks whether today's AI systems grasp humor or merely produce joke-like text. To probe this, the authors build three benchmarks from 14 years of the weekly New Yorker Cartoon Caption Contest, each harder than the last. The first is matching : pick the caption actually written for a cartoon out of five… 54. **[Reframing Human-AI Collaboration for Generating Free-Text Explanations](https://md.research.elamur.ai/free/en/papers/wiegreffe-2021-reframing-human-ai-collaboration-generating/)** *(2021 · 111 citations)* Free-text explanations for NLP decisions are usually produced by models trained on crowdsourced explanation corpora. 55. **[SwiftSage: A Generative Agent with Fast and Slow Thinking for Complex Interactive Tasks](https://md.research.elamur.ai/free/en/papers/lin-2023-swiftsage-generative-agent-fast-slow/)** *(2023 · 110 citations)* Agents in text-based simulated environments struggle with long, multi-step tasks. They must plan over many steps, remember context, break goals into subgoals and recover when something goes wrong. Imitation-trained models pick up action patterns well but plan and recover poorly. Prompted LLMs plan well but have… 56. **[We're Afraid Language Models Aren't Modeling Ambiguity](https://md.research.elamur.ai/free/en/papers/liu-2023-we-re-afraid-language-models/)** *(2023 · 110 citations)* The paper asks whether pretrained language models can notice when a sentence has more than one plausible meaning and separate those meanings. 57. **[NeuroLogic Decoding: (Un)supervised Neural Text Generation with Predicate Logic Constraints](https://md.research.elamur.ai/free/en/papers/lu-2021-neurologic-decoding-un-supervised-neural/)** *(2021 · 110 citations)* Large pretrained language models finetuned on task data still often ignore lexical constraints. A recipe generator, for example, may skip listed ingredients or invent new ones, and adding much more finetuning data barely helps. The authors introduce NeuroLogic Decoding, an inference-time algorithm that works with… 58. **[WildBench: Benchmarking LLMs with Challenging Tasks from Real Users in the Wild](https://md.research.elamur.ai/free/en/papers/lin-2024-wildbench-benchmarking-llms-challenging-tasks/)** *(2024 · 105 citations)* Existing automatic LLM benchmarks poorly reflect what real users actually ask. MT-bench is tiny, AlpacaEval is dominated by easy information-seeking prompts, and ArenaHard leans heavily toward coding. WildBench addresses this by drawing hard, naturally distributed tasks from WildChat, a log of roughly one million… 59. **[Personalized Soups: Personalized Large Language Model Alignment via Post-hoc Parameter Merging](https://md.research.elamur.ai/free/en/papers/jang-2023-personalized-soups-personalized-large-language/)** *(2023 · 102 citations)* Standard RLHF trains a policy against a single reward model that reflects averaged annotator taste. This ignores that individual users may want different, even opposing, qualities in a response. The authors frame this as Reinforcement Learning from Personalized Human Feedback (RLPHF) and treat it as a Multi-Objective… 60. **[Simulating Action Dynamics with Neural Process Networks](https://md.research.elamur.ai/free/en/papers/bosselut-2017-simulating-action-dynamics-neural-process/)** *(2017 · 102 citations)* Procedural texts such as recipes often leave the consequences of actions unstated. For example, baking implies that an ingredient becomes hot and ends up in the oven. Most NLU models cannot infer such implicit effects. The authors propose Neural Process Networks (NPN), which read instructions sentence by sentence and… 61. **[CommonsenseQA 2.0: Exposing the Limits of AI through Gamification](https://md.research.elamur.ai/free/en/papers/talmor-2022-commonsenseqa-2-0-exposing-limits/)** *(2022 · 94 citations)* Modern pre-trained language models reach human-level scores on many NLU benchmarks. Yet they remain fragile on adversarial inputs and still make basic commonsense mistakes, partly because static crowdsourced datasets contain exploitable artifacts. The authors turn data collection into a game. Players write yes/no… 62. **[WildTeaming at Scale: From In-the-Wild Jailbreaks to (Adversarially) Safer Language Models](https://md.research.elamur.ai/free/en/papers/jiang-2024-wildteaming-scale-wild-jailbreaks-adversarially/)** *(2024 · 92 citations)* Frontier LLMs can still be jailbroken, and open safety-training data is scarce, since even open-weight models rarely publish theirs. 63. **[It's not Rocket Science: Interpreting Figurative Language in Narratives](https://md.research.elamur.ai/free/en/papers/chakrabarty-2022-it-s-not-rocket-science/)** *(2022 · 91 citations)* Transformer-based language models handle literal text well but often fail on figurative expressions. Idioms and similes are hard because their meaning cannot be built from the meanings of their parts. The authors frame this as a story-continuation problem, following the Story Cloze setup. They took short passages… 64. **[Do Membership Inference Attacks Work on Large Language Models?](https://md.research.elamur.ai/free/en/papers/duan-2024-membership-inference-attacks-work-large/)** *(2024 · 87 citations)* This paper asks whether standard membership inference attacks (MIAs) can tell if a given text was in an LLM's pre-training data. 65. **[Back to the Future: Unsupervised Backprop-based Decoding for Counterfactual and Abductive Commonsense Reasoning](https://md.research.elamur.ai/free/en/papers/qin-2020-back-future-unsupervised-backprop-based/)** *(2020 · 86 citations)* Abductive reasoning asks what plausibly happened between an observed past and an observed future. Counterfactual reasoning asks how a story's later events should change when an earlier event is altered. Both require a model to condition on context on both sides of the text it generates. Standard left-to-right… 66. **[Rainier: Reinforced Knowledge Introspector for Commonsense Question Answering](https://md.research.elamur.ai/free/en/papers/liu-2022-rainier-reinforced-knowledge-introspector-commonsense/)** *(2022 · 82 citations)* Commonsense question answering often improves when a model receives extra background knowledge alongside the question. 67. **[Tuning Language Models by Proxy](https://md.research.elamur.ai/free/en/papers/liu-2024-tuning-language-models-proxy/)** *(2024 · 79 citations)* Fine-tuning large language models is expensive, and it is impossible for closed models whose weights are hidden. 68. **[OpinionFinder: A System for Subjectivity Analysis](https://md.research.elamur.ai/free/en/papers/wilson-2005-opinionfinder-system-subjectivity-analysis/)** *(2005 · 78 citations)* OpinionFinder is a demonstration system that automatically detects subjective language in English text, meaning opinions, emotions, speculations and other private states. 69. **[Minding Language Models' (Lack of) Theory of Mind: A Plug-and-Play Multi-Character Belief Tracker](https://md.research.elamur.ai/free/en/papers/sclar-2023-minding-language-models-lack-theory/)** *(2023 · 76 citations)* Large language models still handle theory of mind (ToM) poorly. They struggle to track what each character in a story believes, especially when a belief is false, and what one character assumes about another's beliefs. Earlier fixes trained models on templated benchmarks, and those models tend to exploit dataset… 70. **[NaturalProofs: Mathematical Theorem Proving in Natural Language](https://md.research.elamur.ai/free/en/papers/welleck-2021-naturalproofs-mathematical-theorem-proving-natural/)** *(2021 · 76 citations)* Most machine-learning work on theorem proving targets formal systems such as Lean, Mizar or Metamath. It largely ignores the informal mathematics people actually write, where LaTeX formulas sit inside ordinary prose. NaturalProofs fills this gap with a corpus of theorems, definitions and proofs drawn from three kinds… 71. **[Modular Pluralism: Pluralistic Alignment via Multi-LLM Collaboration](https://md.research.elamur.ai/free/en/papers/feng-2024-modular-pluralism-pluralistic-alignment-multi/)** *(2024 · 68 citations)* Standard alignment tends to push LLMs toward one averaged human preference. This leaves the diverse values of different cultures, demographics and communities poorly represented. Existing fixes usually require retraining or weight access, so they cannot be used with proprietary black-box models. Patching a newly… 72. **[Do Neural Language Representations Learn Physical Commonsense?](https://md.research.elamur.ai/free/en/papers/forbes-2019-do-neural-language-representations-learn/)** *(2019 · 64 citations)* This paper asks how much physical commonsense large text-trained language representations actually acquire. 73. **[Infini-gram: Scaling Unbounded n-gram Language Models to a Trillion Tokens](https://md.research.elamur.ai/free/en/papers/liu-2024-infini-gram-scaling-unbounded-n/)** *(2024 · 62 citations)* Classical n-gram language models are usually limited to small n (often 5) and modest data, because count tables grow explosively with n. 74. **[NaturalProver: Grounded Mathematical Proof Generation with Language Models](https://md.research.elamur.ai/free/en/papers/welleck-2022-naturalprover-grounded-mathematical-proof-generation/)** *(2022 · 62 citations)* The paper tackles theorem proving in informal mathematics, where proofs mix prose and LaTeX, a setting that modern generative models had barely explored. 75. **[ZebraLogic: On the Scaling Limits of LLMs for Logical Reasoning](https://md.research.elamur.ai/free/en/papers/lin-2025-zebralogic-scaling-limits-llms-logical/)** *(2025 · 60 citations)* It is still unclear how well large language models handle real deductive reasoning, and how that ability changes as models or compute grow. [← prev](https://md.research.elamur.ai/free/en/authors/yejin-choi/page/2/) [1](https://md.research.elamur.ai/free/en/authors/yejin-choi/) [2](https://md.research.elamur.ai/free/en/authors/yejin-choi/page/2/) **3** [4](https://md.research.elamur.ai/free/en/authors/yejin-choi/page/4/) [next →](https://md.research.elamur.ai/free/en/authors/yejin-choi/page/4/)