--- title: "Luke Zettlemoyer — papers, page 3 · University of Washington" description: "Luke Zettlemoyer, University of Washington: 87 papers on Elamur Research (2008–2025), h-index 91 — affiliations, topics, co-authors and…" canonical: "https://research.elamur.ai/free/en/authors/luke-zettlemoyer/page/3/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/authors/luke-zettlemoyer/page/3/) [home](https://md.research.elamur.ai/free/en/) / [authors](https://md.research.elamur.ai/free/en/authors/) / [Luke Zettlemoyer](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/) / page 3 # Luke Zettlemoyer · page 3 87 papers in our library · 2008–2025 · page 3 of 4 [The profile, affiliations and co-authors →](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/) ## Papers The most cited first (OpenAlex and our library). [← prev](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/page/2/) [1](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/) [2](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/page/2/) **3** [4](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/page/4/) [next →](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/page/4/) 51. **[Improving Passage Retrieval with Zero-Shot Question Generation](https://md.research.elamur.ai/free/en/papers/sachan-2022-improving-passage-retrieval-zero-shot/)** *(2022 · 199 citations)* Open-domain question answering relies on a first-stage retriever, and better ranking usually requires labelled question–passage pairs. 52. **[Neural Metaphor Detection in Context](https://md.research.elamur.ai/free/en/papers/gao-2018-neural-metaphor-detection-context/)** *(2018 · 198 citations)* The paper addresses metaphor detection: deciding whether a word in a sentence is used figuratively. Earlier systems often reduced context to subject–verb–object triples or relied on shallow unigram features. The authors show that the full sentence gives the model more useful evidence. They test two setups: tagging… 53. **[One Embedder, Any Task: Instruction-Finetuned Text Embeddings](https://md.research.elamur.ai/free/en/papers/su-2022-one-embedder-any-task-instruction/)** *(2022 · 194 citations)* Most text embedding models are built for one kind of job. A retriever such as DPR handles semantic similarity poorly, SimCSE has the reverse weakness, and both lose quality when moved to new domains like finance or medicine. The authors propose INSTRUCTOR, a single encoder that takes a natural-language description of… 54. **[Memorization Without Overfitting: Analyzing the Training Dynamics of Large Language Models](https://md.research.elamur.ai/free/en/papers/tirumala-2022-memorization-overfitting-analyzing-training-dynamics/)** *(2022 · 190 citations)* The paper looks at how large Transformer language models memorize their training data while they train, not only once training is finished, and how this changes as models grow. 55. **[Pre-training via Paraphrasing](https://md.research.elamur.ai/free/en/papers/lewis-2020-pre-training-paraphrasing/)** *(2020 · 187 citations)* Almost all large-scale NLP pre-training relies on masked language modeling, where parts of a text are hidden and then restored. 56. **[8-bit Optimizers via Block-wise Quantization](https://md.research.elamur.ai/free/en/papers/dettmers-2021-8-bit-optimizers-block-wise/)** *(2021 · 181 citations)* Stateful optimizers such as Adam and SGD with momentum store per-parameter gradient statistics. These statistics take up a large share of training memory (33–75% by the authors' estimate), which limits how big a model can fit on given hardware. Dettmers et al. show that these statistics can be kept in 8 bits instead… 57. **[Deep RNNs Encode Soft Hierarchical Syntax](https://md.research.elamur.ai/free/en/papers/blevins-2018-deep-rnns-encode-soft-hierarchical/)** *(2018 · 179 citations)* This short paper asks whether deep recurrent networks trained on ordinary NLP objectives learn syntactic structure that they were never directly taught. 58. **[Binding Language Models in Symbolic Languages](https://md.research.elamur.ai/free/en/papers/cheng-2022-binding-language-models-symbolic-languages/)** *(2022 · 170 citations)* End-to-end neural systems perform well on NLP tasks but are hard to interpret and fragile. Symbolic programs are transparent but limited to what their grammar can express. Earlier neural-symbolic hybrids needed custom languages, task-specific modules and large training sets. The authors introduce BINDER, a framework… 59. **[ROSCOE: A Suite of Metrics for Scoring Step-by-Step Reasoning](https://md.research.elamur.ai/free/en/papers/golovneva-2022-roscoe-suite-metrics-scoring-step/)** *(2022 · 155 citations)* Language models that write out their reasoning before answering often do better on downstream tasks. However, there has been no dependable automatic way to check whether the intermediate steps are actually sound, separately from whether the final answer is right. Golovneva et al. (Meta AI Research) address this gap… 60. **[RA-DIT: Retrieval-Augmented Dual Instruction Tuning](https://md.research.elamur.ai/free/en/papers/lin-2023-ra-dit-retrieval-augmented-dual/)** *(2023 · 149 citations)* Large language models store knowledge only in their weights, so they handle rare facts poorly and are costly to update. 61. **[Large-Scale QA-SRL Parsing](https://md.research.elamur.ai/free/en/papers/fitzgerald-2018-large-scale-qa-srl-parsing/)** *(2018 · 143 citations)* The paper releases QA-SRL Bank 2.0, which has 265,140 valid question-answer pairs for 64,018 sentences and 133,479 verbs across Wikipedia, Wikinews and science textbooks. 62. **[An Information Bottleneck Approach for Controlling Conciseness in Rationale Extraction](https://md.research.elamur.ai/free/en/papers/paranjape-2020-information-bottleneck-approach-controlling-conciseness/)** *(2020 · 140 citations)* Rationale-based models make predictions from a selected part of the input text, which makes them easier to interpret. 63. **[Self-Alignment with Instruction Backtranslation](https://md.research.elamur.ai/free/en/papers/li-2023-self-alignment-instruction-backtranslation/)** *(2023 · 139 citations)* Instruction-tuning LLMs usually depends on costly human-written demonstrations or on outputs distilled from stronger proprietary models, and high-quality annotation is hard to scale. 64. **[Grounded Adaptation for Zero-shot Executable Semantic Parsing](https://md.research.elamur.ai/free/en/papers/zhong-2020-grounded-adaptation-zero-shot-executable/)** *(2020 · 136 citations)* Executable semantic parsers, such as text-to-SQL systems, tend to fail when deployed on a database they never encountered during training, and labelled examples for each new schema are rarely available. 65. **[The case for 4-bit precision: k-bit Inference Scaling Laws](https://md.research.elamur.ai/free/en/papers/dettmers-2022-case-4-bit-precision-k/)** *(2022 · 123 citations)* Quantizing an LLM shrinks its memory use and, at small batch sizes, its latency. This raises a practical question: for a fixed budget of total model bits, is it better to run a larger model at low precision or a smaller one at higher precision? Dettmers and Zettlemoyer frame this as inference scaling laws that relate… 66. **[XLM-V: Overcoming the Vocabulary Bottleneck in Multilingual Masked Language Models](https://md.research.elamur.ai/free/en/papers/liang-2023-xlm-v-overcoming-vocabulary-bottleneck/)** *(2023 · 120 citations)* Multilingual encoders such as XLM-R have grown in depth and parameter count, yet their vocabularies have stayed near 250K tokens shared by more than 100 languages. 67. **[LSTM CCG Parsing](https://md.research.elamur.ai/free/en/papers/lewis-2016-lstm-ccg-parsing/)** *(2016 · 120 citations)* The paper asks whether a CCG parser needs an explicit model of bilexical dependencies. Its parser has none: a bidirectional LSTM supertagger assigns lexical categories, a deterministic grammar combines them, and A* search finds the best parse. Semi-supervised tri-training adds training signal. On CCGbank section 23… 68. **[Crowdsourcing Question-Answer Meaning Representations](https://md.research.elamur.ai/free/en/papers/michael-2018-crowdsourcing-question-answer-meaning-representations/)** *(2018 · 114 citations)* A QAMR represents a sentence's predicate-argument structure as a set of free-form wh-question / answer-span pairs about all its content words, not only verbs («How old is Pierre Vinken? – 61 years old», «What will he join the board as? – nonexecutive director»). 69. **[Dictionary-based Phrase-level Prompting of Large Language Models for Machine Translation](https://md.research.elamur.ai/free/en/papers/ghazvininejad-2023-dictionary-based-phrase-level-prompting/)** *(2023 · 95 citations)* Prompted LLMs translate surprisingly well without MT-specific training, but they often mistranslate rare words, which are common in low-resource language pairs and out-of-domain text. 70. **[A Probabilistic Model of Syntactic and Semantic Acquisition from Child-Directed Utterances and their Meanings](https://md.research.elamur.ai/free/en/papers/kwiatkowski-2012-probabilistic-model-syntactic-semantic-acquisition/)** *(2012 · 92 citations)* The authors model how a child might acquire grammar and word meanings simultaneously when hearing utterances paired with several possible interpretations of the situation, without knowing which one is correct. 71. **[Joint Coreference Resolution and Named-Entity Linking with Multi-Pass Sieves](https://md.research.elamur.ai/free/en/papers/hajishirzi-2013-joint-coreference-resolution-named-entity/)** *(2013 · 88 citations)* Coreference resolvers often fail because they lack background knowledge about entities. Named-entity linking (NEL) systems, in turn, frequently misjudge context when they examine each mention in isolation. The authors argue that the two tasks can help each other and propose NECO, a deterministic joint system with no… 72. **[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. 73. **[Controlled Crowdsourcing for High-Quality QA-SRL Annotation](https://md.research.elamur.ai/free/en/papers/roit-2020-controlled-crowdsourcing-high-quality-qa/)** *(2020 · 86 citations)* The paper shows that the crowdsourced QA-SRL data of 2018 has low coverage, with an estimated recall in the low 70s, and proposes a controlled protocol to fix it. 74. **[JuICe: A Large Scale Distantly Supervised Dataset for Open Domain Context-based Code Generation](https://md.research.elamur.ai/free/en/papers/agashe-2019-juice-large-scale-distantly-supervised/)** *(2019 · 75 citations)* The paper addresses code generation in interactive computing, where a programmer writes code cell by cell in Jupyter notebooks. 75. **[CREPE: Open-Domain Question Answering with False Presuppositions](https://md.research.elamur.ai/free/en/papers/yu-2023-crepe-open-domain-question-answering/)** *(2023 · 66 citations)* CREPE has 8,400 «explain like I'm five» questions from Reddit. Annotators, guided by the most upvoted answer, mark whether the question rests on a false presupposition and, if so, write the presupposition and a correction. About 25% of real information-seeking questions contain one. A sample of 50 shows the kinds: a… [← prev](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/page/2/) [1](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/) [2](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/page/2/) **3** [4](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/page/4/) [next →](https://md.research.elamur.ai/free/en/authors/luke-zettlemoyer/page/4/)