Hannaneh Hajishirzi · page 2
Papers
- SciREX: A Challenge Dataset for Document-Level Information Extraction (2020 · 220 citations)
Most information extraction (IE) benchmarks work at the level of single sentences or short paragraphs, yet many facts in scientific papers can only be recovered by reading the whole article. - Solving Geometry Problems: Combining Text and Diagram Interpretation (2015 · 220 citations)
The paper presents GeoS, described as the first automated system that solves unmodified SAT plane-geometry multiple-choice questions. - Real-Time Open-Domain Question Answering with Dense-Sparse Phrase Index (2019 · 218 citations)
Standard open-domain QA systems such as DrQA first retrieve a handful of documents and then run a neural reader over them for each new question. - XOR QA: Cross-lingual Open-Retrieval Question Answering (2021 · 210 citations)
Most multilingual QA benchmarks assume the answer is written in the same language as the question. In practice, many languages have few relevant articles (information scarcity), and native speakers often ask about foreign entities that local Wikipedias barely cover (information asymmetry). The authors define XOR QA… - Camels in a Changing Climate: Enhancing LM Adaptation with Tulu 2 (2023 · 207 citations)
After the first TÜLU release, open instruction tuning moved quickly. Stronger base models, better distilled datasets and simpler adaptation methods all appeared. This work from AI2 and the University of Washington tests these advances one by one and keeps the useful ones in TÜLU 2, a suite of LLAMA-2 models at 7B… - Is Reinforcement Learning (Not) for Natural Language Processing: Benchmarks, Baselines, and Building Blocks for Natural Language Policy Optimization (2022 · 174 citations)
The paper asks whether reinforcement learning is a practical way to align pre-trained language models with human preferences. - An Information Bottleneck Approach for Controlling Conciseness in Rationale Extraction (2020 · 140 citations)
Rationale-based models make predictions from a selected part of the input text, which makes them easier to interpret. - Task-aware Retrieval with Instructions (2023 · 132 citations)
A search query alone often hides what the user actually wants. The same query might call for code, an answer, or a duplicate question. Usually this is handled by training a separate retriever for each intent. Asai et al. instead propose "retrieval with instructions," where a natural-language description of the search… - ATTEMPT: Parameter-Efficient Multi-task Tuning via Attentional Mixtures of Soft Prompts (2022 · 122 citations)
Parameter-efficient tuning methods such as prompt tuning change only a tiny share of a language model's weights. However, they often trail full fine-tuning and learn each task in isolation, missing what large, well-resourced datasets could offer. - Logic-Guided Data Augmentation and Regularization for Consistent Question Answering (2020 · 113 citations)
Comparison questions ask how two entities or events relate, for example which is larger or what effect a cause has. - Data Engineering for Scaling Language Models to 128K Context (2024 · 109 citations)
Open-source LLMs that claim support for 100K+ tokens often fail the Needle-in-a-Haystack test, which checks whether a model can recall a fact hidden anywhere in a very long document. - Fine-grained Hallucination Detection and Editing for Language Models (2024 · 100 citations)
Language models frequently state false facts. Most existing checkers give only a yes/no verdict for an entire response, so every error is treated the same way. The authors argue that factual mistakes come in distinct kinds, each requiring a different amount of verification effort, and they organize these kinds into a… - Probing Across Time: What Does RoBERTa Know and When? (2021 · 96 citations)
The paper asks when, during pretraining, a masked language model acquires different kinds of knowledge, rather than only what a finished model knows. - CONQRR: Conversational Query Rewriting for Retrieval with Reinforcement Learning (2022 · 95 citations)
Open-domain conversational question answering (CQA) needs passage retrieval for questions that only make sense in light of earlier turns. - SciFact-Open: Towards open-domain scientific claim verification (2022 · 94 citations)
Current scientific claim verification systems score very well on small benchmarks, but nobody had checked how they behave against literature collections of realistic size. - Joint Coreference Resolution and Named-Entity Linking with Multi-Pass Sieves (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… - Do Membership Inference Attacks Work on Large Language Models? (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. - BUFFET: Benchmarking Large Language Models for Few-shot Cross-lingual Transfer (2024 · 86 citations)
Research on few-shot learning with large language models has focused mostly on English. It has remained unclear how in-context learning (ICL) compares with fine-tuning when only a few target-language examples are available. The authors present BUFFET, a benchmark whose name refers to evaluating few-shot transfer… - Rainier: Reinforced Knowledge Introspector for Commonsense Question Answering (2022 · 82 citations)
Commonsense question answering often improves when a model receives extra background knowledge alongside the question. - CREPE: Open-Domain Question Answering with False Presuppositions (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… - Infini-gram: Scaling Unbounded n-gram Language Models to a Trillion Tokens (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. - Evidentiality-guided Generation for Knowledge-Intensive NLP Tasks (2022 · 60 citations)
Retrieval-augmented generators such as FiD often produce answers from retrieved passages that look relevant but do not actually support the output. - Generalizing Verifiable Instruction Following (2025 · 59 citations)
Language models are commonly tested on whether they obey checkable output rules, such as word counts or required formats. - Unpacking DPO and PPO: Disentangling Best Practices for Learning from Preference Feedback (2024 · 50 citations)
Preference-based training is now a standard final stage for language models. Published setups, however, combine different datasets, algorithms and evaluation suites, so it is unclear which ingredient actually drives the gains. The authors divide the pipeline into four components: preference data, the learning… - Don't throw away your value model! Generating more preferable text with Value-Guided Monte-Carlo Tree Search decoding (2023 · 46 citations)
Language models aligned with PPO are usually deployed using only the policy network. The value network that PPO trains alongside it is typically thrown away. The authors argue that this byproduct is well suited to scoring candidates during guided decoding. It is trained to estimate the expected return of unfinished…
This page in Markdown → https://md.research.elamur.ai/free/en/authors/hannaneh-hajishirzi/page/2/