Pinzhen Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Pinzhen Chen has a total of 33 co-authors. Here are the top co-authors by number of shared publications:
- Alexandra Birch
School of Informatics University of Edinburgh
4 shared publications - Barry Haddow
School of Informatics University of Edinburgh
4 shared publications - Nikolay Bogoychev
School of Informatics University of Edinburgh
3 shared publications Yijun Yang
School of Informatics University of Edinburgh UK
3 shared publications- Zheng Zhao
University of Edinburgh
3 shared publications - Gerasimos Lampouras
Huawei Noah's Ark Lab London UK
2 shared publications
Latest Publications
- Open Access
MatheMagic: Generating Dynamic Mathematics Benchmarks Robust to Memorization
Conducting contamination-free evaluation of mathematical capabilities can be difficult for two reasons: models may memorize a test set once it is made public, and current mathematical benchmarks are prone to overfitting due to having limited diversity of symbols and rules, coupled with closed-ended answers. This paper proposes a method to leverage these shortcomings as useful features to a construct dynamic, counterfactual benchmark, which can be used to both reveal overfitting and measure true reasoning. We demonstrate this via MatheMagic, which generates math test instances with the interpretations of numbers and operators altered, yet has automatically verifiable answers. Test instances are randomly seeded and constructed at test time to evaluate a model's induction or deduction capability, offering stability, extensibility, comparability, and robustness to overfitting. Our experiments find that models solve deduction more easily than induction, but they revert to standard math. Further analysis reveals that math-adapted models fail to exhibit a general "skill" of reasoning, and fine-tuning on induction tasks generalizes poorly.
Open AccessMatheMagic: Generating Dynamic Mathematics Benchmarks Robust to Memorization
Conducting contamination-free evaluation of mathematical capabilities can be difficult for two reasons: models may memorize a test set once it is made public, and current mathematical benchmarks are prone to overfitting due to having limited diversity of symbols and rules, coupled with closed-ended answers. This paper proposes a method to leverage these shortcomings as useful features to a construct dynamic, counterfactual benchmark, which can be used to both reveal overfitting and measure true reasoning. We demonstrate this via MatheMagic, which generates math test instances with the interpretations of numbers and operators altered, yet has automatically verifiable answers. Test instances are randomly seeded and constructed at test time to evaluate a model's induction or deduction capability, offering stability, extensibility, comparability, and robustness to overfitting. Our experiments find that models solve deduction more easily than induction, but they revert to standard math. Further analysis reveals that math-adapted models fail to exhibit a general "skill" of reasoning, and fine-tuning on induction tasks generalizes poorly.
- Open Access
Generalizing From Short to Long: Effective Data Synthesis for Long-Context Instruction Tuning
Long-context modelling for large language models (LLMs) has been a key area of recent research because many real world use cases require reasoning over longer inputs such as documents. The focus of research into modelling long context has been on how to model position and there has been little investigation into other important aspects of language modelling such as instruction tuning. Long context training examples are challenging and expensive to create and use. In this paper, we investigate how to design instruction data for the post-training phase of a long context pre-trained model: how much and what type of context is needed for optimal and efficient post-training. Our controlled study reveals that models instruction-tuned on short contexts can effectively generalize to longer ones, while also identifying other critical factors such as instruction difficulty and context composition. Based on these findings, we propose context synthesis, a novel data synthesis framework that leverages off-the-shelf LLMs to generate extended background contexts for high-quality instruction-answer pairs. Experiment results on the document-level benchmark (LongBench) demonstrate that our proposed approach outperforms previous instruction synthesis approaches and comes close to the performance of human-annotated long-context instruction data. The project will be available at: https://github.com/NJUNLP/context-synthesis.
Open AccessGeneralizing From Short to Long: Effective Data Synthesis for Long-Context Instruction Tuning
Long-context modelling for large language models (LLMs) has been a key area of recent research because many real world use cases require reasoning over longer inputs such as documents. The focus of research into modelling long context has been on how to model position and there has been little investigation into other important aspects of language modelling such as instruction tuning. Long context training examples are challenging and expensive to create and use. In this paper, we investigate how to design instruction data for the post-training phase of a long context pre-trained model: how much and what type of context is needed for optimal and efficient post-training. Our controlled study reveals that models instruction-tuned on short contexts can effectively generalize to longer ones, while also identifying other critical factors such as instruction difficulty and context composition. Based on these findings, we propose context synthesis, a novel data synthesis framework that leverages off-the-shelf LLMs to generate extended background contexts for high-quality instruction-answer pairs. Experiment results on the document-level benchmark (LongBench) demonstrate that our proposed approach outperforms previous instruction synthesis approaches and comes close to the performance of human-annotated long-context instruction data. The project will be available at: https://github.com/NJUNLP/context-synthesis.
- Top Cited PapersOpen Access
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 5589-5610
Fine-Tuning Large Language Models with Sequential Instructions
Hanxu Hu, Simon Yu, Pinzhen Chen, Edoardo Ponti. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 5589-5610
Top Cited PapersOpen AccessFine-Tuning Large Language Models with Sequential Instructions
Hanxu Hu, Simon Yu, Pinzhen Chen, Edoardo Ponti. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.
- Open Access
Context and System Fusion in Post-ASR Emotion Recognition with Large Language Models
Large language models (LLMs) have started to play a vital role in modelling speech and text. To explore the best use of context and multiple systems' outputs for post-ASR speech emotion prediction, we study LLM prompting on a recent task named GenSEC. Our techniques include ASR transcript ranking, variable conversation context, and system output fusion. We show that the conversation context has diminishing returns and the metric used to select the transcript for prediction is crucial. Finally, our best submission surpasses the provided baseline by 20% in absolute accuracy.
Open AccessContext and System Fusion in Post-ASR Emotion Recognition with Large Language Models
Large language models (LLMs) have started to play a vital role in modelling speech and text. To explore the best use of context and multiple systems' outputs for post-ASR speech emotion prediction, we study LLM prompting on a recent task named GenSEC. Our techniques include ASR transcript ranking, variable conversation context, and system output fusion. We show that the conversation context has diminishing returns and the metric used to select the transcript for prediction is crucial. Finally, our best submission surpasses the provided baseline by 20% in absolute accuracy.
- Open Access
UniArk: Improving Generalisation and Consistency for Factual Knowledge Extraction through Debiasing
Several recent papers have investigated the potential of language models as knowledge bases as well as the existence of severe biases when extracting factual knowledge. In this work, we focus on the factual probing performance over unseen prompts from tuning, and using a probabilistic view we show the inherent misalignment between pre-training and downstream tuning objectives in language models for probing knowledge. We hypothesize that simultaneously debiasing these objectives can be the key to generalisation over unseen prompts. We propose an adapter-based framework, UniArk, for generalised and consistent factual knowledge extraction through simple methods without introducing extra parameters. Extensive experiments show that UniArk can significantly improve the model's out-of-domain generalisation as well as consistency under various prompts. Additionally, we construct ParaTrex, a large-scale and diverse dataset for measuring the inconsistency and out-of-domain generation of models. Further, ParaTrex offers a reference method for constructing paraphrased datasets using large language models.
Open AccessUniArk: Improving Generalisation and Consistency for Factual Knowledge Extraction through Debiasing
Several recent papers have investigated the potential of language models as knowledge bases as well as the existence of severe biases when extracting factual knowledge. In this work, we focus on the factual probing performance over unseen prompts from tuning, and using a probabilistic view we show the inherent misalignment between pre-training and downstream tuning objectives in language models for probing knowledge. We hypothesize that simultaneously debiasing these objectives can be the key to generalisation over unseen prompts. We propose an adapter-based framework, UniArk, for generalised and consistent factual knowledge extraction through simple methods without introducing extra parameters. Extensive experiments show that UniArk can significantly improve the model's out-of-domain generalisation as well as consistency under various prompts. Additionally, we construct ParaTrex, a large-scale and diverse dataset for measuring the inconsistency and out-of-domain generation of models. Further, ParaTrex offers a reference method for constructing paraphrased datasets using large language models.
- Open Access
EEE-QA: Exploring Effective and Efficient Question-Answer Representations
Current approaches to question answering rely on pre-trained language models (PLMs) like RoBERTa. This work challenges the existing question-answer encoding convention and explores finer representations. We begin with testing various pooling methods compared to using the begin-of-sentence token as a question representation for better quality. Next, we explore opportunities to simultaneously embed all answer candidates with the question. This enables cross-reference between answer choices and improves inference throughput via reduced memory usage. Despite their simplicity and effectiveness, these methods have yet to be widely studied in current frameworks. We experiment with different PLMs, and with and without the integration of knowledge graphs. Results prove that the memory efficacy of the proposed techniques with little sacrifice in performance. Practically, our work enhances 38-100% throughput with 26-65% speedups on consumer-grade GPUs by allowing for considerably larger batch sizes. Our work sends a message to the community with promising directions in both representation quality and efficiency for the question-answering task in natural language processing.
Open AccessEEE-QA: Exploring Effective and Efficient Question-Answer Representations
Current approaches to question answering rely on pre-trained language models (PLMs) like RoBERTa. This work challenges the existing question-answer encoding convention and explores finer representations. We begin with testing various pooling methods compared to using the begin-of-sentence token as a question representation for better quality. Next, we explore opportunities to simultaneously embed all answer candidates with the question. This enables cross-reference between answer choices and improves inference throughput via reduced memory usage. Despite their simplicity and effectiveness, these methods have yet to be widely studied in current frameworks. We experiment with different PLMs, and with and without the integration of knowledge graphs. Results prove that the memory efficacy of the proposed techniques with little sacrifice in performance. Practically, our work enhances 38-100% throughput with 26-65% speedups on consumer-grade GPUs by allowing for considerably larger batch sizes. Our work sends a message to the community with promising directions in both representation quality and efficiency for the question-answering task in natural language processing.
- Open Access
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 388-409
Fine-Tuning Large Language Models to Translate: Will a Touch of Noisy Data in Misaligned Languages Suffice?
Dawei Zhu, Pinzhen Chen, Miaoran Zhang, Barry Haddow, Xiaoyu Shen, Dietrich Klakow. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 388-409
Open AccessFine-Tuning Large Language Models to Translate: Will a Touch of Noisy Data in Misaligned Languages Suffice?
Dawei Zhu, Pinzhen Chen, Miaoran Zhang, Barry Haddow, Xiaoyu Shen, Dietrich Klakow. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.
- Open Access
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 9706-9726
Is It Good Data for Multilingual Instruction Tuning or Just Bad Multilingual Evaluation for Large Language Models?
Pinzhen Chen, Simon Yu, Zhicheng Guo, Barry Haddow. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 9706-9726
Open AccessIs It Good Data for Multilingual Instruction Tuning or Just Bad Multilingual Evaluation for Large Language Models?
Pinzhen Chen, Simon Yu, Zhicheng Guo, Barry Haddow. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.
- Open Access
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 686-691
Cher at KSAA-CAD 2024: Compressing Words and Definitions into the Same Space for Arabic Reverse Dictionary
Pinzhen Chen, Zheng Zhao, Shun Shao. Proceedings of The Second Arabic Natural Language Processing Conference. 2024.
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 686-691
Open AccessCher at KSAA-CAD 2024: Compressing Words and Definitions into the Same Space for Arabic Reverse Dictionary
Pinzhen Chen, Zheng Zhao, Shun Shao. Proceedings of The Second Arabic Natural Language Processing Conference. 2024.
- Open Access
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 7018-7035
UniArk: Improving Generalisation and Consistency for Factual Knowledge Extraction through Debiasing
Yijun Yang, Jie He, Pinzhen Chen, Victor Gutierrez Basulto, Jeff Pan. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 7018-7035
Open AccessUniArk: Improving Generalisation and Consistency for Factual Knowledge Extraction through Debiasing
Yijun Yang, Jie He, Pinzhen Chen, Victor Gutierrez Basulto, Jeff Pan. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.