Papers with model generalization
Scaling Parameter-Constrained Language Models with Quality Data (2024.emnlp-industry)
Copied to clipboard
Ernie Chang, Matteo Paltenghi, Yang Li, Pin-Jie Lin, Changsheng Zhao, Patrick Huber, Zechun Liu, Rastislav Rabatin, Yangyang Shi, Vikas Chandra
| Challenge: | Scaling laws in language modeling quantify training loss as a function of dataset size and model parameters, but neglect the critical role of data quality in model generalization. |
| Approach: | They propose to use effective training tokens as a combination of text diversity and syntheticity as measured by a teacher model to calculate scaling laws. |
| Outcome: | The proposed term effective training tokens is a combination of two readily-computed indicators of text diversity and syntheticity as measured by a teacher model. |
Data-driven Model Generalizability in Crosslinguistic Low-resource Morphological Segmentation (2022.tacl-1)
Copied to clipboard
| Challenge: | morphological segmentation is a common method of evaluation for multilingual tasks . authors often examine models with one data set that is representative of all possible data . |
| Approach: | They compare three broad classes of models with different parameterizations using morphological segmentation as the test case. |
| Outcome: | The results show that the extent of model generalization depends on the characteristics of the data set, and does not necessarily rely heavily on the data sets size. |
Hierarchical Curriculum Learning for AMR Parsing (2022.acl-short)
Copied to clipboard
| Challenge: | Abstract Meaning Representation (AMR) parsing aims to translate sentences to semantic representation with a hierarchical structure, but there is a gap between their flat training objective and the hierarchic structure, which limits the model generalization. |
| Approach: | They propose a Hierarchical Curriculum Learning framework with Structure-level (SC) and Instance-level curricula (IC) that aims to translate sentences to semantic representation with a hierarchical structure. |
| Outcome: | Experiments on AMR2.0, AMR3.0, structure-complex and out-of-distribution situations confirm the effectiveness of the proposed framework. |
Uncertainty Estimation and Reduction of Pre-trained Models for Text Regression (2022.tacl-1)
Copied to clipboard
| Challenge: | State-of-the-art classification and regression models are often not well calibrated and can be inaccurate. |
| Approach: | They quantify calibration of pre- trained language models for text regression . they apply uncertainty estimates to augment training data in low-resource domains . |
| Outcome: | The proposed model calibrations improve performance and generalizability in low-resource settings. |
Token-Level Self-Evolution Training for Sequence-to-Sequence Learning (2023.acl-short)
Copied to clipboard
| Challenge: | Adaptive training approaches do not consider the variation of learning difficulty in different training steps, making the learning deterministic and sub-optimal. |
| Approach: | They propose a dynamic token-level self-evolution training method that reweighs the training losses of different target tokens based on priors. |
| Outcome: | Empirically, the proposed method yields significant improvements on three translation tasks. |
Posterior Differential Regularization with f-divergence for Improving Model Robustness (2021.naacl-main)
Copied to clipboard
| Challenge: | Recent studies show that pre-trained models suffer catastrophic degradation in out-of-domain generalization to datasets with domain shift or adversarial scenarios. |
| Approach: | They propose to regularize the posterior difference between clean and noisy inputs by using a Jacobian regularization framework and a virtual adversarial training framework. |
| Outcome: | The proposed framework can improve model robustness in fully supervised and semi-supervised settings. |
Improving Model Generalization: A Chinese Named Entity Recognition Case Study (2021.acl-short)
Copied to clipboard
| Challenge: | Named Entity Recognition (NER) is a fundamental building block for various downstream natural language processing tasks due to the ambiguous word boundaries and complex composition. |
| Approach: | They propose to resample entities within the same category to encourage a model to leverage both name and context knowledge in the training process. |
| Outcome: | The proposed method significantly improves a model’s ability to detect unseen entities, especially for company, organization and position categories. |
Improving Prompt Generalization for Cross-prompt Essay Trait Scoring from the Scoring-invariance Perspective (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing research on cross-prompt trait essay scoring focuses on improving model generalization by obtaining prompt-invariant representations. |
| Approach: | They propose a scoring-invariant learning objective that encourages the model to focus on intrinsic information within the essay that reflects its quality during training, thereby learning generic scoring features. |
| Outcome: | The proposed scoring-invariant learning objective encourages the model to focus on intrinsic information within the essay that reflects its quality during training, thereby learning generic scoring features. |
Less Is Better: Recovering Intended-Feature Subspace to Robustify NLU Models (2022.coling-1)
Copied to clipboard
| Challenge: | Existing approaches to debiase datasets rely on knowledge of bias attributes . current approaches focus on how to leverage kinds of supervision effectively . |
| Approach: | They propose to extend the supervision on bias by extending it into feature space. |
| Outcome: | Empirical results show that a low-dimensional subspace with intended features can represent biased datasets. |
Span Pointer Networks for Non-Autoregressive Task-Oriented Semantic Parsing (2021.findings-emnlp)
Copied to clipboard
Akshat Shrivastava, Pierce Chuang, Arun Babu, Shrey Desai, Abhinav Arora, Alexander Zotov, Ahmed Aly
| Challenge: | a novel approach to map utterances to semantic frames is based on non-autoregressive parsers that shift the decoding task from text generation to span prediction. |
| Approach: | They propose a non-autoregressive, task-oriented parser which shifts the decoding task from text generation to span prediction and produces endpoints as opposed to text. |
| Outcome: | The proposed model bridges the quality gap between non-autoregressive and autoregressive parsers, achieving 87 EM on TOPv2 and shows a 70% reduction in latency and 83% reduction in memory at beam size 5 compared to prior non-regressives. |
Quantifying Train-Evaluation Overlap with Nearest Neighbors (2023.findings-acl)
Copied to clipboard
| Challenge: | Using a novel metric, we quantify train-evaluation overlap in benchmark datasets to evaluate model generalization. |
| Approach: | They quantify train-evaluation overlap as a measure of an individual dataset’s adequacy to evaluate model generalization over a wide range of datasets. |
| Outcome: | The proposed metric quantifies train-evaluation overlap, providing insights for constructing datasets to study generalization. |
Learning Mutually Informed Representations for Characters and Subwords (2024.findings-naacl)
Copied to clipboard
| Challenge: | Pretrained language models rely on subword tokenization to process text as a sequence of subwords. |
| Approach: | They propose a character-subword language model that integrates character and subword modalities into one model. |
| Outcome: | The proposed model outperforms its backbone language models on English sequence labeling and classification tasks. |
Tailor: Generating and Perturbing Text with Semantic Controls (2022.acl-long)
Copied to clipboard
| Challenge: | Existing studies train task-specific generators, relying on training a model for every perturbation. |
| Approach: | They propose a semantically-controlled text generation system that modifies sentences to match target attributes. |
| Outcome: | The proposed system produces textual outputs conditioned on control codes derived from semantic representations. |
Learning Physical Common Sense as Knowledge Graph Completion via BERT Data Augmentation and Constrained Tucker Factorization (2020.emnlp-main)
Copied to clipboard
| Challenge: | Physical commonsense learning is an essential part of human-robot interaction . existing methods of learning physical commons sense suffer from generalization . |
| Approach: | They propose to use physical commonsense learning as a knowledge graph completion problem to better use latent relationships among training samples. |
| Outcome: | The proposed method outperforms existing methods in the human-robot interaction problem. |
Multilingual Mix: Example Interpolation Improves Multilingual Neural Machine Translation (2022.acl-long)
Copied to clipboard
| Challenge: | Existing approaches to train multilingual models to learn the inductive bias of a shared vocabulary and set of parameters across languages. |
| Approach: | They propose to use a multilingual crossover encoder-decoder to fuse language pairs at an instance level to encourage sharing of input and output spaces. |
| Outcome: | The proposed approach improves quality on English-to-Many, Many-to English and zero-shot translation tasks from +0.5 BLEU up to +5.5 BLUE points. |
Improving Sharpness-Aware Minimization with Fisher Mask for Better Generalization on Language Models (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for fine-tuning pretrained language models suffer from poor generalization . however, they add a perturbation to each model parameter equally, which is sub-optimal . |
| Approach: | They propose a sharpness-aware minimization optimization procedure that introduces a Fisher mask to improve the efficiency of SAM. |
| Outcome: | The proposed method outperforms the vanilla sharpness-aware minimization method on GLUE and SuperGLUE benchmarks. |
Learning to Learn to be Right for the Right Reasons (2021.naacl-main)
Copied to clipboard
| Challenge: | Recent work shows that models trained on held-out data perform poorly on hard instances . previous methods have resorted to manual methods of encouraging models not to overfit to superficial cues . |
| Approach: | They propose to explicitly learn a model that does well on both easy and hard tests . they use Choice of Plausible Alternatives and Commonsense Explanation to evaluate the model . |
| Outcome: | The proposed model performs well on easy and hard tests with superficial cues but performs poorly on hard ones without superficial cuings. |
Dual Adversarial Neural Transfer for Low-Resource Named Entity Recognition (P19-1)
Copied to clipboard
| Challenge: | Named entity recognition (NER) is an important step in most natural language processing (NLP) applications. |
| Approach: | They propose a dual-adversarial neural transfer method for addressing low-resource Named Entity Recognition (NER) they propose 'Generalized Resource-Adversarial Discriminator' and 'accidental training' |
| Outcome: | The proposed method improves on low-resource Named Entity Recognition (NER) with two variants, i.e., DATNet-F and DATNET-P, and adversarial training is adopted to boost model generalization. |
ARCH: Efficient Adversarial Regularized Training with Caching (2021.findings-emnlp)
Copied to clipboard
Simiao Zuo, Chen Liang, Haoming Jiang, Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen, Tuo Zhao
| Challenge: | Existing approaches to regularize models require generating a perturbation for each sample in each epoch. |
| Approach: | They propose an adversarial regularization method where perturbations are generated and cached once every several epochs. |
| Outcome: | The proposed method significantly eases the computational burden (saves up to 70% of computational time) it produces a notably better (in most of the tasks) or comparable model generalization. |
Domain-Oriented Prefix-Tuning: Towards Efficient and Generalizable Fine-tuning for Zero-Shot Dialogue Summarization (2022.naacl-main)
Copied to clipboard
| Challenge: | Existing methods for domain adaptation of abstractive dialogue summarization lack generalization ability on new domains. |
| Approach: | They propose a domain-oriented prefix-tuning model that uses a prefix module to alleviate domain entanglement and discrete prompts to guide the model to focus on key contents of dialogues. |
| Outcome: | The proposed model can be generalized to two multi-domain dialogue summarization datasets. |
Sharpness-Aware Minimization with Dynamic Reweighting (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Deep neural networks are often overparameterized and can overfit training data. |
| Approach: | They propose an adversarial weight minimization algorithm that conducts adversarials and finds a common adversaria per-batch. |
| Outcome: | The proposed algorithm finds a common adversarial weight perturbation per-batch. |
Adaptive Meta-learner via Gradient Similarity for Few-shot Text Classification (2022.coling-1)
Copied to clipboard
| Challenge: | Existing methods for few-shot text classification suffer from overfitting due to the lack of matching between the few amount of samples and complicated models. |
| Approach: | They propose a method to improve model generalization ability to a new task by leveraging a meta-learner via gradient similarity method. |
| Outcome: | The proposed method improves few-shot text classification performance on several benchmarks. |
Can Data Diversity Enhance Learning Generalization? (2022.coling-1)
Copied to clipboard
| Challenge: | a diversity advanced actor-critical reinforcement learning framework is used to improve NLP generalization and accuracy. |
| Approach: | They introduce Diversity Advanced Actor-Critic reinforcement learning framework to improve NLP generalization and accuracy. |
| Outcome: | The proposed framework outperforms domain adaptation and generalization baselines without using any target domain knowledge. |
Kill two birds with one stone: generalized and robust AI-generated text detection via dynamic perturbations (2025.naacl-long)
Copied to clipboard
| Challenge: | Existing methods focus on model generalization or focus on robustness. |
| Approach: | They propose a model-based AIGT detection method that can be generalized and robust under two adversarial attacks. |
| Outcome: | The proposed method outperforms state-of-the-art methods for generalization and robustness under two text adversarial attacks. |
STGN: an Implicit Regularization Method for Learning with Noisy Labels in Natural Language Processing (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing work to mitigate the effect of noisy labels is limited to specific tasks or training procedures, making it hard to be widely used. |
| Approach: | They propose a stochastic tailor-made gradient noise to mitigate the effect of noisy labels by introducing benign noise into stochistic gradient descent. |
| Outcome: | The proposed method can be used to discriminate correct samples from incorrect ones and boost existing training methods. |
SoRFT: Issue Resolving with Subtask-oriented Reinforced Fine-Tuning (2025.acl-long)
Copied to clipboard
| Challenge: | Existing issue-resolving frameworks rely on commercial models, leading to high costs and privacy concerns. |
| Approach: | They propose a training approach to enhance issue resolving capability of LLMs by decomposing issue reasolving into subtasks. |
| Outcome: | The proposed approach improves issue-resolving performance and generalizes model . it is cost-effective and provides a cost-efficient alternative to commercial models . |
In2Core: Leveraging Influence Functions for Coreset Selection in Instruction Finetuning of Large Language Models (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) exhibit surprising abilities across a variety of language tasks. |
| Approach: | They propose an algorithm which selects a coreset by analyzing correlation between training and evaluation samples with a trained model. |
| Outcome: | The proposed algorithm can achieve similar performance with just 50% of the training data while preserving the accuracy of the existing model. |
CACL: Community-Aware Heterogeneous Graph Contrastive Learning for Social Media Bot Detection (2024.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for social media bot detection neglect community structure and poor model generalization due to the relatively small scale of the dataset. |
| Approach: | They propose a framework that constructs social networks as heterogeneous graphs and uses community-aware modules to mine hard positive and hard negative samples for supervised graph contrastive learning. |
| Outcome: | The proposed framework outperforms baselines on three social media bot benchmarks. |
Enhancing Explainable Rating Prediction through Annotated Macro Concepts (2024.acl-long)
Copied to clipboard
| Challenge: | Existing models learn user and item embeddings and generate reasons based on these embedds. |
| Approach: | They propose a concept-based explanation framework that leverages macro concepts to bridge the gap between the user/item embeddings and the recommendation reasons. |
| Outcome: | Extensive experiments on three datasets prove the proposed model is superior to existing models. |
Adaptive Label Smoothing with Self-Knowledge in Natural Language Generation (2022.emnlp-main)
Copied to clipboard
| Challenge: | Overconfidence in model generalization and calibration has been shown to impair model generalisation and calibration. |
| Approach: | They propose a regularization scheme that takes model probability into account and takes it into account . they use a prior label distribution to smooth target labels . |
| Outcome: | The proposed model improves model generalization and calibration by taking model probability into account. |
Hard Gate Knowledge Distillation - Leverage Calibration for Robust and Reliable Language Model (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing knowledge distillation schemes focus on a teacher as a source of knowledge and a gauge to detect miscalibration of a student. |
| Approach: | They propose a method that uses a teacher model as a source of knowledge and a model as an error detector to detect miscalibration of a student. |
| Outcome: | The proposed scheme improves model generalization and significantly lowers calibration error. |
Zero-shot Sharpness-Aware Quantization for Pre-trained Language Models (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing zero-shot quantization methods are based on overfitting problem in adversarial learning process, leading to sub-optimal performance. |
| Approach: | They propose a zero-shot sharpness-aware quantization framework for the quantization of various PLMs by optimizing a minimax problem. |
| Outcome: | The proposed framework can achieve significant performance gains on discriminative and generative PLMs. |
Improving generalization in large langue model by learning prefix subspaces (2023.findings-emnlp)
Copied to clipboard
| Challenge: | emergence of large language models has significantly transformed the applications of deep learning methods in natural language processing. |
| Approach: | They propose to improve LLMs' generalization by optimizing entire models in parameter space by learning entire simplexes of continous prefixes. |
| Outcome: | The proposed method improves generalization of large language models in the scarce data regime. |
Directions for NLP Practices Applied to Online Hate Speech Detection (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing approaches to address hate speech in online spaces have relied on conventions and practices from NLP. |
| Approach: | They argue that many conventions in NLP are poorly suited for the problem and encourage researchers to develop methods that are more appropriate for the task. |
| Outcome: | The proposed methods are poorly suited for the problem and should be adapted to address the propagation of online harms. |
Few-Shot Named Entity Recognition: An Empirical Baseline Study (2021.emnlp-main)
Copied to clipboard
Jiaxin Huang, Chunyuan Li, Krishan Subudhi, Damien Jose, Shobana Balakrishnan, Weizhu Chen, Baolin Peng, Jianfeng Gao, Jiawei Han
| Challenge: | Existing methods to build named entity recognition systems with limited labeled data are lacking. |
| Approach: | They propose three orthogonal schemes to build named entity recognition systems when labeled data is limited. |
| Outcome: | The proposed NER systems outperform existing methods on few-shot and training-free settings. |
Plug-and-Play Data Module for Code RL: Adaptive Ambiguity Replay (2026.findings-acl)
Copied to clipboard
Jianqing Zhang, Wei Xia, Zhezheng Hao, Hong Wang, Hande Dong, Qiang Lin, Yang Liu, Jian Cao, Qiang Yang
| Challenge: | Existing approaches to reinforcement learning (RL) rely on static, in-epoch metrics that overlook training dynamics, often introducing low-utility or outdated data. |
| Approach: | They propose a plug-and-play module that prioritizes cross-epoch ambiguous samples to neutralize the noise from stale experiences. |
| Outcome: | Extensive experiments on nine LLMs show that Adaptive Ambiguity Replay outperforms state-of-the-art baselines on real-world code editing tasks. |
Leveraging Contrastive Learning and Knowledge Distillation for Incomplete Modality Rumor Detection (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing rumor detection models neglect the semantic coherence between text and image components in multimodal posts . Existing models neglect incomplete modalities in single modal posts, such as missing text or images . |
| Approach: | They propose a framework for incomplete modality rumor detection that captures semantic consistency between text and image pairs while enhancing model generalization to incomplete modalities within individual posts. |
| Outcome: | The proposed framework outperforms state-of-the-art methods on two English and two Chinese benchmark datasets for rumor detection in social media. |
Challenges to Evaluating the Generalization of Coreference Resolution Models: A Measurement Modeling Perspective (2024.findings-acl)
Copied to clipboard
| Challenge: | a recent study shows that evaluations of CR models on multiple datasets conflate different factors concerning what is being measured. |
| Approach: | They propose to view evaluations through the lens of measurement modeling . they show that evaluations risk conflating different factors concerning what is being measured . |
| Outcome: | The evaluations on seven datasets show that models that reflect coreference generalization are often correlated with differences in how coreference is defined and operationalized. |
MoNMT: Modularly Leveraging Monolingual and Bilingual Knowledge for Neural Machine Translation (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing models for multi-domain translation tasks only use monolingual data, whereas bilingual data is indispensable for improving the models. |
| Approach: | They propose a modular strategy that facilitates the cooperation of monolingual and bilingual knowledge in translation tasks by avoiding catastrophic forgetting. |
| Outcome: | The proposed model exhibits superior generalization and robustness over the conventional approach. |
Cross-Cultural Transfer of Emoji Semantics and Sentiment in Financial Social Media (2026.findings-acl)
Copied to clipboard
| Challenge: | Emojis are widely used in online financial communication, but it is unclear whether they provide transferable sentiment signals across languages, platforms, and asset communities. |
| Approach: | They examine whether emoji usage, semantics, and sentiment polarity remain stable across financial communities and asset communities. |
| Outcome: | The results show that emojis provide compact, language-independent sentiment cues that improve model generalization across markets and platforms. |
Adaptation Odyssey in LLMs: Why Does Additional Pretraining Sometimes Fail to Improve? (2024.emnlp-main)
Copied to clipboard
| Challenge: | In the last decade, the generalization and adaptation abilities of deep learning models were evaluated on fixed training and test distributions. |
| Approach: | They propose to train large language models on unlabeled text corpora and train them online. |
| Outcome: | The proposed model training on a text domain could degrade its perplexity on the test portion of the same domain. |
Understanding the Influence of Synthetic Data for Text Embedders (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in general purpose text embedders have been driven by training on synthetic training data. |
| Approach: | They propose to use GPT-4 to produce high quality synthetic data that expands existing training datasets for embeddings to new tasks. |
| Outcome: | The proposed dataset is high quality and leads to consistent improvements in performance. |
KaFT: Knowledge-aware Fine-tuning for Boosting LLMs’ Domain-specific Question-Answering Performance (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent literature reveals that supervised fine-tuning (SFT) is suboptimal for domain-specific question-answering tasks. |
| Approach: | They propose a query diversification strategy for robust conflict detection and a knowledge-aware fine-tuning approach to effectively boost LLMs’ performance. |
| Outcome: | The proposed approach improves the model generalization and alleviates the hallucination. |
Relaxing the Constraints: A Dual-Importance Projection Mechanism for Lifelong Model Editing (2026.findings-acl)
Copied to clipboard
Zhenghai Chen, Senbin Xu, Jiaxi Tan, Xinhua Wu, Yan Zhang, Xiawu Zheng, Shengchuan Zhang, Ke Li, Sicheng Zhao, Liujuan Cao, Rongrong Ji
| Challenge: | Existing knowledge editing methods rely on strict orthogonal projection to preserve previously edited knowledge, but this constraint limits gradient expressiveness, resulting in degradation of model generalization and overall performance as the number of edits increases. |
| Approach: | They propose a method that leverages Singular Value Decomposition to identify critical gradient subspaces and introduces a dual mechanism comprising "accumulated importance" and "projection importance" |
| Outcome: | Extensive experiments on five mainstream LLMs show that the proposed method achieves an average comprehensive performance improvement of 10.36% and effectively maintains the model’s general capabilities on downstream tasks. |