Tag
The paper proposes Parallel Pretraining (PPT), which concatenates documents with their translations in the same training sequence, and finds that a two-stage PPT-then-standard-pretraining approach consistently improves general language understanding and cross-lingual performance over standard pretraining at matched token budgets.
一项覆盖14种语言的大规模研究，探讨跨语言仇恨言论检测的迁移效果，发现语言相似性是最强预测因素，而文化线索与交际规范可部分解释剩余差异。
The paper introduces Multilingual GSM-Symbolic, a symbolic-template dataset covering 15 languages and 30k item-matched QA pairs, to quantify what determines cross-lingual capability transfer in LLMs. It finds model size, language resource level, reasoning, and typological distance are key drivers, and shows a framework explaining 92% of between-language variation while predicting unseen-language performance within 6.0pp.
The paper presents the first experiments on isolated sign language recognition for Icelandic Sign Language using a low-resource dataset, showing that cross-lingual transfer from other sign languages significantly improves performance.
This paper explores improving cross-lingual transfer for sequential sentence classification in research papers by leveraging structural similarity, proposing methods that enhance performance in multilingual settings based on experiments with encoder-based and generative models.
The paper introduces PolERo, a dataset of 3,574 annotated question-answer pairs from Romanian presidential transcripts, to study political evasion. It evaluates various NLP classification models and examines cross-lingual transfer, finding that fine-tuned encoders are competitive and ambivalent evasion categories are challenging.
The study evaluates cross-lingual transfer in legal comprehension for Tulu using transliteration and RAG with Kannada legal papers, revealing script-dependent improvements and challenges like fact substitution and confabulation.
This paper investigates cross-lingual alignment in small Japanese-English bilingual language models by probing cultural and pragmatic reasoning, introducing the J-PragEval-v0 benchmark and proposing Pragmatic Representation Steering for inference-time interventions.
This paper introduces DonorRank, a learning-to-rank framework for selecting effective donor languages in low-resource cross-lingual speech recognition, evaluated on Indic and African language corpora. It demonstrates improved donor selection over genetic-similarity and high-resource heuristics, and provides insights into transfer patterns for multilingual ASR.
This paper studies cross-lingual transfer for machine translation among five Turkic languages using pairwise transfer matrices with mT5, finding that transfer is strongest between closely related pairs and that Latinization helps in script-mismatched settings.
This paper presents a systematic study of cross-lingual transfer for biomedical machine translation into low-resource Arabic-script languages, using Arabic and Persian as pivots. The authors evaluate LoRA adapter merging as a zero-data transfer strategy, showing it works surprisingly well for closely related languages like Dari.
This paper addresses the uncontrolled latent variable of transcription style in ASR models by introducing a method using coverage-aware decoder task tokens to enable controllable verbatim transcription with accurate word-level timing, achieving high disfluency detection F1 via zero-shot cross-lingual transfer.
This paper introduces KyrgyzLLM-Bench, a benchmark suite for evaluating large language models in the Kyrgyz language, comprising both natively authored and translated datasets, and provides a systematic evaluation of 26 models.
Presents Tatoxa, a state-of-the-art system for text detoxification in the Tatar language, outperforming existing LLMs. Introduces a new dataset and shows that cross-lingual transfer performs worse than native data.
This paper proposes a benchmark suite grounded in Pāṇinian grammar to unify Indic language processing across languages, aiming to improve accuracy, data efficiency, and transferability.
This paper empirically compares several LoRA variants for multilingual instruction tuning and finds no significant advantage of complex variants over basic LoRA in balancing cross-lingual transfer and knowledge retention.
This paper introduces PyLang, a programming language absent from all pretraining corpora, and shows that LLMs fine-tuned on it can learn syntax but fail to transfer algorithmic reasoning, resulting in an 'implementation fidelity gap' where models understand algorithms but cannot express them in an unfamiliar language.