English
Related papers

Related papers: Cross-Tokenizer Likelihood Scoring Algorithms for …

200 papers

Maximum-likelihood estimation (MLE) is widely used in sequence to sequence tasks for model training. It uniformly treats the generation/prediction of each target token as multi-class classification, and yields non-smooth prediction…

Computation and Language · Computer Science 2018-12-13 Chengyue Gong , Xu Tan , Di He , Tao Qin

Model ensembling is a technique to combine the predicted distributions of two or more models, often leading to improved robustness and performance. For ensembling in text generation, the next token's probability distribution is derived from…

Computation and Language · Computer Science 2025-03-03 Rachel Wicks , Kartik Ravisankar , Xinchen Yang , Philipp Koehn , Matt Post

Subword tokenization methods, such as Byte-Pair Encoding (BPE), significantly impact the performance and efficiency of large language models (LLMs). The standard approach involves training a general-purpose tokenizer that uniformly…

Computation and Language · Computer Science 2026-01-30 Vijini Liyanage , François Yvon

Existing techniques for accelerating language model inference, such as speculative decoding, require training auxiliary speculator models and building and deploying complex inference pipelines. We consider a new approach for converting a…

Computation and Language · Computer Science 2026-04-27 John Kirchenbauer , Abhimanyu Hans , Brian Bartoldson , Micah Goldblum , Ashwinee Panda , Tom Goldstein

Knowledge Distillation (KD) has emerged as a prominent technique for model compression. However, conventional KD approaches primarily focus on homogeneous architectures with identical tokenizers, constraining their applicability in…

Computation and Language · Computer Science 2025-02-18 Yijie Chen , Yijin Liu , Fandong Meng , Yufeng Chen , Jinan Xu , Jie Zhou

This paper addresses the challenges of high computational cost and slow inference in deploying large language models. It proposes a distillation strategy guided by multiple teacher models. The method constructs several teacher models and…

Computation and Language · Computer Science 2025-07-22 Xiandong Meng , Yan Wu , Yexin Tian , Xin Hu , Tianze Kang , Junliang Du

Efficiency and safety of Large Language Models (LLMs), among other factors, rely on the quality of tokenization. A good tokenizer not only improves inference speed and language understanding but also provides extra defense against jailbreak…

Computation and Language · Computer Science 2026-04-16 Pavel Chizhov , Egor Bogomolov , Ivan P. Yamshchikov

Tokenization is the first -- and often least scrutinized -- step of most NLP pipelines. Standard algorithms for learning tokenizers rely on frequency-based objectives, which favor languages dominant in the training data and consequently…

Computation and Language · Computer Science 2025-08-25 Negar Foroutan , Clara Meister , Debjit Paul , Joel Niklaus , Sina Ahmadi , Antoine Bosselut , Rico Sennrich

Knowledge distillation, transferring knowledge from a teacher model to a student model, has emerged as a powerful technique in neural machine translation for compressing models or simplifying training targets. Knowledge distillation…

Computation and Language · Computer Science 2024-04-24 Jingxuan Wei , Linzhuang Sun , Yichong Leng , Xu Tan , Bihui Yu , Ruifeng Guo

Language models typically tokenize text into subwords, using a deterministic, hand-engineered heuristic of combining characters into longer surface-level strings such as 'ing' or whole words. Recent literature has repeatedly shown the…

Computation and Language · Computer Science 2023-10-19 Avijit Thawani , Saurabh Ghanekar , Xiaoyuan Zhu , Jay Pujara

Traditional greedy tokenization methods have been a critical step in Natural Language Processing (NLP), influencing how text is converted into tokens and directly impacting model performance. While subword tokenizers like Byte-Pair Encoding…

Computation and Language · Computer Science 2025-05-05 Bharath Raj , Garvit Suri , Vikrant Dewangan , Raghav Sonavane

Transformer-based large language models (LLMs) rely on contextual embeddings which generate different (continuous) representations for the same token depending on its surrounding context. Nonetheless, words and tokens typically have a…

Computation and Language · Computer Science 2025-07-10 Qitong Wang , Mohammed J. Zaki , Georgios Kollias , Vasileios Kalantzis

Prior research has demonstrated noticeable performance gains through the use of probabilistic tokenizations, an approach that involves employing multiple tokenizations of the same input string during the training phase of a language model.…

Computation and Language · Computer Science 2024-07-08 Ashutosh Sathe , Divyanshu Aggarwal , Sunayana Sitaram

The Tsetlin Machine (TM) is a propositional logic based model that uses conjunctive clauses to learn patterns from data. As with typical neural networks, the performance of a Tsetlin Machine is largely dependent on its parameter count, with…

Artificial Intelligence · Computer Science 2025-04-03 Calvin Kinateder

The assumption across nearly all language model (LM) tokenization schemes is that tokens should be subwords, i.e., contained within word boundaries. While providing a seemingly reasonable inductive bias, is this common practice limiting the…

Computation and Language · Computer Science 2025-08-28 Alisa Liu , Jonathan Hayase , Valentin Hofmann , Sewoong Oh , Noah A. Smith , Yejin Choi

Leveraging shared learning through Massively Multilingual Models, state-of-the-art machine translation models are often able to adapt to the paucity of data for low-resource languages. However, this performance comes at the cost of…

Computation and Language · Computer Science 2022-11-10 Harshita Diddee , Sandipan Dandapat , Monojit Choudhury , Tanuja Ganu , Kalika Bali

Ensembles of machine learning models yield improved system performance as well as robust and interpretable uncertainty estimates; however, their inference costs may often be prohibitively high. \emph{Ensemble Distribution Distillation} is…

Machine Learning · Computer Science 2021-05-17 Max Ryabinin , Andrey Malinin , Mark Gales

Many applications of large language models (LLMs) require long-context understanding, but models continue to struggle with such tasks. We hypothesize that conventional next-token prediction training could contribute to this, because each…

Computation and Language · Computer Science 2025-03-13 Falko Helm , Nico Daheim , Iryna Gurevych

Tokenization imposes a fixed granularity on the input text, freezing how a language model operates on data and how far in the future it predicts. Byte Pair Encoding (BPE) and similar schemes split text once, build a static vocabulary, and…

Computation and Language · Computer Science 2025-06-18 Mathurin Videau , Badr Youbi Idrissi , Alessandro Leite , Marc Schoenauer , Olivier Teytaud , David Lopez-Paz

Recent dynamic tokenisation methods operate directly on bytes and pool their latent representations into patches. This bears similarities to computational models of word segmentation that determine lexical boundaries using spikes in an…

Computation and Language · Computer Science 2025-06-24 Zébulon Goriely , Suchir Salhan , Pietro Lesci , Julius Cheng , Paula Buttery