English
Related papers

Related papers: Model Merging in Pre-training of Large Language Mo…

200 papers

The pre-trained language models are continually fine-tuned to better support downstream applications. However, this operation may result in significant performance degeneration on general tasks beyond the targeted domain. To overcome this…

Computation and Language · Computer Science 2023-12-11 Shitao Xiao , Zheng Liu , Peitian Zhang , Xingrun Xing

The fine-tuning of pre-trained language models has resulted in the widespread availability of task-specific models. Model merging offers an efficient way to create multi-task models by combining these fine-tuned models at the parameter…

Computation and Language · Computer Science 2025-04-29 Sanwoo Lee , Jiahao Liu , Qifan Wang , Jingang Wang , Xunliang Cai , Yunfang Wu

Mixing datasets for fine-tuning large models (LMs) has become critical for maximizing performance on downstream tasks. However, composing effective dataset mixtures typically relies on heuristics and trial-and-error, often requiring…

Machine Learning · Computer Science 2025-05-23 Zhixu Silvia Tao , Kasper Vinken , Hao-Wei Yeh , Avi Cooper , Xavier Boix

Large Language Models (LLMs) have demonstrated exceptional capabilities across diverse natural language processing (NLP) tasks. The release of open-source LLMs like LLaMA and Qwen has triggered the development of numerous fine-tuned models…

Computation and Language · Computer Science 2025-06-17 Zichuan Fu , Xian Wu , Yejing Wang , Wanyu Wang , Shanshan Ye , Hongzhi Yin , Yi Chang , Yefeng Zheng , Xiangyu Zhao

Determining an effective data mixture is a key factor in Large Language Model (LLM) pre-training, where models must balance general competence with proficiency on hard tasks such as math and code. However, identifying an optimal mixture…

Computation and Language · Computer Science 2026-05-18 Shengrui Li , Fei Zhao , Kaiyan Zhao , Jieying Ye , Haifeng Liu , Fangcheng Shi , Zheyong Xie , Yao Hu , Shaosheng Cao

Multi-task model merging offers a promising paradigm for integrating multiple expert models into a unified model without additional training. Existing state-of-the-art techniques, such as Task Arithmetic and its variants, merge models by…

Artificial Intelligence · Computer Science 2025-05-15 Wenju Sun , Qingyong Li , Yangli-ao Geng , Boyang Li

Large language models (LLMs) show best-in-class performance across a wide range of natural language processing applications. Training these models is an extremely computationally expensive task; frontier Artificial Intelligence (AI)…

Distributed, Parallel, and Cluster Computing · Computer Science 2025-10-10 Alexander Interrante-Grant , Carla Varela-Rosa , Suhaas Narayan , Chris Connelly , Albert Reuther

Large Language Models (LLMs) are often English-centric due to the disproportionate distribution of languages in their pre-training data. Enhancing non-English language capabilities through post-pretraining often results in catastrophic…

Computation and Language · Computer Science 2024-08-22 Hao Zhou , Zhijun Wang , Shujian Huang , Xin Huang , Xue Han , Junlan Feng , Chao Deng , Weihua Luo , Jiajun Chen

Recently, fine-tuning pre-trained language models (e.g., multilingual BERT) to downstream cross-lingual tasks has shown promising results. However, the fine-tuning process inevitably changes the parameters of the pre-trained model and…

Computation and Language · Computer Science 2020-10-06 Zihan Liu , Genta Indra Winata , Andrea Madotto , Pascale Fung

Model merging constructs versatile models by integrating task-specific models without requiring labeled data or expensive joint retraining. Although recent methods improve adaptability to heterogeneous tasks by generating customized merged…

Machine Learning · Computer Science 2026-02-09 Haiyun Qiu , Xingyu Wu , Liang Feng , Kay Chen Tan

Large language models are classically trained in stages: pretraining on raw text followed by post-training for instruction following and reasoning. However, this separation creates a fundamental limitation: many desirable behaviors such as…

We propose a method to optimize language model pre-training data mixtures through efficient approximation of the cross-entropy loss corresponding to each candidate mixture via a Mixture of Data Experts (MDE). We use this approximation as a…

Machine Learning · Computer Science 2025-02-25 Lior Belenki , Alekh Agarwal , Tianze Shi , Kristina Toutanova

Training data compositions for Large Language Models (LLMs) can significantly affect their downstream performance. However, a thorough data ablation study exploring large sets of candidate data mixtures is typically prohibitively expensive…

Computation and Language · Computer Science 2024-12-10 Clara Na , Ian Magnusson , Ananya Harsh Jha , Tom Sherborne , Emma Strubell , Jesse Dodge , Pradeep Dasigi

Merging has become a widespread way to cheaply combine individual models into a single model that inherits their capabilities and attains better performance. This popularity has spurred rapid development of many new merging methods, which…

Machine Learning · Computer Science 2024-09-30 Derek Tam , Yash Kant , Brian Lester , Igor Gilitschenski , Colin Raffel

Model merging acquires general capabilities without extra data or training by combining multiple models' parameters. Previous approaches achieve linear mode connectivity by aligning parameters into the same loss basin using permutation…

Machine Learning · Computer Science 2025-03-28 Yi-Kai Zhang , Jin Wang , Xu-Xiang Zhong , De-Chuan Zhan , Han-Jia Ye

Biological multimodal large language models (MLLMs) have emerged as powerful foundation models for scientific discovery. However, existing models are specialized to a single modality, limiting their ability to solve inherently cross-modal…

Machine Learning · Computer Science 2026-03-17 Wonbin Lee , Dongki Kim , Sung Ju Hwang

For most languages of the world, language model pre-training operates in a data-constrained regime where models must repeat their training data many times, degrading generalization. Two remedies exist: aggressive hyperparameter tuning such…

Machine Learning · Computer Science 2026-05-14 Paul Jeha , Anastasiia Sedova , Louis Béthune , Skyler Seto , Jes Frellsen , Pierre Ablin , Natalie Schluter

State-of-the-art vision-and-language models consist of many parameters and learn from enormous datasets, surpassing the amounts of linguistic data that children are exposed to as they acquire a language. This paper presents our approach to…

Computation and Language · Computer Science 2025-10-03 Ece Takmaz , Lisa Bylinina , Jakub Dotlacil

Multilingual Large Language Models (LLMs) struggle with cross-lingual tasks due to data imbalances between high-resource and low-resource languages, as well as monolingual bias in pre-training. Existing methods, such as bilingual…

Computation and Language · Computer Science 2026-04-14 Weihua Zheng , Chang Liu , Zhengyuan Liu , Xin Huang , Kui Wu , Muhammad Huzaifah Md Shahrin , Aiti Aw , Roy Ka-Wei Lee

Mixture-of-Experts (MoE) has gained increasing popularity as a promising framework for scaling up large language models (LLMs). However, training MoE from scratch in a large-scale setting still suffers from data-hungry and instability…

Computation and Language · Computer Science 2024-06-25 Tong Zhu , Xiaoye Qu , Daize Dong , Jiacheng Ruan , Jingqi Tong , Conghui He , Yu Cheng
‹ Prev 1 3 4 5 6 7 10 Next ›