English
Related papers

Related papers: Suppressing Final Layer Hidden State Jumps in Tran…

200 papers

We analyze the operation of transformer language adapters, which are small modules trained on top of a frozen language model to adapt its predictions to new target languages. We show that adapted predictions mostly evolve in the source…

Computation and Language · Computer Science 2024-06-11 Jesujoba O. Alabi , Marius Mosbach , Matan Eyal , Dietrich Klakow , Mor Geva

Modern large language models (LLMs) rely on system prompts to establish behavioral constraints and safety rules. Standard causal self-attention treats privileged instructions and untrusted user content with equal structural priority -- a…

Machine Learning · Computer Science 2026-05-12 Li Lixing

Last-layer retraining (LLR) methods -- wherein the last layer of a neural network is reinitialized and retrained on a held-out set following ERM training -- have garnered interest as an efficient approach to rectify dependence on spurious…

Machine Learning · Computer Science 2026-05-15 John C. Hill , Tyler LaBonte , Xinchen Zhang , Vidya Muthukumar

Handling the ever-increasing scale of contemporary deep learning and transformer-based models poses a significant challenge. Overparameterized Transformer networks outperform prior art in Natural Language processing and Computer Vision.…

Machine Learning · Computer Science 2024-11-05 Soheil Zibakhsh Shabgahi , Mohammad Sohail Shariff , Farinaz Koushanfar

Model merging has emerged as a lightweight paradigm for enhancing Large Language Models (LLMs), yet its underlying mechanisms remain poorly understood. In this work, we analyze late-stage pre-training trajectories and uncover a…

Machine Learning · Computer Science 2026-05-27 Wenjie Zhou , Bohan Wang , Hongtao Zhang , Chenxi Jia , Wei Chen , Xueqi Cheng

Fine-tuning safety-aligned language models for downstream tasks often leads to substantial degradation of refusal behavior, making models vulnerable to adversarial misuse. While prior work has shown that safety-relevant features are encoded…

Machine Learning · Computer Science 2026-05-05 Sadia Asif , Mohammad Mohammadi Amiri

Large Language Models demonstrate remarkable mathematical capabilities but at the same time struggle with abstract reasoning and planning. In this study, we explore whether Transformers can learn to abstract and generalize the rules…

Neural and Evolutionary Computing · Computer Science 2024-12-03 Mikhail Burtsev

Language models with the Transformers structure have shown great performance in natural language processing. However, there still poses problems when fine-tuning pre-trained language models on downstream tasks, such as over-fitting or…

Computation and Language · Computer Science 2023-05-12 Hongyi Yuan , Zheng Yuan , Chuanqi Tan , Fei Huang , Songfang Huang

The Transformer is widely used in natural language processing tasks. To train a Transformer however, one usually needs a carefully designed learning rate warm-up stage, which is shown to be crucial to the final performance but will slow…

Machine Learning · Computer Science 2020-06-30 Ruibin Xiong , Yunchang Yang , Di He , Kai Zheng , Shuxin Zheng , Chen Xing , Huishuai Zhang , Yanyan Lan , Liwei Wang , Tie-Yan Liu

It is generally thought that transformer-based large language models benefit from pre-training by learning generic linguistic knowledge that can be focused on a specific task during fine-tuning. However, we propose that much of the benefit…

Computation and Language · Computer Science 2024-06-19 Anna C. Marbut , John W. Chandler , Travis J. Wheeler

While machine learning models become more capable in discriminative tasks at scale, their ability to overcome biases introduced by training data has come under increasing scrutiny. Previous results suggest that there are two extremes of…

Machine Learning · Computer Science 2025-06-26 Nathan Stromberg , Christos Thrampoulidis , Lalitha Sankar

Despite achieving state-of-the-art performance on many NLP tasks, the high energy cost and long inference delay prevent Transformer-based pretrained language models (PLMs) from seeing broader adoption including for edge and mobile…

Computation and Language · Computer Science 2022-11-30 Canwen Xu , Julian McAuley

Training instability remains a critical challenge in large language model (LLM) pretraining, often manifesting as sudden gradient explosions that waste significant computational resources. We study training failures in a 5M-parameter…

Machine Learning · Computer Science 2026-02-03 Lianhai Ren , Yucheng Ding , Xiao Liu , Qianxiao Li , Peng Cheng , Yeyun Gong

Transformer-based Large Language Models (LLMs) have become a fixture in modern machine learning. Correspondingly, significant resources are allocated towards research that aims to further advance this technology, typically resulting in…

Machine Learning · Computer Science 2023-12-22 Pratyusha Sharma , Jordan T. Ash , Dipendra Misra

Quantizing optimizer states is becoming an important ingredient of memory-efficient large-scale pre-training, but the resulting optimizer dynamics remain only partially understood. We study low-precision exponential moving average (EMA)…

Machine Learning · Computer Science 2026-03-18 Kristi Topollai , Anna Choromanska

In spite of their huge success, transformer models remain difficult to scale in depth. In this work, we develop a unified signal propagation theory and provide formulae that govern the moments of the forward and backward signal through the…

Computation and Language · Computer Science 2024-07-19 Akhil Kedia , Mohd Abbas Zaidi , Sushil Khyalia , Jungho Jung , Harshith Goka , Haejun Lee

In recent years, great progress has been made in a variety of application domains thanks to the development of increasingly deeper neural networks. Unfortunately, the huge number of units of these networks makes them expensive both…

Computer Vision and Pattern Recognition · Computer Science 2018-10-12 Jose M. Alvarez , Mathieu Salzmann

Representation Misdirection for Unlearning (RMU), which steers model representation in the intermediate layer to a target random representation, is an effective method for large language model (LLM) unlearning. Despite its high performance,…

Computation and Language · Computer Science 2025-02-07 Dang Huu-Tien , Trung-Tin Pham , Hoang Thanh-Tung , Naoya Inoue

We evaluate three simple, normalization-centric changes to improve Transformer training. First, we show that pre-norm residual connections (PreNorm) and smaller initializations enable warmup-free, validation-based training with large…

Computation and Language · Computer Science 2020-01-01 Toan Q. Nguyen , Julian Salazar

Current transformer language models are trained with uniform computational budgets across all layers, implicitly assuming layer homogeneity. We challenge this assumption through empirical analysis of SmolLM2-135M, a 30-layer, 135M-parameter…

Machine Learning · Computer Science 2026-03-23 Tomasz Wietrzykowski