English
Related papers

Related papers: Weight Patching: Toward Source-Level Mechanistic L…

200 papers

Large Language Models (LLMs) are known to exhibit social, demographic, and gender biases, often as a consequence of the data on which they are trained. In this work, we adopt a mechanistic interpretability approach to analyze how such…

Computation and Language · Computer Science 2025-06-09 Bhavik Chandna , Zubair Bashir , Procheta Sen

The ability of Large Language Models (LLMs) to use external tools unlocks powerful real-world interactions, making rigorous evaluation essential. However, current benchmarks primarily report final accuracy, revealing what models can do but…

Computation and Language · Computer Science 2026-01-29 Qihao Wang , Yue Hu , Mingzhe Lu , Jiayue Wu , Yanbing Liu , Yuanmin Tang

Methods based on weight compensation, which iteratively apply quantization and weight compensation to minimize the output error, have recently demonstrated remarkable success in quantizing Large Language Models (LLMs). The representative…

Machine Learning · Computer Science 2026-04-10 Shuaiting Li , Juncan Deng , Kedong Xu , Rongtao Deng , Hong Gu , Minghan Jiang , Haibin Shen , Kejie Huang

Task arithmetic is a powerful technique for transferring skills between Large Language Models (LLMs), but it often suffers from negative interference when models have diverged during training. We address this limitation by first aligning…

Computation and Language · Computer Science 2025-11-17 Stefan Horoi , Sangwoo Cho , Supriyo Chakraborty , Shi-Xiong Zhang , Sambit Sahu , Guy Wolf , Genta Indra Winata

The fields of explainable AI and mechanistic interpretability aim to uncover the internal structure of neural networks, with circuit discovery as a central tool for understanding model computations. Existing approaches, however, rely on…

Machine Learning · Computer Science 2026-03-05 Elena Golimblevskaia , Aakriti Jain , Bruno Puri , Ammar Ibrahim , Wojciech Samek , Sebastian Lapuschkin

Despite the continuous research and evolution of language models, they sometimes underperform previous versions. Existing approaches to overcome these challenges are resource-intensive, highlighting the need for alternatives that enable…

Computation and Language · Computer Science 2026-02-19 Namkyung Yoon , Kyeonghyun Yoo , Wooyong Jung , Sanghong Kim , Hwangnam Kim

Model merging aims to efficiently combine the weights of multiple expert models, each trained on a specific task, into a single multi-task model, with strong performance across all tasks. When applied to all but the last layer of weights,…

Machine Learning · Computer Science 2024-10-17 Ekansh Sharma , Daniel M. Roy , Gintare Karolina Dziugaite

Although it is known that transformer language models (LMs) pass features from early layers to later layers, it is not well understood how this information is represented and routed by the model. We analyze a mechanism used in two LMs to…

Computation and Language · Computer Science 2025-05-12 Jack Merullo , Carsten Eickhoff , Ellie Pavlick

While training large language models (LLMs) from scratch can generate models with distinct functionalities and strengths, it comes at significant costs and may result in redundant capabilities. Alternatively, a cost-effective and compelling…

Computation and Language · Computer Science 2024-01-23 Fanqi Wan , Xinting Huang , Deng Cai , Xiaojun Quan , Wei Bi , Shuming Shi

Weight binarization has emerged as a promising strategy to reduce the complexity of large language models (LLMs). Existing approaches fall into post-training binarization, which is simple but causes severe performance loss, and…

Machine Learning · Statistics 2026-04-22 Ba-Hien Tran , Van Minh Nguyen

Large Language Models (LLMs) have shown high capabilities in several software development-related tasks such as program repair, documentation, code refactoring, debugging, and testing. However, training these models requires massive amount…

Software Engineering · Computer Science 2025-06-10 Meghdad Dehghan , Jie JW Wu , Fatemeh H. Fard , Ali Ouni

Weight tying, i.e. sharing parameters between input and output embedding matrices, is common practice in language model design, yet its impact on the learned embedding space remains poorly understood. In this paper, we show that tied…

Computation and Language · Computer Science 2026-03-30 Antonio Lopardo , Avyukth Harish , Catherine Arnett , Akshat Gupta

Foundation models are trained on broad data distributions, yielding generalist capabilities that enable many downstream applications but also expand the space of potential misuse and failures. This position paper argues that capability…

Machine Learning · Computer Science 2026-02-06 Shoaib Ahmed Siddiqui , Eleni Triantafillou , David Krueger , Adrian Weller

Pre-trained Large Language Models (LLMs) encapsulate large amounts of knowledge and take enormous amounts of compute to train. We make use of this resource, together with the observation that LLMs are able to transfer knowledge and…

Machine Learning · Computer Science 2025-01-14 Malcolm L. Wolff , Shenghao Yang , Kari Torkkola , Michael W. Mahoney

Large Language Models (LLMs) suffer from huge number of parameters, which restricts their deployment on edge devices. Weight sharing is one promising solution that encourages weight reuse, effectively reducing memory usage with less…

Computation and Language · Computer Science 2024-10-25 Zouying Cao , Yifei Yang , Hai Zhao

Transformer-based language models excel in NLP tasks, but fine-grained control remains challenging. This paper explores methods for manipulating transformer models through principled interventions at three levels: prompts, activations, and…

Computation and Language · Computer Science 2025-09-08 Faruk Alpay , Taylan Alpay

Multi-task learning (MTL) leverages a shared model to accomplish multiple tasks and facilitate knowledge transfer. Recent research on task arithmetic-based MTL demonstrates that merging the parameters of independently fine-tuned models can…

Machine Learning · Computer Science 2024-10-30 Li Shen , Anke Tang , Enneng Yang , Guibing Guo , Yong Luo , Lefei Zhang , Xiaochun Cao , Bo Du , Dacheng Tao

Recent studies have attempted to refine the Transformer architecture to demonstrate its effectiveness in Long-Term Time Series Forecasting (LTSF) tasks. Despite surpassing many linear forecasting models with ever-improving performance, we…

Machine Learning · Computer Science 2024-12-30 Peiwang Tang , Weitai Zhang

Recent advancements in building domain-specific large language models (LLMs) have shown remarkable success, especially in tasks requiring reasoning abilities like logical inference over complex relationships and multi-step problem solving.…

Modern deep learning usually treats models as separate artifacts: trained independently, specialized for particular purposes, and replaced when improved versions appear. This thesis studies model merging as an alternative paradigm:…

Machine Learning · Computer Science 2026-05-05 Donato Crisostomi