Enhancing LLM Character-Level Manipulation via Divide and Conquer

2025-02-12 · Zhen Xiong, Yujun Cai, Bryan Hooi, Nanyun Peng, Zhecheng Li, Yiwei Wang

Large Language Models (LLMs) have demonstrated strong generalization capabilities across a wide range of natural language processing (NLP) tasks. However, they exhibit notable weaknesses in character-level string manipulation, struggling with fundamental operations such as character deletion, insertion, and substitution. These challenges stem primarily from tokenization constraints, despite the critical role of such operations in data preprocessing and code generation. Through systematic analysis, we derive two key insights: (1) LLMs face significant difficulties in leveraging intrinsic token knowledge for character-level reasoning, and (2) atomized word structures can substantially enhance LLMs' ability to process token-level structural information. Building on these insights, we propose Character-Level Manipulation via Divide and Conquer, a novel approach designed to bridge the gap between token-level processing and character-level manipulation. Our method decomposes complex operations into explicit character-level subtasks coupled with controlled token reconstruction phases, leading to significant improvements in accuracy. Without additional training, our method significantly improves accuracies on the $\texttt{Deletion}$, $\texttt{Insertion}$, and $\texttt{Substitution}$ tasks. To support further research, we open-source our implementation and benchmarks.

📄 PDF Abstract BibTeX arXiv:2502.08180

Code (0)

등록된 구현이 없습니다.

Tasks

Code Generation

Component Divide-and-Conquer for Real-World Image Super-Resolution

2020-08-05 · ECCV 2020 8 · Pengxu Wei, Ziwei Xie, Hannan Lu, Zongyuan Zhan 외

In this paper, we present a large-scale Diverse Real-world image Super-Resolution dataset, i.e., DRealSR, as well as a divide-and-conquer Super-Resolution (SR) network, exploring the utility of guiding SR model with low-…

Image Super-Resolution Super-Resolution

Divide-and-Conquer Reinforcement Learning

2017-11-27 · ICLR 2018 1 · Dibya Ghosh, Avi Singh, Aravind Rajeswaran, Vikash Kumar 외

Standard model-free deep reinforcement learning (RL) algorithms sample a new initial state for each trial, allowing them to optimize policies that can perform well even in highly stochastic environments. However, problem…

Deep Reinforcement Learning Policy Gradient Methods reinforcement-learning Reinforcement Learning+1

Divide-Fuse-Conquer: Eliciting "Aha Moments" in Multi-Scenario Games

2025-05-22 · Xiaoqing Zhang, Huabin Zheng, Ang Lv, YuHan Liu 외

Large language models (LLMs) have been observed to suddenly exhibit advanced reasoning abilities during reinforcement learning (RL), resembling an ``aha moment'' triggered by simple outcome-based rewards. While RL has pr…

Reinforcement Learning (RL)

DC-Net: Divide-and-Conquer for Salient Object Detection

2023-05-24 · JiaYi Zhu, Xuebin Qin, Abdulmotaleb Elsaddik

In this paper, we introduce Divide-and-Conquer into the salient object detection (SOD) task to enable the model to learn prior knowledge that is for predicting the saliency map. We design a novel network, Divide-and-Conq…

Decoder Object object-detection Object Detection+1

Divide-and-Conquer for Enhancing Unlabeled Learning, Stability, and Plasticity in Semi-supervised Continual Learning

2025-08-07 · Yue Duan, Taicai Chen, Lei Qi, Yinghuan Shi arxiv

Semi-supervised continual learning (SSCL) seeks to leverage both labeled and unlabeled data in a sequential learning setup, aiming to reduce annotation costs while managing continual data arrival. SSCL introduces complex…

Continual Learning