paper-with-me

홈 › Papers

The Best of Both Worlds: Bridging Quality and Diversity in Data Selection with Bipartite Graph

2024-10-16 · Minghao Wu, Thuy-Trang Vu, Lizhen Qu, Gholamreza Haffari

The performance of large language models (LLMs) in natural language processing (NLP) tasks is significantly influenced by the quality and diversity of data used for supervised fine-tuning (SFT). Current data selection methods often focus solely on quality or diversity, leading to underperforming models due to suboptimal training data. In this paper, we introduce GraphFilter, a novel method that represents the dataset as a bipartite graph, linking sentences to their constituent n-grams. This representation effectively captures the relationships between sentences and linguistic patterns, facilitating the selection of sentences that enhance n-gram diversity. To balance quality and diversity during selection, we propose a priority function that combines the quality metric with the diversity metric in a multiplicative manner. GraphFilter iteratively selects high-priority sentences, updates the bipartite graph by removing covered n-grams, and re-calculates priorities to reflect the evolving data landscape. We conduct extensive experiments using three model backbones across six widely used benchmarks. The results demonstrate that GraphFilter outperforms all nine baseline approaches, achieving superior model performance and computational efficiency. Our analyses validate the effectiveness of our design choices, examine the subsets selected by GraphFilter and other methods, highlight the importance of instruction diversity, and explore the role of quality and diversity in relation to subset sizes. GraphFilter establishes a new foundation for effective data selection strategies, encouraging further research in data selection for LLMs.

📄 PDF Abstract BibTeX arXiv:2410.12458

Code (0)

등록된 구현이 없습니다.

Tasks

Computational EfficiencyDiversity

Methods 이 논문이 사용한 방법론

Focus 설명 없음

Similar Papers 제목 키워드 기반

WorldSense: Evaluating Real-world Omnimodal Understanding for Multimodal LLMs

2025-02-06 · Jack Hong, Shilin Yan, Jiayin Cai, XiaoLong Jiang 외

In this paper, we introduce WorldSense, the first benchmark to assess the multi-modal video understanding, that simultaneously encompasses visual, audio, and text inputs. In contrast to existing benchmarks, our WorldSens…

Video Understanding

Markov Chain Monte Carlo and Variational Inference: Bridging the Gap

2014-10-23 · Tim Salimans, Diederik P. Kingma, Max Welling

Recent advances in stochastic gradient variational inference have made it possible to perform variational Bayesian inference with posterior approximations containing auxiliary random variables. This enables us to explore…

Bayesian InferenceVariational Inference

WorldStereo: Bridging Camera-Guided Video Generation and Scene Reconstruction via 3D Geometric Memories

2026-03-02 · Yisu Zhang, Chenjie Cao, Tengfei Wang, Xuhui Zuo 외 arxiv

Recent advances in foundational Video Diffusion Models (VDMs) have yielded significant progress. Yet, despite the remarkable visual quality of generated videos, reconstructing consistent 3D scenes from these outputs rema…

3D ReconstructionScene GenerationVideo GenerationPoint Clouds

GEP-PG: Decoupling Exploration and Exploitation in Deep Reinforcement Learning Algorithms

2018-02-14 · ICML 2018 7 · Cédric Colas, Olivier Sigaud, Pierre-Yves Oudeyer

In continuous action domains, standard deep reinforcement learning algorithms like DDPG suffer from inefficient exploration when facing sparse or deceptive reward problems. Conversely, evolutionary and developmental meth…

Deep Reinforcement LearningDiversityreinforcement-learningReinforcement Learning+1

Adversarial Training and Provable Defenses: Bridging the Gap

2020-05-01 · ICLR 2020 1 · Mislav Balunovic, Martin Vechev

We present COLT, a new method to train neural networks based on a novel combination of adversarial training and provable defenses. The key idea is to model neural network training as a procedure which includes both, the …