paper-with-me

홈 › Papers

Balancing Synthetic Data and Replay for Enhancing Task-Specific Capabilities

2025-10-13 · Urs Spiegelhalter, Jörg K. H. Franke, Frank Hutter arxiv

Adapting language models to new tasks through continued pretraining faces a fundamental trade-off: models must learn new capabilities while avoiding catastrophic forgetting of existing knowledge. While prior work has studied synthetic data generation techniques, the optimal replay ratios for balancing task performance and knowledge retention under computational constraints remain poorly understood. We present a comprehensive empirical study investigating the interplay between replay ratio configuration and computational budget when adapting language models to new tasks. Using the bAbI reasoning tasks as our target objective, we apply synthetic data generation and systematically evaluate different total token budgets and replay ratio configurations. We analyze their effects on both task mastery and general knowledge retention. Our experiments reveal an optimal configuration that balances task-specific performance with general knowledge retention. Based on our findings, we provide empirically-grounded guidelines for selecting replay ratios based on computational budget, enabling practitioners to achieve strong task adaptation with significantly reduced training costs.

📄 PDF Abstract BibTeX arXiv:2510.11842

Code (0)

등록된 구현이 없습니다.

Tasks

Synthetic Data GenerationGeneral Knowledge

Similar Papers 제목 키워드 기반

Balancing the Past and Present: A Coordinated Replay Framework for Federated Class-Incremental Learning

2025-07-10 · Zhuang Qi, Lei Meng, Han Yu

Federated Class Incremental Learning (FCIL) aims to collaboratively process continuously increasing incoming tasks across multiple clients. Among various approaches, data replay has become a promising solution, which can…

class-incremental learningClass Incremental LearningIncremental LearningPrivacy Preserving

Enhancing Continual Learning for Software Vulnerability Prediction: Addressing Catastrophic Forgetting via Hybrid-Confidence-Aware Selective Replay for Temporal LLM Fine-Tuning

2026-02-27 · Xuhui Dou, Hayretdin Bahsi, Alejandro Guerra-Manzanares arxiv

Recent work applies Large Language Models (LLMs) to source-code vulnerability detection, but most evaluations still rely on random train-test splits that ignore time and overestimate real-world performance. In practice, …

Vulnerability DetectionContinual Learning

LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning

2025-07-17 · Kaihong Wang, Donghyun Kim, Margrit Betke arxiv

Continual learning for vision-language models has achieved remarkable performance through synthetic replay, where samples are generated using Stable Diffusion to regularize during finetuning and retain knowledge. However…

Incremental LearningContinual Learning

CacheRoute: Planned Prefix-Affinity Routing for Large-Scale LLM Serving

2026-08-20 · Huang Cheng arxiv

Prefix caching avoids prefill only when a repeated request returns to a server that still holds the prefix KV. Cache-blind balancing disperses that reuse; fixed affinity preserves it but can overload a server. CacheRoute…

Synthetic Information towards Maximum Posterior Ratio for deep learning on Imbalanced Data

2024-01-05 · Hung Nguyen, Morris Chang

This study examines the impact of class-imbalanced data on deep learning models and proposes a technique for data balancing by generating synthetic data for the minority class. Unlike random-based oversampling, our metho…

Deep Learning