paper-with-me

홈 › Papers

DistiLLM: Towards Streamlined Distillation for Large Language Models

2024-02-06 · Jongwoo Ko, Sungnyun Kim, Tianyi Chen, Se-Young Yun

Knowledge distillation (KD) is widely used for compressing a teacher model to a smaller student model, reducing its inference cost and memory footprint while preserving model capabilities. However, current KD methods for auto-regressive sequence models (e.g., large language models) suffer from missing a standardized objective function. Moreover, the recent use of student-generated outputs to address training-inference mismatches has significantly escalated computational costs. To tackle these issues, we introduce DistiLLM, a more effective and efficient KD framework for auto-regressive language models. DistiLLM comprises two components: (1) a novel skew Kullback-Leibler divergence loss, where we unveil and leverage its theoretical properties, and (2) an adaptive off-policy approach designed to enhance the efficiency in utilizing student-generated outputs. Extensive experiments, including instruction-following tasks, demonstrate the effectiveness of DistiLLM in building high-performing student models while achieving up to 4.3$\times$ speedup compared to recent KD methods.

📄 PDF Abstract BibTeX arXiv:2402.03898

Code (5)

jongwooko/distillm 공식 구현 jax
ghwang-s/abkd pytorch
jongwooko/distillm-2 pytorch
mingyukim87/synergynerf pytorch
tianyic/only_train_once pytorch

Tasks

Instruction FollowingKnowledge Distillation

Similar Papers 제목 키워드 기반

DistiLLM-2: A Contrastive Approach Boosts the Distillation of LLMs

2025-03-10 · Jongwoo Ko, Tianyi Chen, Sungnyun Kim, Tianyu Ding 외

Despite the success of distillation in large language models (LLMs), most prior work applies identical loss functions to both teacher- and student-generated data. These strategies overlook the synergy between loss formul…

Code GenerationInstruction Following

DistillMatch: Leveraging Knowledge Distillation from Vision Foundation Model for Multimodal Image Matching

2025-09-19 · Meng Yang, Fan Fan, Zizhuo Li, Songchu Deng 외 arxiv

Multimodal image matching seeks pixel-level correspondences between images of different modalities, crucial for cross-modal perception, fusion and analysis. However, the significant appearance differences between modalit…

Knowledge DistillationData AugmentationImage Matching

Rank-DistiLLM: Closing the Effectiveness Gap Between Cross-Encoders and LLMs for Passage Re-Ranking

2024-05-13 · Ferdinand Schlatt, Maik Fröbe, Harrisen Scells, Shengyao Zhuang 외

Cross-encoders distilled from large language models (LLMs) are often more effective re-rankers than cross-encoders fine-tuned on manually labeled data. However, distilled models do not match the effectiveness of their te…

Language ModellingLarge Language ModelPassage Re-RankingRe-Ranking

Dual-Model Defense: Safeguarding Diffusion Models from Membership Inference Attacks through Disjoint Data Splitting

2024-10-22 · Bao Q. Tran, Viet Nguyen, Anh Tran, Toan Tran

Diffusion models have demonstrated remarkable capabilities in image synthesis, but their recently proven vulnerability to Membership Inference Attacks (MIAs) poses a critical privacy concern. This paper introduces two no…

Image GenerationMemorization

Memory-Efficient Semi-Supervised Continual Learning: The World is its Own Replay Buffer

2021-01-23 · James Smith, Jonathan Balloch, Yen-Chang Hsu, Zsolt Kira

Rehearsal is a critical component for class-incremental continual learning, yet it requires a substantial memory budget. Our work investigates whether we can significantly reduce this memory budget by leveraging unlabele…

Continual LearningKnowledge DistillationOut of Distribution (OOD) Detection