Beyond Logits: Aligning Feature Dynamics for Effective Knowledge Distillation (2025.acl-long)
Copied to clipboard
Guoqiang Gong, Jiaxing Wang, Jin Xu, Deping Xiang, Zicheng Zhang, Leqi Shen, Yifeng Zhang, JunhuaShu JunhuaShu, ZhaolongXing ZhaolongXing, Zhen Chen, Pengzhang Liu, Ke Zhang
| Challenge: | Knowledge distillation (KD) compresses large language models into lightweight versions called student models. |
| Approach: | They propose to align the entire feature dynamics between teacher and student models by using two additional loss terms to achieve this. |
| Outcome: | The proposed method matches the entire feature dynamics between teacher and student models rather than just the final states. |
Similar Papers
On the Generalization vs Fidelity Paradox in Knowledge Distillation (2025.findings-acl)
Copied to clipboard
| Challenge: | Knowledge distillation (KD) is a key technique for compressing large language models into smaller ones while preserving performance. |
| Approach: | They propose to use knowledge distillation to compress large language models into smaller ones while preserving performance. |
| Outcome: | The proposed technique improves the performance of smaller models by 10% while providing only marginal benefits for larger models. |
Revisiting Knowledge Distillation for Autoregressive Language Models (2024.acl-long)
Copied to clipboard
| Challenge: | Autoregressive language models (LMs) are expensive and memory intensive, preventing the development of industrial applications. |
| Approach: | They propose an adaptive teaching approach to improve the KD of autoregressive language models by distilling knowledge into a small student model. |
| Outcome: | The proposed method can achieve consistent and significant performance gains across all model types and sizes. |
Knowledge Distillation for Language Models (2025.naacl-tutorial)
Copied to clipboard
| Challenge: | Knowledge distillation (KD) aims to transfer knowledge from a teacher to a student . this tutorial will cover topics ranging from LLM sequence compression to LLM self-distillation . |
| Approach: | They propose to introduce intermediate-layer matching and prediction matching . they will then present advanced techniques such as reinforcement learning-based KD and multi-teacher distillation . |
| Outcome: | This tutorial aims to provide participants with a comprehensive understanding of the techniques and applications of knowledge distillation for language models. |
Dynamic Knowledge Distillation for Pre-trained Language Models (2021.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods conduct knowledge distillation statically, e.g., student model aligns output distribution to teacher model on pre-defined training dataset. |
| Approach: | They propose a dynamic knowledge distillation that empowers the student to adjust the learning procedure according to its competency . they find it is promising and provide discussions on potential future directions towards more efficient methods . |
| Outcome: | The proposed method can boost student model performance while accelerating training . the proposed method reduces memory usage and accelerates model inference . |
AD-KD: Attribution-Driven Knowledge Distillation for Language Model Compression (2023.acl-long)
Copied to clipboard
| Challenge: | Existing knowledge distillation methods focus on the transfer of model-specific knowledge but overlook data-specific information. |
| Approach: | They propose an attribution-driven knowledge distillation approach which explores the token-level rationale behind the teacher model and transfers attribution knowledge to the student model. |
| Outcome: | The proposed method outperforms state-of-the-art methods on the GLUE benchmark and shows that it is more efficient than existing methods. |
Enhancing Knowledge Distillation of Large Language Models through Efficient Multi-Modal Distribution Alignment (2025.coling-main)
Copied to clipboard
| Challenge: | Existing knowledge distillation techniques for large language models are causing difficulties for student models to learn multi-modal probability distributions. |
| Approach: | They propose a ranking loss-based knowledge distillation method that encourages consistency of the ranking of peak predictions between teacher and student models. |
| Outcome: | The proposed method improves student models' ability to learn multi-modal distributions. |
Why Skip If You Can Combine: A Simple Knowledge Distillation Technique for Intermediate Layers (2020.emnlp-main)
Copied to clipboard
| Challenge: | Existing knowledge distillation techniques are not suitable for deep learning tasks due to memory constraints. |
| Approach: | They propose to combine knowledge from a large teacher network into a student network (S) they propose to use a combinatorial mechanism to inject layer-level supervision from T to S . |
| Outcome: | The proposed model outperforms existing models in PortugueseEnglish, TurkishEnglish and EnglishGerman directions and students trained using it have 50% fewer parameters and can deliver comparable results to 12-layer teachers. |
Self-Evolution Knowledge Distillation for LLM-based Machine Translation (2025.coling-main)
Copied to clipboard
| Challenge: | Existing knowledge distillation strategies for large language models minimize output distributions between student and teacher models indiscriminately for each token. |
| Approach: | They propose a distillation strategy that integrates teacher and one-hot distribution of ground truth into the student distribution as prior knowledge, which promotes the distillation process. |
| Outcome: | The proposed method brings an average improvement of approximately 1.4 SacreBLEU points across four translation directions in the WMT22 test sets. |
Pre-training Distillation for Large Language Models: A Design Space Exploration (2025.acl-long)
Copied to clipboard
| Challenge: | Knowledge distillation (KD) aims to transfer knowledge from a large teacher model to a smaller student model for model compression. |
| Approach: | They extend knowledge distillation to the pre-training phase of large language models . they first conduct an experiment using a teacher LLM to distill a 1.9B student LLM . |
| Outcome: | The proposed model can be used to distill a 1.9B student model using a teacher LLM. |
Continuation KD: Improved Knowledge Distillation through the Lens of Continuation Optimization (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for knowledge distillation (KD) do not mitigate the noise in the teacher’s output: modeling the noisy behaviour of the teacher can distract the student from learning more useful features. |
| Approach: | They propose a method that optimizes the highly non-convex KD objective by starting with the smoothed version of this objective and making it more complex as the training proceeds. |
| Outcome: | The proposed method achieves state-of-the-art performance on NLU and computer vision tasks. |