paper-with-me

Papers

Sparse Teachers Can Be Dense with Knowledge

2022-10-08 · Yi Yang, Chen Zhang, Dawei Song

Recent advances in distilling pretrained language models have discovered that, besides the expressiveness of knowledge, the student-friendliness should be taken into consideration to realize a truly knowledgable teacher. Based on a pilot study, we find that over-parameterized teachers can produce expressive yet student-unfriendly knowledge and are thus limited in overall knowledgableness. To remove the parameters that result in student-unfriendliness, we propose a sparse teacher trick under the guidance of an overall knowledgable score for each teacher parameter. The knowledgable score is essentially an interpolation of the expressiveness and student-friendliness scores. The aim is to ensure that the expressive parameters are retained while the student-unfriendly ones are removed. Extensive experiments on the GLUE benchmark show that the proposed sparse teachers can be dense with knowledge and lead to students with compelling performance in comparison with a series of competitive baselines.

📄 PDF Abstract BibTeX arXiv:2210.03923

Code (1)

genezc/stark 공식 구현 pytorch

Similar Papers 제목 키워드 기반

Distilling Dense Representations for Ranking using Tightly-Coupled Teachers

2020-10-22 · Sheng-Chieh Lin, Jheng-Hong Yang, Jimmy Lin

We present an approach to ranking with dense representations that applies knowledge distillation to improve the recently proposed late-interaction ColBERT model. Specifically, we distill the knowledge from ColBERT's expr…

Knowledge Distillation

Driving Beyond Privilege: Distilling Dense-Reward Knowledge into Sparse-Reward Policies

2025-12-03 · Feeza Khan Khanzada, Jaerock Kwon arxiv

We study how to exploit dense simulator-defined rewards in vision-based autonomous driving without inheriting their misalignment with deployment metrics. In realistic simulators such as CARLA, privileged state (e.g., lan…

Reinforcement LearningAutonomous Driving

Monitored Distillation for Positive Congruent Depth Completion

2022-03-30 · Tian Yu Liu, Parth Agrawal, Allison Chen, Byung-Woo Hong 외

We propose a method to infer a dense depth map from a single image, its calibration, and the associated sparse point cloud. In order to leverage existing models (teachers) that produce putative depth maps, we propose an …

Depth CompletionImage ReconstructionKnowledge DistillationModel Selection

From Proprietary to Open-Source: Bridging the Distribution Gap via Multi-Agent Protocol Distillation in Agentic Search

2026-07-27 · Junlin Liu, Jiangwang Chen, Zixin Song, Shuaiyu Zhou 외 hf

Agentic search enables large language models to solve knowledge-intensive tasks by interleaving multi-step reasoning with retrieval, yet optimizing this with outcome-based reinforcement learning (RL) provides only sparse…

Knowledge DistillationReinforcement Learning

Born Again Neural Networks

2018-05-12 · ICML 2018 7 · Tommaso Furlanello, Zachary C. Lipton, Michael Tschannen, Laurent Itti 외

Knowledge distillation (KD) consists of transferring knowledge from one machine learning model (the teacher}) to another (the student). Commonly, the teacher is a high-capacity model with formidable performance, while th…

Image ClassificationKnowledge DistillationLanguage Modeling