SOTAVerified

Knowledge Distillation

Knowledge distillation is the process of transferring knowledge from a large model to a smaller one. While large models (such as very deep neural networks or ensembles of many models) have higher knowledge capacity than small models, this capacity might not be fully utilized.

Papers

Showing 601650 of 4240 papers

TitleStatusHype
Boosting Multi-Label Image Classification with Complementary Parallel Self-DistillationCode1
PointDistiller: Structured Knowledge Distillation Towards Efficient and Compact 3D DetectionCode1
IDEAL: Query-Efficient Data-Free Learning from Black-box ModelsCode1
Knowledge Distillation via the Target-aware TransformerCode1
Knowledge Distillation from A Stronger TeacherCode1
Exploring Extreme Parameter Compression for Pre-trained Language ModelsCode1
Directed Acyclic Transformer for Non-Autoregressive Machine TranslationCode1
Knowledge Distillation Meets Open-Set Semi-Supervised LearningCode1
DistilProtBert: A distilled protein language model used to distinguish between real proteins and their randomly shuffled counterpartsCode1
Spot-adaptive Knowledge DistillationCode1
Nearest Neighbor Knowledge Distillation for Neural Machine TranslationCode1
Curriculum Learning for Dense Retrieval DistillationCode1
Conformer and Blind Noisy Students for Improved Image Quality AssessmentCode1
Proto2Proto: Can you recognize the car, the way I do?Code1
On-Device Next-Item Recommendation with Self-Supervised Knowledge DistillationCode1
Eliminating Backdoor Triggers for Deep Neural Networks Using Attention Relation Graph DistillationCode1
Modeling Missing Annotations for Incremental Learning in Object DetectionCode1
DialoKG: Knowledge-Structure Aware Task-Oriented Dialogue GenerationCode1
MoEBERT: from BERT to Mixture-of-Experts via Importance-Guided AdaptationCode1
LRH-Net: A Multi-Level Knowledge Distillation Approach for Low-Resource Heart NetworkCode1
Overcoming Catastrophic Forgetting in Incremental Object Detection via Elastic Response DistillationCode1
Class-Incremental Learning by Knowledge Distillation with Adaptive Feature ConsolidationCode1
End-to-End Zero-Shot HOI Detection via Vision and Language Knowledge DistillationCode1
Distill-VQ: Learning Retrieval Oriented Vector Quantization By Distilling Knowledge from Dense EmbeddingsCode1
Feature Structure Distillation with Centered Kernel Alignment in BERT TransferringCode1
It's All In the Teacher: Zero-Shot Quantization Brought Closer to the TeacherCode1
Self-Distillation from the Last Mini-Batch for Consistency RegularizationCode1
Rainbow Keywords: Efficient Incremental Learning for Online Spoken Keyword SpottingCode1
Monitored Distillation for Positive Congruent Depth CompletionCode1
Instance Relation Graph Guided Source-Free Domain Adaptive Object DetectionCode1
Uncertainty-aware Contrastive Distillation for Incremental Semantic SegmentationCode1
Knowledge Distillation with the Reused Teacher ClassifierCode1
PCA-Based Knowledge Distillation Towards Lightweight and Content-Style Balanced Photorealistic Style Transfer ModelsCode1
Model LEGO: Creating Models Like Disassembling and Assembling Building BlocksCode1
Rich Feature Construction for the Optimization-Generalization DilemmaCode1
Ensembling and Knowledge Distilling of Large Sequence Taggers for Grammatical Error CorrectionCode1
R-DFCIL: Relation-Guided Representation Learning for Data-Free Class Incremental LearningCode1
SSD-KD: A Self-supervised Diverse Knowledge Distillation Method for Lightweight Skin Lesion Classification Using Dermoscopic ImagesCode1
DQ-BART: Efficient Sequence-to-Sequence Model via Joint Distillation and QuantizationCode1
Document-Level Relation Extraction with Adaptive Focal Loss and Knowledge DistillationCode1
Open-Vocabulary One-Stage Detection with Hierarchical Visual-Language Knowledge DistillationCode1
Fine-tuning Global Model via Data-Free Knowledge Distillation for Non-IID Federated LearningCode1
When Chosen Wisely, More Data Is What You Need: A Universal Sample-Efficient Strategy For Data AugmentationCode1
Graph Flow: Cross-layer Graph Flow Distillation for Dual Efficient Medical Image SegmentationCode1
SATS: Self-Attention Transfer for Continual Semantic SegmentationCode1
Unified Visual Transformer CompressionCode1
Representation Compensation Networks for Continual Semantic SegmentationCode1
Knowledge Distillation as Efficient Pre-training: Faster Convergence, Higher Data-efficiency, and Better TransferabilityCode1
Prediction-Guided Distillation for Dense Object DetectionCode1
Overcoming Catastrophic Forgetting beyond Continual Learning: Balanced Training for Neural Machine TranslationCode1
Show:102550
← PrevPage 13 of 85Next →

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1ScaleKD (T:BEiT-L S:ViT-B/14)Top-1 accuracy %86.43Unverified
2ScaleKD (T:Swin-L S:ViT-B/16)Top-1 accuracy %85.53Unverified
3ScaleKD (T:Swin-L S:ViT-S/16)Top-1 accuracy %83.93Unverified
4ScaleKD (T:Swin-L S:Swin-T)Top-1 accuracy %83.8Unverified
5KD++(T: regnety-16GF S:ViT-B)Top-1 accuracy %83.6Unverified
6VkD (T:RegNety 160 S:DeiT-S)Top-1 accuracy %82.9Unverified
7SpectralKD (T:Swin-S S:Swin-T)Top-1 accuracy %82.7Unverified
8ScaleKD (T:Swin-L S:ResNet-50)Top-1 accuracy %82.55Unverified
9DiffKD (T:Swin-L S: Swin-T)Top-1 accuracy %82.5Unverified
10DIST (T: Swin-L S: Swin-T)Top-1 accuracy %82.3Unverified
#ModelMetricClaimedVerifiedStatus
1SRD (T:resnet-32x4, S:shufflenet-v2)Top-1 Accuracy (%)79.86Unverified
2shufflenet-v2(T:resnet-32x4, S:shufflenet-v2)Top-1 Accuracy (%)78.76Unverified
3MV-MR (T: CLIP/ViT-B-16 S: resnet50)Top-1 Accuracy (%)78.6Unverified
4resnet8x4 (T: resnet32x4 S: resnet8x4)Top-1 Accuracy (%)78.28Unverified
5resnet8x4 (T: resnet32x4 S: resnet8x4 [modified])Top-1 Accuracy (%)78.08Unverified
6ReviewKD++(T:resnet-32x4, S:shufflenet-v2)Top-1 Accuracy (%)77.93Unverified
7ReviewKD++(T:resnet-32x4, S:shufflenet-v1)Top-1 Accuracy (%)77.68Unverified
8resnet8x4 (T: resnet32x4 S: resnet8x4)Top-1 Accuracy (%)77.5Unverified
9resnet8x4 (T: resnet32x4 S: resnet8x4)Top-1 Accuracy (%)76.68Unverified
10resnet8x4 (T: resnet32x4 S: resnet8x4)Top-1 Accuracy (%)76.31Unverified
#ModelMetricClaimedVerifiedStatus
1LSHFM (T: ResNet101 S: ResNet50)mAP93.17Unverified
2LSHFM (T: ResNet101 S: MobileNetV2)mAP90.14Unverified
#ModelMetricClaimedVerifiedStatus
1TIE-KD (T: Adabins S: MobileNetV2)RMSE2.43Unverified