Image Classification on ImageNet-1K (with DeiT-S)

Metric: Top 1 Accuracy (higher is better)

LeaderboardDataset

Loading chart...

Results

Sort:

#	Model↕	Top 1 Accuracy▼	Extra Data	Paper	Date↕	Code
1	MCTF ($r=16$)	80.1	No	Multi-criteria Token Fusion with One-step-ahead ...	2024-03-15	Code
2	dTPS	80.1	No	Joint Token Pruning and Squeezing Towards More A...	2023-04-21	Code
3	MCTF ($r=18$)	79.9	No	Multi-criteria Token Fusion with One-step-ahead ...	2024-03-15	Code
4	DiffRate	79.8	No	DiffRate : Differentiable Compression Rate for E...	2023-05-29	Code
5	PPT	79.8	No	PPT: Token Pruning and Pooling for Efficient Vis...	2023-10-03	Code
6	DynamicViT (80%)	79.8	No	DynamicViT: Efficient Vision Transformers with D...	2021-06-03	Code
7	EViT (80%)	79.8	No	Not All Patches are What You Need: Expediting Vi...	2022-02-16	Code
8	LTMP (80%)	79.8	No	Learned Thresholds Token Merging and Pruning for...	2023-07-20	Code
9	SPViT (3.9G)	79.8	No	SPViT: Enabling Faster Vision Transformers via S...	2021-12-27	Code
10	EViT (90%)	79.8	No	Not All Patches are What You Need: Expediting Vi...	2022-02-16	Code
11	DynamicViT (90%)	79.8	No	DynamicViT: Efficient Vision Transformers with D...	2021-06-03	Code
12	Base (DeiT-S)	79.8	No	Training data-efficient image transformers & dis...	2020-12-23	Code
13	ATS	79.7	No	Adaptive Token Sampling For Efficient Vision Tra...	2021-11-30	Code
14	eTPS	79.7	No	Joint Token Pruning and Squeezing Towards More A...	2023-04-21	Code
15	ToMe ($r=8$)	79.7	No	Token Merging: Your ViT But Faster	2022-10-17	Code
16	BAT (70%)	79.6	No	Beyond Attentive Tokens: Incorporating Token Imp...	2022-11-21	Code
17	AS-DeiT-S (65%)	79.6	No	Adaptive Sparse ViT: Towards Learnable Adaptive ...	2022-09-28	Code
18	LTMP (60%)	79.6	No	Learned Thresholds Token Merging and Pruning for...	2023-07-20	Code
19	MCTF ($r=20$)	79.5	No	Multi-criteria Token Fusion with One-step-ahead ...	2024-03-15	Code
20	DPS-ViT	79.5	No	Patch Slimming for Efficient Vision Transformers	2021-06-05	-
21	EViT (70%)	79.5	No	Not All Patches are What You Need: Expediting Vi...	2022-02-16	Code
22	PS-ViT	79.4	No	Patch Slimming for Efficient Vision Transformers	2021-06-05	-
23	ToMe ($r=13$)	79.4	No	Token Merging: Your ViT But Faster	2022-10-17	Code
24	EvoViT	79.4	No	Evo-ViT: Slow-Fast Token Evolution for Dynamic V...	2021-08-03	Code
25	SPViT (2.6G)	79.3	No	SPViT: Enabling Faster Vision Transformers via S...	2021-12-27	Code
26	BAT (60%)	79.3	No	Beyond Attentive Tokens: Incorporating Token Imp...	2022-11-21	Code
27	DynamicViT (70%)	79.3	No	DynamicViT: Efficient Vision Transformers with D...	2021-06-03	Code
28	S$^2$ViTE	79.2	No	Chasing Sparsity in Vision Transformers: An End-...	2021-06-08	Code
29	ToMe ($r=16$)	79.1	No	Token Merging: Your ViT But Faster	2022-10-17	Code
30	IA-RED$^2$	79.1	No	IA-RED$^2$: Interpretability-Aware Redundancy Re...	2021-06-23	-
31	BAT (50%)	79	No	Beyond Attentive Tokens: Incorporating Token Imp...	2022-11-21	Code
32	EViT (60%)	78.9	No	Not All Patches are What You Need: Expediting Vi...	2022-02-16	Code
33	AS-DeiT-S (50%)	78.7	No	Adaptive Sparse ViT: Towards Learnable Adaptive ...	2022-09-28	Code
34	BAT (40%)	78.6	No	Beyond Attentive Tokens: Incorporating Token Imp...	2022-11-21	Code
35	LTMP (45%)	78.6	No	Learned Thresholds Token Merging and Pruning for...	2023-07-20	Code
36	A-ViT	78.6	No	AdaViT: Adaptive Tokens for Efficient Vision Tra...	2021-12-14	Code
37	EViT (50%)	78.5	No	Not All Patches are What You Need: Expediting Vi...	2022-02-16	Code
38	HVT-S-1	78.3	No	Scalable Vision Transformers with Hierarchical P...	2021-03-19	Code
39	SPViT	78.3	No	Pruning Self-attentions into Convolutional Layer...	2021-11-23	Code
40	BAT (30%)	77.8	No	Beyond Attentive Tokens: Incorporating Token Imp...	2022-11-21	Code
41	BAT (20%)	76.4	No	Beyond Attentive Tokens: Incorporating Token Imp...	2022-11-21	Code

#1MCTF ($r=16$)
80.1
Top 1 Accuracy· 2024-03-15
Multi-criteria Token Fusion with One-step-ahead Attention for Efficient Vision Transformers Code
#2dTPSSOTA
80.1
Top 1 Accuracy· 2023-04-21
Joint Token Pruning and Squeezing Towards More Aggressive Compression of Vision Transformers Code
#3MCTF ($r=18$)
79.9
Top 1 Accuracy· 2024-03-15
Multi-criteria Token Fusion with One-step-ahead Attention for Efficient Vision Transformers Code
#4DiffRate
79.8
Top 1 Accuracy· 2023-05-29
DiffRate : Differentiable Compression Rate for Efficient Vision Transformers Code
#5PPT
79.8
Top 1 Accuracy· 2023-10-03
PPT: Token Pruning and Pooling for Efficient Vision Transformers Code
#6DynamicViT (80%)
79.8
Top 1 Accuracy· 2021-06-03
DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification Code
#7EViT (80%)
79.8
Top 1 Accuracy· 2022-02-16
Not All Patches are What You Need: Expediting Vision Transformers via Token Reorganizations Code
#8LTMP (80%)
79.8
Top 1 Accuracy· 2023-07-20
Learned Thresholds Token Merging and Pruning for Vision Transformers Code
#9SPViT (3.9G)
79.8
Top 1 Accuracy· 2021-12-27
SPViT: Enabling Faster Vision Transformers via Soft Token Pruning Code
#10EViT (90%)
79.8
Top 1 Accuracy· 2022-02-16
Not All Patches are What You Need: Expediting Vision Transformers via Token Reorganizations Code
#11DynamicViT (90%)
79.8
Top 1 Accuracy· 2021-06-03
DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification Code
#12Base (DeiT-S)SOTA
79.8
Top 1 Accuracy· 2020-12-23
Training data-efficient image transformers & distillation through attention Code
#13ATS
79.7
Top 1 Accuracy· 2021-11-30
Adaptive Token Sampling For Efficient Vision Transformers Code
#14eTPS
79.7
Top 1 Accuracy· 2023-04-21
Joint Token Pruning and Squeezing Towards More Aggressive Compression of Vision Transformers Code
#15ToMe ($r=8$)
79.7
Top 1 Accuracy· 2022-10-17
Token Merging: Your ViT But Faster Code
#16BAT (70%)
79.6
Top 1 Accuracy· 2022-11-21
Beyond Attentive Tokens: Incorporating Token Importance and Diversity for Efficient Vision Transformers Code
#17AS-DeiT-S (65%)
79.6
Top 1 Accuracy· 2022-09-28
Adaptive Sparse ViT: Towards Learnable Adaptive Token Pruning by Fully Exploiting Self-Attention Code
#18LTMP (60%)
79.6
Top 1 Accuracy· 2023-07-20
Learned Thresholds Token Merging and Pruning for Vision Transformers Code
#19MCTF ($r=20$)
79.5
Top 1 Accuracy· 2024-03-15
Multi-criteria Token Fusion with One-step-ahead Attention for Efficient Vision Transformers Code
#20DPS-ViT
79.5
Top 1 Accuracy· 2021-06-05
Patch Slimming for Efficient Vision Transformers
#21EViT (70%)
79.5
Top 1 Accuracy· 2022-02-16
Not All Patches are What You Need: Expediting Vision Transformers via Token Reorganizations Code
#22PS-ViT
79.4
Top 1 Accuracy· 2021-06-05
Patch Slimming for Efficient Vision Transformers
#23ToMe ($r=13$)
79.4
Top 1 Accuracy· 2022-10-17
Token Merging: Your ViT But Faster Code
#24EvoViT
79.4
Top 1 Accuracy· 2021-08-03
Evo-ViT: Slow-Fast Token Evolution for Dynamic Vision Transformer Code
#25SPViT (2.6G)
79.3
Top 1 Accuracy· 2021-12-27
SPViT: Enabling Faster Vision Transformers via Soft Token Pruning Code
#26BAT (60%)
79.3
Top 1 Accuracy· 2022-11-21
Beyond Attentive Tokens: Incorporating Token Importance and Diversity for Efficient Vision Transformers Code
#27DynamicViT (70%)
79.3
Top 1 Accuracy· 2021-06-03
DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification Code
#28S$^2$ViTE
79.2
Top 1 Accuracy· 2021-06-08
Chasing Sparsity in Vision Transformers: An End-to-End Exploration Code
#29ToMe ($r=16$)
79.1
Top 1 Accuracy· 2022-10-17
Token Merging: Your ViT But Faster Code
#30IA-RED$^2$
79.1
Top 1 Accuracy· 2021-06-23
IA-RED$^2$: Interpretability-Aware Redundancy Reduction for Vision Transformers
#31BAT (50%)
79
Top 1 Accuracy· 2022-11-21
Beyond Attentive Tokens: Incorporating Token Importance and Diversity for Efficient Vision Transformers Code
#32EViT (60%)
78.9
Top 1 Accuracy· 2022-02-16
Not All Patches are What You Need: Expediting Vision Transformers via Token Reorganizations Code
#33AS-DeiT-S (50%)
78.7
Top 1 Accuracy· 2022-09-28
Adaptive Sparse ViT: Towards Learnable Adaptive Token Pruning by Fully Exploiting Self-Attention Code
#34BAT (40%)
78.6
Top 1 Accuracy· 2022-11-21
Beyond Attentive Tokens: Incorporating Token Importance and Diversity for Efficient Vision Transformers Code
#35LTMP (45%)
78.6
Top 1 Accuracy· 2023-07-20
Learned Thresholds Token Merging and Pruning for Vision Transformers Code
#36A-ViT
78.6
Top 1 Accuracy· 2021-12-14
AdaViT: Adaptive Tokens for Efficient Vision Transformer Code
#37EViT (50%)
78.5
Top 1 Accuracy· 2022-02-16
Not All Patches are What You Need: Expediting Vision Transformers via Token Reorganizations Code
#38HVT-S-1
78.3
Top 1 Accuracy· 2021-03-19
Scalable Vision Transformers with Hierarchical Pooling Code
#39SPViT
78.3
Top 1 Accuracy· 2021-11-23
Pruning Self-attentions into Convolutional Layers in Single Path Code
#40BAT (30%)
77.8
Top 1 Accuracy· 2022-11-21
Beyond Attentive Tokens: Incorporating Token Importance and Diversity for Efficient Vision Transformers Code
#41BAT (20%)
76.4
Top 1 Accuracy· 2022-11-21
Beyond Attentive Tokens: Incorporating Token Importance and Diversity for Efficient Vision Transformers Code