ML Research Wiki / Benchmarks / Model Compression / ImageNet

ImageNet

Model Compression Benchmark

Performance Over Time

📊 Showing 12 results | 📏 Metric: Top-1

Top Performing Models

Rank	Model	Paper	Top-1	Date	Code
1	ADLIK-MO-ResNet50+W4A4	Learned Step Size Quantization	77.88	2019-02-21	📦 zhutmost/lsq-net 📦 hustzxd/LSQuantization 📦 ZouJiu1/LSQplus
2	ADLIK-MO-ResNet50+W3A4	Learned Step Size Quantization	77.34	2019-02-21	📦 zhutmost/lsq-net 📦 hustzxd/LSQuantization 📦 ZouJiu1/LSQplus
3	ResNet-18 + 4bit-1dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compression and Quantization	70.52	2023-03-14	-
4	MobileNet-v1 + 4bit-1dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compression and Quantization	69.63	2023-03-14	-
5	ResNet-18 + 2bit-1dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compression and Quantization	68.63	2023-03-14	-
6	MobileNet-v1 + 2bit-1dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compression and Quantization	67.62	2023-03-14	-
7	ResNet-18 + 4bit-4dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compression and Quantization	66.10	2023-03-14	-
8	ResNet-18 + 2bit-2dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compression and Quantization	64.70	2023-03-14	-
9	MobileNet-v1 + 4bit-4dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compression and Quantization	61.40	2023-03-14	-
10	ResNet-18 + 1bit-1dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compression and Quantization	59.70	2023-03-14	-

All Papers (12)

Learned Step Size Quantization

2019

ADLIK-MO-ResNet50+W4A4

zhutmost/lsq-net hustzxd/LSQuantization

Learned Step Size Quantization

2019

ADLIK-MO-ResNet50+W3A4

zhutmost/lsq-net hustzxd/LSQuantization

R2 Loss: Range Restriction Loss for Model Compression and Quantization

2023

ResNet-18 + 4bit-1dim model compression using DKM

R2 Loss: Range Restriction Loss for Model Compression and Quantization

2023

MobileNet-v1 + 4bit-1dim model compression using DKM

R2 Loss: Range Restriction Loss for Model Compression and Quantization

2023

ResNet-18 + 2bit-1dim model compression using DKM

R2 Loss: Range Restriction Loss for Model Compression and Quantization

2023

MobileNet-v1 + 2bit-1dim model compression using DKM

R2 Loss: Range Restriction Loss for Model Compression and Quantization

2023

ResNet-18 + 4bit-4dim model compression using DKM

R2 Loss: Range Restriction Loss for Model Compression and Quantization

2023

ResNet-18 + 2bit-2dim model compression using DKM

R2 Loss: Range Restriction Loss for Model Compression and Quantization

2023

MobileNet-v1 + 4bit-4dim model compression using DKM

R2 Loss: Range Restriction Loss for Model Compression and Quantization

2023

ResNet-18 + 1bit-1dim model compression using DKM

R2 Loss: Range Restriction Loss for Model Compression and Quantization

2023

MobileNet-v1 + 2bit-2dim model compression using DKM

R2 Loss: Range Restriction Loss for Model Compression and Quantization

2023

MobileNet-v1 + 1bit-1dim model compression using DKM

ImageNet

Performance Over Time

Edit Benchmark Results

Edit Result

Top Performing Models

All Papers (12)

Learned Step Size Quantization

Learned Step Size Quantization

R2 Loss: Range Restriction Loss for Model Compression and Quantization

R2 Loss: Range Restriction Loss for Model Compression and Quantization

R2 Loss: Range Restriction Loss for Model Compression and Quantization

R2 Loss: Range Restriction Loss for Model Compression and Quantization

R2 Loss: Range Restriction Loss for Model Compression and Quantization

R2 Loss: Range Restriction Loss for Model Compression and Quantization

R2 Loss: Range Restriction Loss for Model Compression and Quantization

R2 Loss: Range Restriction Loss for Model Compression and Quantization

R2 Loss: Range Restriction Loss for Model Compression and Quantization

R2 Loss: Range Restriction Loss for Model Compression and Quantization

Model	Paper	Top-1	Date
ADLIK-MO-ResNet50+W4A4	Learned Step Size Quantization	77.88	2019-02-21
ADLIK-MO-ResNet50+W3A4	Learned Step Size Quantization	77.34	2019-02-21
ResNet-18 + 4bit-1dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compres…	70.52	2023-03-14
MobileNet-v1 + 4bit-1dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compres…	69.63	2023-03-14
ResNet-18 + 2bit-1dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compres…	68.63	2023-03-14
MobileNet-v1 + 2bit-1dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compres…	67.62	2023-03-14
ResNet-18 + 4bit-4dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compres…	66.10	2023-03-14
ResNet-18 + 2bit-2dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compres…	64.70	2023-03-14
MobileNet-v1 + 4bit-4dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compres…	61.40	2023-03-14
ResNet-18 + 1bit-1dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compres…	59.70	2023-03-14
MobileNet-v1 + 2bit-2dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compres…	53.99	2023-03-14
MobileNet-v1 + 1bit-1dim model compression using DKM	R2 Loss: Range Restriction Loss for Model Compres…	52.58	2023-03-14