HyperAI超神经

首页算力平台文档资讯论文教程数据集百科 SOTA LLM 模型天梯 GPU 天梯顶会

中文

HyperAI超神经

Model Compression On Imagenet

评估指标

Top-1

评测结果

各个模型在此基准测试上的表现结果

		Paper Title	Repository
ADLIK-MO-ResNet50+W4A4	77.878	Learned Step Size Quantization
ADLIK-MO-ResNet50+W3A4	77.34	Learned Step Size Quantization
ResNet-18 + 4bit-1dim model compression using DKM	70.52	R2 Loss: Range Restriction Loss for Model Compression and Quantization	-
MobileNet-v1 + 4bit-1dim model compression using DKM	69.63	R2 Loss: Range Restriction Loss for Model Compression and Quantization	-
ResNet-18 + 2bit-1dim model compression using DKM	68.63	R2 Loss: Range Restriction Loss for Model Compression and Quantization	-
MobileNet-v1 + 2bit-1dim model compression using DKM	67.62	R2 Loss: Range Restriction Loss for Model Compression and Quantization	-
ResNet-18 + 4bit-4dim model compression using DKM	66.1	R2 Loss: Range Restriction Loss for Model Compression and Quantization	-
ResNet-18 + 2bit-2dim model compression using DKM	64.7	R2 Loss: Range Restriction Loss for Model Compression and Quantization	-
MobileNet-v1 + 4bit-4dim model compression using DKM	61.4	R2 Loss: Range Restriction Loss for Model Compression and Quantization	-
ResNet-18 + 1bit-1dim model compression using DKM	59.7	R2 Loss: Range Restriction Loss for Model Compression and Quantization	-
MobileNet-v1 + 2bit-2dim model compression using DKM	53.99	R2 Loss: Range Restriction Loss for Model Compression and Quantization	-
MobileNet-v1 + 1bit-1dim model compression using DKM	52.58	R2 Loss: Range Restriction Loss for Model Compression and Quantization	-

0 of 12 row(s) selected.

Model Compression On Imagenet | SOTA | HyperAI超神经