Image Classification on ImageNet 34 (val)
0.7303Feature AccuracyAdaQuant
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| AdaQuantType=PTQ, Per-channel=false, Power-of-Two=false, Backbone=MobileNetV22021.09 | 0.7303 | 0.7303 | 0 | |
| ZeroQType=PTQ, Per-channel=false, Power-of-Two=false, Backbone=MobileNetV22021.09 | 0.7303 | 0.7291 | 0.12 | |
| QTType=QAT, Per-channel=false, Power-of-Two=false, Backbone=MobileNetV22021.09 | 0.719 | 0.709 | 1 | |
| SSBDType=PTQ, Per-channel=false, Power-of-Two=false, Backbone=MobileNetV22021.09 | 0.719 | 0.7129 | 0.61 | |
| KrishnamoorthiType=PTQ, Per-channel=true, Power-of-Two=false, Backbone=MobileNetV22021.09 | 0.719 | 0.697 | 2.2 | |
| Wu et alType=PTQ, Per-channel=true, Power-of-Two=false, Backbone=MobileNetV22021.09 | 0.7188 | 0.7114 | 0.74 | |
| HPTQType=PTQ, Per-channel=true, Power-of-Two=true, Backbone=MobileNetV22021.09 | 0.7181 | 0.7146 | 0.352 | |
| Nagel et alType=PTQ, Per-channel=false, Power-of-Two=false, Backbone=MobileNetV22021.09 | 0.7172 | 0.7099 | 0.73 | |
| Nagel et alType=PTQ, Per-channel=true, Power-of-Two=false, Backbone=MobileNetV22021.09 | 0.7172 | 0.7116 | 0.56 | |
| DFQType=PTQ, Per-channel=false, Power-of-Two=false, Backbone=MobileNetV22021.09 | 0.7172 | 0.7092 | 0.8 | |
| TQTType=QAT, Per-channel=false, Power-of-Two=true, Backbone=MobileNetV22021.09 | 0.717 | 0.718 | -0.1 | |
| Lee et alType=PTQ, Per-channel=false, Power-of-Two=false, Backbone=MobileNetV22021.09 | 0.7123 | 0.695 | 1.73 | |
| RVQuantType=QAT, Per-channel=false, Power-of-Two=false, Backbone=MobileNetV22021.09 | 0.701 | 0.7029 | -0.19 |