Image Classification on ImageNet-1K Subset Eval Original (val)
79.85Base AccuracyI-ViT
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| I-ViTBackbone=DeiT-Small, Year=2023, Compression Type=INT8 quant., Protocol=Original, Quantization Mode=PTQ2026.06 | 79.85 | 0.68 | 4 | — | |
| I-ViTBackbone=DeiT-Small, Year=2023, Compression Type=INT8 quant., Protocol=Original, Quantization Mode=QAT2026.06 | 79.85 | 0.27 | 4 | — | |
| HiReLCBackbone=DeiT-Small, Year=2026, Compression Type=W3–8 + prune size, Protocol=Subset eval.2026.06 | 77.18 | 2.17 | 5.59 | — | |
| AMCBackbone=ResNet-50, Year=2018, Compression Type=FLOPs pruning, Protocol=Original2026.06 | 76.15 | 1.81 | 2 | — | |
| HAQBackbone=ResNet-50, Year=2019, Compression Type=latency quant., Protocol=Original2026.06 | 76.15 | 0.81 | 1.4 | — | |
| HAWQ-V2Backbone=ResNet-50, Year=2020, Compression Type=W2–8 size, Protocol=Original2026.06 | 76.13 | 0.37 | 3.7 | — | |
| HiReLCBackbone=ResNet-50, Year=2026, Compression Type=W3–8 + prune size, Protocol=Subset eval.2026.06 | 74.12 | 0 | 6.32 | — |