Image Classification on Oxford Flowers (Top-1/Top-5 Accuracy)
74.03Top-1 AccuracySimple Averaging w/ E-PMQ
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Simple Averaging w/ E-PMQMerging Strategy=Simple Averaging, Quantization Method=E-PMQ, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 74.03 | — | |
| TIES-Merging w/ E-PMQMerging Strategy=TIES-Merging, Quantization Method=E-PMQ, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 72.37 | — | |
| WUDI-Merging w/ E-PMQMerging Strategy=WUDI-Merging, Quantization Method=E-PMQ, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 70.37 | — | |
| Task Arithmetic w/ E-PMQMerging Strategy=Task Arithmetic, Quantization Method=E-PMQ, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 69.69 | — | |
| Simple AveragingMerging Strategy=Simple Averaging, Quantization Method=None, Model Backbone=CLIP-ViT-B/32, Bit-width=Full2026.05 | 67.4 | — | |
| Simple Averaging w/ GPTQMerging Strategy=Simple Averaging, Quantization Method=GPTQ, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 64.35 | — | |
| WUDI-MergingMerging Strategy=WUDI-Merging, Quantization Method=None, Model Backbone=CLIP-ViT-B/32, Bit-width=Full2026.05 | 63.8 | — | |
| WUDI-Merging w/ GPTQMerging Strategy=WUDI-Merging, Quantization Method=GPTQ, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 61.26 | — | |
| Simple Averaging w/ RTNMerging Strategy=Simple Averaging, Quantization Method=RTN, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 60.56 | — | |
| Simple Averaging w/ AWQMerging Strategy=Simple Averaging, Quantization Method=AWQ, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 59.88 | — | |
| TIES-MergingMerging Strategy=TIES-Merging, Quantization Method=None, Model Backbone=CLIP-ViT-B/32, Bit-width=Full2026.05 | 58.2 | — | |
| WUDI-Merging w/ AWQMerging Strategy=WUDI-Merging, Quantization Method=AWQ, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 55.68 | — | |
| TIES-Merging w/ GPTQMerging Strategy=TIES-Merging, Quantization Method=GPTQ, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 55.6 | — | |
| WUDI-Merging w/ RTNMerging Strategy=WUDI-Merging, Quantization Method=RTN, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 54.92 | — | |
| TIES-Merging w/ RTNMerging Strategy=TIES-Merging, Quantization Method=RTN, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 50.54 | — | |
| TIES-Merging w/ AWQMerging Strategy=TIES-Merging, Quantization Method=AWQ, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 50.2 | — | |
| Task ArithmeticMerging Strategy=Task Arithmetic, Quantization Method=None, Model Backbone=CLIP-ViT-B/32, Bit-width=Full2026.05 | 38.6 | — | |
| Task Arithmetic w/ GPTQMerging Strategy=Task Arithmetic, Quantization Method=GPTQ, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 35.92 | — | |
| Task Arithmetic w/ AWQMerging Strategy=Task Arithmetic, Quantization Method=AWQ, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 34.62 | — | |
| Task Arithmetic w/ RTNMerging Strategy=Task Arithmetic, Quantization Method=RTN, Model Backbone=CLIP-ViT-B/32, Bit-width=4-bit2026.05 | 32.38 | — | |
| CE+SIL+SupCon2Loss function=Cross-Entropy + Silhouette + SupCon22026.03 | 27.4 | 56.11 | |
| CE+SILLoss function=Cross-Entropy + Silhouette2026.03 | 26.56 | 54.56 | |
| ProxyNCALoss function=Proxy Neighborhood Component Analysis2026.03 | 22.1 | — | |
| SupCon2Loss function=Supervised Contrastive (two-view)2026.03 | 21.06 | 50.25 | |
| CenterLoss function=Center Loss2026.03 | 20.18 | 47.89 | |
| CELoss function=Cross-Entropy2026.03 | 19.6 | 47.21 | |
| SupConLoss function=Supervised Contrastive2026.03 | 17.91 | 46.54 |