Image Classification on Food101 (Top-1 Accuracy and Delta)
92.33Top-1 AccuracyBiSigLIP
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BiSigLIPShots=16, Base Model=SigLIP, Adaptation Method=Bilinear adaptation2026.03 | 92.33 | 0.14 | |
| SigLIPShots=0, Base Model=SigLIP, Adaptation Method=Zero-Shot2026.03 | 92.19 | — | |
| BiCLIPShots=16, Base Model=CLIP, Adaptation Method=Bilinear adaptation2026.03 | 90.09 | 1.36 | |
| CLIPShots=0, Base Model=CLIP, Adaptation Method=Zero-Shot2026.03 | 88.73 | — | |
| Simple Averaging + E-PMQMerging Strategy=Simple Averaging, Quantization Method=E-PMQ, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 82.45 | — | |
| Simple AveragingMerging Strategy=Simple Averaging, Quantization Method=Full-precision, Bit-width=Full, Backbone=CLIP-ViT-B/322026.05 | 80.4 | — | |
| Simple Averaging + GPTQMerging Strategy=Simple Averaging, Quantization Method=GPTQ, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 80.22 | — | |
| TIES-Merging + E-PMQMerging Strategy=TIES-Merging, Quantization Method=E-PMQ, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 79.97 | — | |
| Simple Averaging + RTNMerging Strategy=Simple Averaging, Quantization Method=RTN, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 76.9 | — | |
| Simple Averaging + AWQMerging Strategy=Simple Averaging, Quantization Method=AWQ, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 76.1 | — | |
| Task Arithmetic + E-PMQMerging Strategy=Task Arithmetic, Quantization Method=E-PMQ, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 73.55 | — | |
| WUDI-Merging + E-PMQMerging Strategy=WUDI-Merging, Quantization Method=E-PMQ, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 72.08 | — | |
| WUDI-MergingMerging Strategy=WUDI-Merging, Quantization Method=Full-precision, Bit-width=Full, Backbone=CLIP-ViT-B/322026.05 | 53.1 | — | |
| TIES-MergingMerging Strategy=TIES-Merging, Quantization Method=Full-precision, Bit-width=Full, Backbone=CLIP-ViT-B/322026.05 | 50.1 | — | |
| WUDI-Merging + GPTQMerging Strategy=WUDI-Merging, Quantization Method=GPTQ, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 49.52 | — | |
| TIES-Merging + GPTQMerging Strategy=TIES-Merging, Quantization Method=GPTQ, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 48.66 | — | |
| TIES-Merging + RTNMerging Strategy=TIES-Merging, Quantization Method=RTN, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 46.9 | — | |
| WUDI-Merging + AWQMerging Strategy=WUDI-Merging, Quantization Method=AWQ, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 45.89 | — | |
| WUDI-Merging + RTNMerging Strategy=WUDI-Merging, Quantization Method=RTN, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 45.74 | — | |
| TIES-Merging + AWQMerging Strategy=TIES-Merging, Quantization Method=AWQ, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 44.11 | — | |
| Task ArithmeticMerging Strategy=Task Arithmetic, Quantization Method=Full-precision, Bit-width=Full, Backbone=CLIP-ViT-B/322026.05 | 15.1 | — | |
| Task Arithmetic + GPTQMerging Strategy=Task Arithmetic, Quantization Method=GPTQ, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 15.09 | — | |
| Task Arithmetic + RTNMerging Strategy=Task Arithmetic, Quantization Method=RTN, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 14.34 | — | |
| Task Arithmetic + AWQMerging Strategy=Task Arithmetic, Quantization Method=AWQ, Bit-width=4-bit, Backbone=CLIP-ViT-B/322026.05 | 13.29 | — |