# Classification on IMAGENET-R zero-shot

Source: [Classification on IMAGENET-R zero-shot](<https://www.sota2.com/research/sota/classification-on-imagenet-r-zero-shot>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:32:49.437Z

Evaluation of Classification performance on the IMAGENET-R zero-shot configuration, focusing on accuracy (ACC), Normalized Prediction Difference (NLPD), and Expected Calibration Error (ECE).

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Classification |
| Dataset (including any stated split) | IMAGENET-R zero-shot |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Accuracy](<https://www.sota2.com/research/sota/classification-on-imagenet-r-zero-shot.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [NLPD](<https://www.sota2.com/research/sota/classification-on-imagenet-r-zero-shot.md?metric=NLPD>): Lower is better; unit: SCALAR.

- [ECE](<https://www.sota2.com/research/sota/classification-on-imagenet-r-zero-shot.md?metric=ECE>): Lower is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| TTA | 90.91 | Not recorded | 68fa7bc61e209131cea8a5eb |
| BayesVLM | 89.72 | Not recorded | 68fa7bc61e209131cea8a5eb |
| SigLIP | 89.71 | Not recorded | 68fa7bc61e209131cea8a5eb |
| SigLIP (temp. scaling) | 89.71 | Not recorded | 68fa7bc61e209131cea8a5eb |

### TTA

Accuracy: 90.91. Result ID: 699a620a5698665ec4ea1349.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bc61e209131cea8a5eb.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75901e209131cea68a6d.md>)

[Implementation](<https://github.com/FarinaMatteo/zero>)

Recorded evaluation configuration: {"Backbone":"SigLIP-Base","Zero-shot":true}. This record does not assert independent reproduction.

### BayesVLM

Accuracy: 89.72. Result ID: 699a620a5698665ec4ea134a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bc61e209131cea8a5eb.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bc61e209131cea8a5eb.md>)

Recorded evaluation configuration: {"Backbone":"SigLIP-Base","Zero-shot":true}. This record does not assert independent reproduction.

### SigLIP

Accuracy: 89.71. Result ID: 699a620a5698665ec4ea1347.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bc61e209131cea8a5eb.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70ae1e209131cea4c7a4.md>)

Recorded evaluation configuration: {"Backbone":"SigLIP-Base","Zero-shot":true}. This record does not assert independent reproduction.

### SigLIP (temp. scaling)

Accuracy: 89.71. Result ID: 699a620a5698665ec4ea1348.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bc61e209131cea8a5eb.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa638b1e209131ce9ff6d1.md>)

Recorded evaluation configuration: {"Backbone":"SigLIP-Base","Zero-shot":true}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.