# Removing Biases on CIFAR-5

Source: [Removing Biases on CIFAR-5](<https://www.sota2.com/research/sota/removing-biases-on-cifar-5>)

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

Data updated: 2026-02-26T06:06:16.798Z

Evaluation of model performance on the CIFAR-5 dataset specifically targeting bias removal, measured by Test, Retain, and Forget errors (mean and standard deviation).

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 | Removing Biases |
| Dataset (including any stated split) | CIFAR-5 |
| Metric | Test Error (mean) |
| Unit | PERCENT |
| Direction | Lower is better |
| Results loaded | 10 of 10 |
| 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

- [Test Error (mean)](<https://www.sota2.com/research/sota/removing-biases-on-cifar-5.md?metric=Test%20Error%20(mean)>): Lower is better; unit: PERCENT.

- [Test Error (std)](<https://www.sota2.com/research/sota/removing-biases-on-cifar-5.md?metric=Test%20Error%20(std)>): Lower is better; unit: PERCENT.

- [Retain Error (mean)](<https://www.sota2.com/research/sota/removing-biases-on-cifar-5.md?metric=Retain%20Error%20(mean)>): Lower is better; unit: PERCENT.

- [Retain Error (std)](<https://www.sota2.com/research/sota/removing-biases-on-cifar-5.md?metric=Retain%20Error%20(std)>): Lower is better; unit: PERCENT.

- [Forget Error (mean)](<https://www.sota2.com/research/sota/removing-biases-on-cifar-5.md?metric=Forget%20Error%20(mean)>): Lower is better; unit: PERCENT.

- [Forget Error (std)](<https://www.sota2.com/research/sota/removing-biases-on-cifar-5.md?metric=Forget%20Error%20(std)>): Lower is better; unit: PERCENT.

## Recorded results

| Method | Test Error (mean) | Result date | Report paper |
| --- | --- | --- | --- |
| CF-k | 22.67 | Not recorded | 68fa71011e209131cea4e7b9 |
| Finetune | 23.48 | Not recorded | 68fa71011e209131cea4e7b9 |
| SCRUB | 23.88 | Not recorded | 68fa71011e209131cea4e7b9 |
| Original | 24.08 | Not recorded | 68fa71011e209131cea4e7b9 |
| NTK | 24.16 | Not recorded | 68fa71011e209131cea4e7b9 |
| Retrain | 24.36 | Not recorded | 68fa71011e209131cea4e7b9 |
| EU-k | 25.87 | Not recorded | 68fa71011e209131cea4e7b9 |
| Bad-T | 25.87 | Not recorded | 68fa71011e209131cea4e7b9 |
| NegGrad+ | 26.07 | Not recorded | 68fa71011e209131cea4e7b9 |
| Fisher | 42.64 | Not recorded | 68fa71011e209131cea4e7b9 |

### CF-k

Test Error (mean): 22.67. Result ID: 699186293a6c31484908b731.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71011e209131cea4e7b9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67551e209131cea158d5.md>)

Recorded evaluation configuration: {"Architecture":"All-CNN","Scale":"Small-scale"}. This record does not assert independent reproduction.

### Finetune

Test Error (mean): 23.48. Result ID: 699186293a6c31484908b72d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71011e209131cea4e7b9.md>)

Recorded evaluation configuration: {"Architecture":"All-CNN","Scale":"Small-scale"}. This record does not assert independent reproduction.

### SCRUB

Test Error (mean): 23.88. Result ID: 699186293a6c31484908b734.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71011e209131cea4e7b9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71011e209131cea4e7b9.md>)

[Implementation](<https://github.com/Meghdad92/SCRUB>)

Recorded evaluation configuration: {"Architecture":"All-CNN","Scale":"Small-scale"}. This record does not assert independent reproduction.

### Original

Test Error (mean): 24.08. Result ID: 699186293a6c31484908b72c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71011e209131cea4e7b9.md>)

Recorded evaluation configuration: {"Architecture":"All-CNN","Scale":"Small-scale"}. This record does not assert independent reproduction.

### NTK

Test Error (mean): 24.16. Result ID: 699186293a6c31484908b72f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71011e209131cea4e7b9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa577a1e209131ce9a5d86.md>)

Recorded evaluation configuration: {"Architecture":"All-CNN","Scale":"Small-scale"}. This record does not assert independent reproduction.

### Retrain

Test Error (mean): 24.36. Result ID: 699186293a6c31484908b72b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71011e209131cea4e7b9.md>)

Recorded evaluation configuration: {"Architecture":"All-CNN","Scale":"Small-scale"}. This record does not assert independent reproduction.

### EU-k

Test Error (mean): 25.87. Result ID: 699186293a6c31484908b732.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71011e209131cea4e7b9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67551e209131cea158d5.md>)

Recorded evaluation configuration: {"Architecture":"All-CNN","Scale":"Small-scale"}. This record does not assert independent reproduction.

### Bad-T

Test Error (mean): 25.87. Result ID: 699186293a6c31484908b733.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71011e209131cea4e7b9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f961e209131cea46745.md>)

Recorded evaluation configuration: {"Architecture":"All-CNN","Scale":"Small-scale"}. This record does not assert independent reproduction.

### NegGrad+

Test Error (mean): 26.07. Result ID: 699186293a6c31484908b730.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71011e209131cea4e7b9.md>)

Recorded evaluation configuration: {"Architecture":"All-CNN","Scale":"Small-scale"}. This record does not assert independent reproduction.

### Fisher

Test Error (mean): 42.64. Result ID: 699186293a6c31484908b72e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa71011e209131cea4e7b9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa577a1e209131ce9a5d86.md>)

Recorded evaluation configuration: {"Architecture":"All-CNN","Scale":"Small-scale"}. This record does not assert independent reproduction.

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