# Continual Learning on Tiny-ImageNet compatible-shift

Source: [Continual Learning on Tiny-ImageNet compatible-shift](<https://www.sota2.com/research/sota/continual-learning-on-tiny-imagenet-compatible-shift>)

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

Data updated: 2026-03-23T11:45:37.674Z

Evaluation of Continual Learning performance on the Tiny-ImageNet compatible-shift dataset, reporting Old-task accuracy, New-task accuracy, and Harmonic mean accuracy.

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 | Continual Learning |
| Dataset (including any stated split) | Tiny-ImageNet compatible-shift |
| Metric | Old-task 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

- [Old-task Accuracy](<https://www.sota2.com/research/sota/continual-learning-on-tiny-imagenet-compatible-shift.md?metric=Old-task%20Accuracy>): Higher is better; unit: PERCENT.

- [New-task Accuracy](<https://www.sota2.com/research/sota/continual-learning-on-tiny-imagenet-compatible-shift.md?metric=New-task%20Accuracy>): Higher is better; unit: PERCENT.

- [Harmonic Mean Accuracy](<https://www.sota2.com/research/sota/continual-learning-on-tiny-imagenet-compatible-shift.md?metric=Harmonic%20Mean%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Old-task Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| SPMA-OG | 30.59 | Not recorded | 69c0c8f00636e5bf4afc3adb |
| ER-512 | 22.18 | Not recorded | 69c0c8f00636e5bf4afc3adb |
| Anchor CE | 19.83 | Not recorded | 69c0c8f00636e5bf4afc3adb |
| Plain FT | 8.38 | Not recorded | 69c0c8f00636e5bf4afc3adb |

### SPMA-OG

Old-task Accuracy: 30.59. Result ID: 69c127d747f34ce924a9b8f2.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8f00636e5bf4afc3adb.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c0c8f00636e5bf4afc3adb.md>)

Recorded evaluation configuration: {"Replay":"512"}. This record does not assert independent reproduction.

### ER-512

Old-task Accuracy: 22.18. Result ID: 69c127d747f34ce924a9b8f1.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8f00636e5bf4afc3adb.md>)

Recorded evaluation configuration: {"Replay":"512"}. This record does not assert independent reproduction.

### Anchor CE

Old-task Accuracy: 19.83. Result ID: 69c127d747f34ce924a9b8f0.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8f00636e5bf4afc3adb.md>)

Recorded evaluation configuration: {"Replay":"256"}. This record does not assert independent reproduction.

### Plain FT

Old-task Accuracy: 8.38. Result ID: 69c127d747f34ce924a9b8ef.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8f00636e5bf4afc3adb.md>)

Recorded evaluation configuration: {"Replay":"640"}. This record does not assert independent reproduction.

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