# Continual Language Modeling on HOPE 8-domain, clean-regime 256M (3 seeds)

Source: [Continual Language Modeling on HOPE 8-domain, clean-regime 256M (3 seeds)](<https://www.sota2.com/research/sota/continual-language-modeling-on-hope-8-domain-clean-regime-256m-3-seeds>)

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

Data updated: 2026-04-27T07:19:45.933Z

Evaluation of Continual Language Modeling performance on the HOPE 8-domain dataset (clean regime, 256M tokens, averaged over 3 seeds), focusing on metrics like average perplexity, forgetting, and adaptation rate.

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 Language Modeling |
| Dataset (including any stated split) | HOPE 8-domain, clean-regime 256M (3 seeds) |
| Metric | Avg PPL |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 19 of 19 |
| 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

- [Avg PPL](<https://www.sota2.com/research/sota/continual-language-modeling-on-hope-8-domain-clean-regime-256m-3-seeds.md?metric=Avg%20PPL>): Lower is better; unit: SCALAR.

- [Forgetting](<https://www.sota2.com/research/sota/continual-language-modeling-on-hope-8-domain-clean-regime-256m-3-seeds.md?metric=Forgetting>): Lower is better; unit: SCALAR.

- [Adapt@500](<https://www.sota2.com/research/sota/continual-language-modeling-on-hope-8-domain-clean-regime-256m-3-seeds.md?metric=Adapt%40500>): Higher is better; unit: PERCENT.

- [Reduction in Forgetting](<https://www.sota2.com/research/sota/continual-language-modeling-on-hope-8-domain-clean-regime-256m-3-seeds.md?metric=Reduction%20in%20Forgetting>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Avg PPL | Result date | Report paper |
| --- | --- | --- | --- |
| Adaptive-OGP | 35.8 | Not recorded | 69eeed5b55a3c32af89f959f |
| OGP | 35.8 | Not recorded | 69eeed5b55a3c32af89f959f |
| Replay 0.5% | 36.2 | Not recorded | 69eeed5b55a3c32af89f959f |
| Replay 0.5% | 36.2 | Not recorded | 69eeed5b55a3c32af89f959f |
| FOPNG | 36.8 | Not recorded | 69eeed5b55a3c32af89f959f |
| Adam-NSCL | 36.8 | Not recorded | 69eeed5b55a3c32af89f959f |
| FOPNG | 36.8 | Not recorded | 69eeed5b55a3c32af89f959f |
| Adam-NSCL | 36.8 | Not recorded | 69eeed5b55a3c32af89f959f |
| A-GEM | 36.9 | Not recorded | 69eeed5b55a3c32af89f959f |
| TRGP | 36.9 | Not recorded | 69eeed5b55a3c32af89f959f |
| ROGO | 37.1 | Not recorded | 69eeed5b55a3c32af89f959f |
| SGP | 37.4 | Not recorded | 69eeed5b55a3c32af89f959f |
| MoFO | 37.4 | Not recorded | 69eeed5b55a3c32af89f959f |
| SGP | 37.4 | Not recorded | 69eeed5b55a3c32af89f959f |
| MoFO | 37.4 | Not recorded | 69eeed5b55a3c32af89f959f |
| EWC | 37.9 | Not recorded | 69eeed5b55a3c32af89f959f |
| EWC | 37.9 | Not recorded | 69eeed5b55a3c32af89f959f |
| Vanilla HOPE | 38.7 | Not recorded | 69eeed5b55a3c32af89f959f |
| Vanilla HOPE | 38.7 | Not recorded | 69eeed5b55a3c32af89f959f |

### Adaptive-OGP

Avg PPL: 35.8. Result ID: 69ef0dea47f34ce924d4d21f.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

Recorded evaluation configuration: {"Type":"OptSt"}. This record does not assert independent reproduction.

### OGP

Avg PPL: 35.8. Result ID: 69ef0dea47f34ce924d4d255.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

Recorded evaluation configuration: {"Type":"OptSt","Model scale":"256M HOPE","Seeds":"3","SVD rank":"Matched","Optimizer":"Matched","Update frequency":"Matched"}. This record does not assert independent reproduction.

### Replay 0.5%

Avg PPL: 36.2. Result ID: 69ef0dea47f34ce924d4d21e.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63d71e209131cea01262.md>)

Recorded evaluation configuration: {"Type":"ER"}. This record does not assert independent reproduction.

### Replay 0.5%

Avg PPL: 36.2. Result ID: 69ef0dea47f34ce924d4d254.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63d71e209131cea01262.md>)

Recorded evaluation configuration: {"Type":"ER","Model scale":"256M HOPE","Seeds":"3"}. This record does not assert independent reproduction.

### FOPNG

Avg PPL: 36.8. Result ID: 69ef0dea47f34ce924d4d21b.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69705d94e25ad2fa9d1ed177.md>)

Recorded evaluation configuration: {"Type":"Proj"}. This record does not assert independent reproduction.

### Adam-NSCL

Avg PPL: 36.8. Result ID: 69ef0dea47f34ce924d4d21c.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69931e209131cea22c9b.md>)

[Implementation](<https://github.com/ShipengWang/Adam-NSCL>)

Recorded evaluation configuration: {"Type":"Proj"}. This record does not assert independent reproduction.

### FOPNG

Avg PPL: 36.8. Result ID: 69ef0dea47f34ce924d4d250.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69705d94e25ad2fa9d1ed177.md>)

Recorded evaluation configuration: {"Type":"Proj","Model scale":"256M HOPE","Seeds":"3","SVD rank":"Matched","Optimizer":"Matched","Update frequency":"Matched"}. This record does not assert independent reproduction.

### Adam-NSCL

Avg PPL: 36.8. Result ID: 69ef0dea47f34ce924d4d252.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69931e209131cea22c9b.md>)

[Implementation](<https://github.com/ShipengWang/Adam-NSCL>)

Recorded evaluation configuration: {"Type":"Proj","Model scale":"256M HOPE","Seeds":"3","SVD rank":"Matched","Optimizer":"Matched","Update frequency":"Matched"}. This record does not assert independent reproduction.

### A-GEM

Avg PPL: 36.9. Result ID: 69ef0dea47f34ce924d4d24d.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65371e209131cea090ac.md>)

[Implementation](<https://github.com/facebookresearch/agem>)

Recorded evaluation configuration: {"Type":"Mem","Model scale":"256M HOPE","Seeds":"3"}. This record does not assert independent reproduction.

### TRGP

Avg PPL: 36.9. Result ID: 69ef0dea47f34ce924d4d251.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6bc21e209131cea3020a.md>)

Recorded evaluation configuration: {"Type":"Proj","Model scale":"256M HOPE","Seeds":"3","SVD rank":"Matched","Optimizer":"Matched","Update frequency":"Matched"}. This record does not assert independent reproduction.

### ROGO

Avg PPL: 37.1. Result ID: 69ef0dea47f34ce924d4d24f.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e791e209131cea4022c.md>)

Recorded evaluation configuration: {"Type":"Proj","Model scale":"256M HOPE","Seeds":"3","SVD rank":"Matched","Optimizer":"Matched","Update frequency":"Matched"}. This record does not assert independent reproduction.

### SGP

Avg PPL: 37.4. Result ID: 69ef0dea47f34ce924d4d21a.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e7b1e209131cea402d0.md>)

Recorded evaluation configuration: {"Type":"Proj"}. This record does not assert independent reproduction.

### MoFO

Avg PPL: 37.4. Result ID: 69ef0dea47f34ce924d4d21d.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d169.md>)

Recorded evaluation configuration: {"Type":"OptSt"}. This record does not assert independent reproduction.

### SGP

Avg PPL: 37.4. Result ID: 69ef0dea47f34ce924d4d24e.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e7b1e209131cea402d0.md>)

Recorded evaluation configuration: {"Type":"Proj","Model scale":"256M HOPE","Seeds":"3","SVD rank":"Matched","Optimizer":"Matched","Update frequency":"Matched"}. This record does not assert independent reproduction.

### MoFO

Avg PPL: 37.4. Result ID: 69ef0dea47f34ce924d4d253.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d169.md>)

Recorded evaluation configuration: {"Type":"OptSt","Model scale":"256M HOPE","Seeds":"3"}. This record does not assert independent reproduction.

### EWC

Avg PPL: 37.9. Result ID: 69ef0dea47f34ce924d4d219.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6cb31e209131cea3555e.md>)

Recorded evaluation configuration: {"Type":"Reg"}. This record does not assert independent reproduction.

### EWC

Avg PPL: 37.9. Result ID: 69ef0dea47f34ce924d4d24c.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6cb31e209131cea3555e.md>)

Recorded evaluation configuration: {"Type":"Reg","Model scale":"256M HOPE","Seeds":"3"}. This record does not assert independent reproduction.

### Vanilla HOPE

Avg PPL: 38.7. Result ID: 69ef0dea47f34ce924d4d218.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

Recorded evaluation configuration: {"Type":"—"}. This record does not assert independent reproduction.

### Vanilla HOPE

Avg PPL: 38.7. Result ID: 69ef0dea47f34ce924d4d24b.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed5b55a3c32af89f959f.md>)

Recorded evaluation configuration: {"Model scale":"256M HOPE","Seeds":"3"}. This record does not assert independent reproduction.

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