# In-context Learning on Fineweb-Edu 16.8B tokens

Source: [In-context Learning on Fineweb-Edu 16.8B tokens](<https://www.sota2.com/research/sota/in-context-learning-on-fineweb-edu-16-8b-tokens>)

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

Data updated: 2026-03-17T10:00:46.067Z

Evaluation of in-context learning capabilities using the 16.8 Billion token subset of the FineWeb-Edu dataset, measured across standard knowledge and reasoning benchmarks (MMLU, ARC, HellaSwag, etc.).

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 | In-context Learning |
| Dataset (including any stated split) | Fineweb-Edu 16.8B tokens |
| Metric | ARC-c Accuracy |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 8 of 8 |
| 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

- [ARC-c Accuracy](<https://www.sota2.com/research/sota/in-context-learning-on-fineweb-edu-16-8b-tokens.md?metric=ARC-c%20Accuracy>): Higher is better; unit: SCALAR.

- [ARC-e Accuracy](<https://www.sota2.com/research/sota/in-context-learning-on-fineweb-edu-16-8b-tokens.md?metric=ARC-e%20Accuracy>): Higher is better; unit: SCALAR.

- [BoolQ Accuracy](<https://www.sota2.com/research/sota/in-context-learning-on-fineweb-edu-16-8b-tokens.md?metric=BoolQ%20Accuracy>): Higher is better; unit: SCALAR.

- [PIQA Accuracy](<https://www.sota2.com/research/sota/in-context-learning-on-fineweb-edu-16-8b-tokens.md?metric=PIQA%20Accuracy>): Higher is better; unit: SCALAR.

- [HellaSwag Accuracy](<https://www.sota2.com/research/sota/in-context-learning-on-fineweb-edu-16-8b-tokens.md?metric=HellaSwag%20Accuracy>): Higher is better; unit: SCALAR.

- [WinoGrande Accuracy](<https://www.sota2.com/research/sota/in-context-learning-on-fineweb-edu-16-8b-tokens.md?metric=WinoGrande%20Accuracy>): Higher is better; unit: SCALAR.

- [OBQA Accuracy](<https://www.sota2.com/research/sota/in-context-learning-on-fineweb-edu-16-8b-tokens.md?metric=OBQA%20Accuracy>): Higher is better; unit: SCALAR.

- [Lambda-OpenAI Score](<https://www.sota2.com/research/sota/in-context-learning-on-fineweb-edu-16-8b-tokens.md?metric=Lambda-OpenAI%20Score>): Higher is better; unit: SCALAR.

- [SciQ Accuracy](<https://www.sota2.com/research/sota/in-context-learning-on-fineweb-edu-16-8b-tokens.md?metric=SciQ%20Accuracy>): Higher is better; unit: SCALAR.

- [MMLU Accuracy](<https://www.sota2.com/research/sota/in-context-learning-on-fineweb-edu-16-8b-tokens.md?metric=MMLU%20Accuracy>): Higher is better; unit: SCALAR.

- [Average Score](<https://www.sota2.com/research/sota/in-context-learning-on-fineweb-edu-16-8b-tokens.md?metric=Average%20Score>): Higher is better; unit: SCALAR.

## Recorded results

| Method | ARC-c Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Spectra-AdEMAMix | 36.86 | Not recorded | 69b8e03d0636e5bf4afbed96 |
| D-Muon | 36.6 | Not recorded | 69b8e03d0636e5bf4afbed96 |
| Spectra-AdamW | 35.67 | Not recorded | 69b8e03d0636e5bf4afbed96 |
| Spectra-Signum | 35.58 | Not recorded | 69b8e03d0636e5bf4afbed96 |
| AdEMAMix | 35.41 | Not recorded | 69b8e03d0636e5bf4afbed96 |
| Mars | 34.98 | Not recorded | 69b8e03d0636e5bf4afbed96 |
| Signum | 34.56 | Not recorded | 69b8e03d0636e5bf4afbed96 |
| AdamW | 33.45 | Not recorded | 69b8e03d0636e5bf4afbed96 |

### Spectra-AdEMAMix

ARC-c Accuracy: 36.86. Result ID: 69b9261747f34ce924a24ecf.

[Reporting paper](<https://www.sota2.com/research/paper/69b8e03d0636e5bf4afbed96.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69b8e03d0636e5bf4afbed96.md>)

Recorded evaluation configuration: {"Model size":"820M","Protocol":"5-shot","Pre-training tokens":"16.8B","Harness":"lm-evaluation-harness"}. This record does not assert independent reproduction.

### D-Muon

ARC-c Accuracy: 36.6. Result ID: 69b9261747f34ce924a24ec9.

[Reporting paper](<https://www.sota2.com/research/paper/69b8e03d0636e5bf4afbed96.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa771a1e209131cea70fcd.md>)

Recorded evaluation configuration: {"Model size":"820M","Protocol":"5-shot","Pre-training tokens":"16.8B","Harness":"lm-evaluation-harness"}. This record does not assert independent reproduction.

### Spectra-AdamW

ARC-c Accuracy: 35.67. Result ID: 69b9261747f34ce924a24ecd.

[Reporting paper](<https://www.sota2.com/research/paper/69b8e03d0636e5bf4afbed96.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69b8e03d0636e5bf4afbed96.md>)

Recorded evaluation configuration: {"Model size":"820M","Protocol":"5-shot","Pre-training tokens":"16.8B","Harness":"lm-evaluation-harness"}. This record does not assert independent reproduction.

### Spectra-Signum

ARC-c Accuracy: 35.58. Result ID: 69b9261747f34ce924a24ecb.

[Reporting paper](<https://www.sota2.com/research/paper/69b8e03d0636e5bf4afbed96.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69b8e03d0636e5bf4afbed96.md>)

Recorded evaluation configuration: {"Model size":"820M","Protocol":"5-shot","Pre-training tokens":"16.8B","Harness":"lm-evaluation-harness"}. This record does not assert independent reproduction.

### AdEMAMix

ARC-c Accuracy: 35.41. Result ID: 69b9261747f34ce924a24ece.

[Reporting paper](<https://www.sota2.com/research/paper/69b8e03d0636e5bf4afbed96.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74f71e209131cea64ede.md>)

Recorded evaluation configuration: {"Model size":"820M","Protocol":"5-shot","Pre-training tokens":"16.8B","Harness":"lm-evaluation-harness"}. This record does not assert independent reproduction.

### Mars

ARC-c Accuracy: 34.98. Result ID: 69b9261747f34ce924a24ec8.

[Reporting paper](<https://www.sota2.com/research/paper/69b8e03d0636e5bf4afbed96.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa5e281e209131ce9dfb23.md>)

Recorded evaluation configuration: {"Model size":"820M","Protocol":"5-shot","Pre-training tokens":"16.8B","Harness":"lm-evaluation-harness"}. This record does not assert independent reproduction.

### Signum

ARC-c Accuracy: 34.56. Result ID: 69b9261747f34ce924a24eca.

[Reporting paper](<https://www.sota2.com/research/paper/69b8e03d0636e5bf4afbed96.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa55161e209131ce9911bc.md>)

Recorded evaluation configuration: {"Model size":"820M","Protocol":"5-shot","Pre-training tokens":"16.8B","Harness":"lm-evaluation-harness"}. This record does not assert independent reproduction.

### AdamW

ARC-c Accuracy: 33.45. Result ID: 69b9261747f34ce924a24ecc.

[Reporting paper](<https://www.sota2.com/research/paper/69b8e03d0636e5bf4afbed96.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa55761e209131ce9946db.md>)

Recorded evaluation configuration: {"Model size":"820M","Protocol":"5-shot","Pre-training tokens":"16.8B","Harness":"lm-evaluation-harness"}. This record does not assert independent reproduction.

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