# File-type classification on magic-files n=1,024 (held-out)

Source: [File-type classification on magic-files n=1,024 (held-out)](<https://www.sota2.com/research/sota/file-type-classification-on-magic-files-n-1-024-held-out>)

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

Data updated: 2026-06-04T11:03:42.813Z

Evaluation of file-type classification performance on the held-out 1,024 sample subset of the magic-files dataset, reporting strict, aligned, and top-level accuracies, along with the performance gap.

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 | File-type classification |
| Dataset (including any stated split) | magic-files n=1,024 (held-out) |
| Metric | Performance Gap (pp) |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| 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

- [Performance Gap (pp)](<https://www.sota2.com/research/sota/file-type-classification-on-magic-files-n-1-024-held-out.md?metric=Performance%20Gap%20(pp)>): Higher is better; unit: SCALAR.

- [Strict Top-1 Acc](<https://www.sota2.com/research/sota/file-type-classification-on-magic-files-n-1-024-held-out.md?metric=Strict%20Top-1%20Acc>): Higher is better; unit: PERCENT.

- [Aligned Top-1 Acc](<https://www.sota2.com/research/sota/file-type-classification-on-magic-files-n-1-024-held-out.md?metric=Aligned%20Top-1%20Acc>): Higher is better; unit: PERCENT.

- [Top-Level Top-1 Acc](<https://www.sota2.com/research/sota/file-type-classification-on-magic-files-n-1-024-held-out.md?metric=Top-Level%20Top-1%20Acc>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Performance Gap (pp) | Result date | Report paper |
| --- | --- | --- | --- |
| MimeLens | 17.5 | Not recorded | 6a2106cbc10484bac23f03a4 |
| MimeLens | 10.7 | Not recorded | 6a2106cbc10484bac23f03a4 |
| MimeLens | 8.7 | Not recorded | 6a2106cbc10484bac23f03a4 |

### MimeLens

Performance Gap (pp): 17.5. Result ID: 6a215b7cfbc373592c1df3a2.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106cbc10484bac23f03a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a2106cbc10484bac23f03a4.md>)

Recorded evaluation configuration: {"matching rule":"strict top-1","model\_variant":"medium","tokenizer":"bpe-16k"}. This record does not assert independent reproduction.

### MimeLens

Performance Gap (pp): 10.7. Result ID: 6a215b7cfbc373592c1df3a4.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106cbc10484bac23f03a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a2106cbc10484bac23f03a4.md>)

Recorded evaluation configuration: {"matching rule":"aligned top-1 (21-class equiv. map, symmetric)","model\_variant":"medium","tokenizer":"bpe-16k"}. This record does not assert independent reproduction.

### MimeLens

Performance Gap (pp): 8.7. Result ID: 6a215b7cfbc373592c1df3a6.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106cbc10484bac23f03a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a2106cbc10484bac23f03a4.md>)

Recorded evaluation configuration: {"matching rule":"top-level top-1 (text / image / application / . . . )","model\_variant":"medium","tokenizer":"bpe-16k"}. This record does not assert independent reproduction.

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