# Explanation Attribution on GPT-2 output-preserving n=30, k=2, overlap=0.7 (test)

Source: [Explanation Attribution on GPT-2 output-preserving n=30, k=2, overlap=0.7 (test)](<https://www.sota2.com/research/sota/explanation-attribution-on-gpt-2-output-preserving-n-30-k-2-overlap-0-7-test>)

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

Data updated: 2026-02-26T05:21:40.947Z

Evaluation of Explanation Attribution methods using specific output-preserving configurations (n=30, k=2, overlap=0.7) derived from GPT-2 outputs, measured by Faithfulness Gap and Overlap SplitIndex.

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 | Explanation Attribution |
| Dataset (including any stated split) | GPT-2 output-preserving n=30, k=2, overlap=0.7 (test) |
| Metric | Faithfulness Gap |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 0 of 0 |
| 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

- [Faithfulness Gap](<https://www.sota2.com/research/sota/explanation-attribution-on-gpt-2-output-preserving-n-30-k-2-overlap-0-7-test.md?metric=Faithfulness%20Gap>): Lower is better; unit: SCALAR.

- [Overlap SplitIndex](<https://www.sota2.com/research/sota/explanation-attribution-on-gpt-2-output-preserving-n-30-k-2-overlap-0-7-test.md?metric=Overlap%20SplitIndex>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Faithfulness Gap | Result date | Report paper |
| --- | --- | --- | --- |

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