# Attribution Quality Evaluation on TITAN autoregressive t+5 (test)

Source: [Attribution Quality Evaluation on TITAN autoregressive t+5 (test)](<https://www.sota2.com/research/sota/attribution-quality-evaluation-on-titan-autoregressive-t-plus-5-test>)

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

Data updated: 2026-04-27T07:39:50.792Z

Evaluation of attribution quality on the test split of the TITAN autoregressive t+5 dataset, measured using metrics such as Sparsity (Gini), Faithfulness (ROAD), and LLE variants.

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 | Attribution Quality Evaluation |
| Dataset (including any stated split) | TITAN autoregressive t+5 (test) |
| Metric | Sparsity (Gini) |
| 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

- [Sparsity (Gini)](<https://www.sota2.com/research/sota/attribution-quality-evaluation-on-titan-autoregressive-t-plus-5-test.md?metric=Sparsity%20(Gini)>): Higher is better; unit: SCALAR.

- [Faithfulness (ROAD)](<https://www.sota2.com/research/sota/attribution-quality-evaluation-on-titan-autoregressive-t-plus-5-test.md?metric=Faithfulness%20(ROAD)>): Higher is better; unit: SCALAR.

- [LLE (L2)](<https://www.sota2.com/research/sota/attribution-quality-evaluation-on-titan-autoregressive-t-plus-5-test.md?metric=LLE%20(L2)>): Lower is better; unit: SCALAR.

- [LLE (Cosine)](<https://www.sota2.com/research/sota/attribution-quality-evaluation-on-titan-autoregressive-t-plus-5-test.md?metric=LLE%20(Cosine)>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Sparsity (Gini) | Result date | Report paper |
| --- | --- | --- | --- |
| WGBary×Grad | 0.75 | Not recorded | 69eeed6955a3c32af89f964c |
| VarGrad | 0.71 | Not recorded | 69eeed6955a3c32af89f964c |
| IntegratedGrad | 0.632 | Not recorded | 69eeed6955a3c32af89f964c |
| BaseGrad | 0.531 | Not recorded | 69eeed6955a3c32af89f964c |
| SmoothGrad | 0.526 | Not recorded | 69eeed6955a3c32af89f964c |
| WGBary | 0.33 | Not recorded | 69eeed6955a3c32af89f964c |

### WGBary×Grad

Sparsity (Gini): 0.75. Result ID: 69ef10fe47f34ce924d4eb1a.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed6955a3c32af89f964c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eeed6955a3c32af89f964c.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### VarGrad

Sparsity (Gini): 0.71. Result ID: 69ef10fe47f34ce924d4eb18.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed6955a3c32af89f964c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64e11e209131cea0707c.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### IntegratedGrad

Sparsity (Gini): 0.632. Result ID: 69ef10fe47f34ce924d4eb16.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed6955a3c32af89f964c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63661e209131ce9fe6bd.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### BaseGrad

Sparsity (Gini): 0.531. Result ID: 69ef10fe47f34ce924d4eb15.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed6955a3c32af89f964c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa614c1e209131ce9f38a9.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### SmoothGrad

Sparsity (Gini): 0.526. Result ID: 69ef10fe47f34ce924d4eb17.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed6955a3c32af89f964c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63661e209131ce9fe6d2.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### WGBary

Sparsity (Gini): 0.33. Result ID: 69ef10fe47f34ce924d4eb19.

[Reporting paper](<https://www.sota2.com/research/paper/69eeed6955a3c32af89f964c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eeed6955a3c32af89f964c.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

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