# Question Answering Calibration on TriviaQA (test)

Source: [Question Answering Calibration on TriviaQA (test)](<https://www.sota2.com/research/sota/question-answering-calibration-on-triviaqa-test>)

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

Data updated: 2026-02-26T05:28:10.787Z

Evaluation of model calibration performance on the test split of the TriviaQA dataset, utilizing specific metrics such as AUC, F1 thresholds, and ΔBow.

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 | Question Answering Calibration |
| Dataset (including any stated split) | TriviaQA (test) |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/question-answering-calibration-on-triviaqa-test.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [AUC](<https://www.sota2.com/research/sota/question-answering-calibration-on-triviaqa-test.md?metric=AUC>): Higher is better; unit: PERCENT.

- [F1@25](<https://www.sota2.com/research/sota/question-answering-calibration-on-triviaqa-test.md?metric=F1%4025>): Higher is better; unit: PERCENT.

- [F1@50](<https://www.sota2.com/research/sota/question-answering-calibration-on-triviaqa-test.md?metric=F1%4050>): Higher is better; unit: PERCENT.

- [F1@75](<https://www.sota2.com/research/sota/question-answering-calibration-on-triviaqa-test.md?metric=F1%4075>): Higher is better; unit: PERCENT.

- [ΔBow](<https://www.sota2.com/research/sota/question-answering-calibration-on-triviaqa-test.md?metric=%CE%94Bow>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| LIMECAL | 72 | Not recorded | 68fa6c011e209131cea319d0 |
| SHAPCAL | 71.8 | Not recorded | 68fa6c011e209131cea319d0 |
| BOWPROP | 71.2 | Not recorded | 68fa6c011e209131cea319d0 |
| KAMATH | 70.6 | Not recorded | 68fa6c011e209131cea319d0 |
| MAXPROB | 67 | Not recorded | 68fa6c011e209131cea319d0 |

### LIMECAL

Accuracy: 72. Result ID: 698eb9f2ffc3901de0a021d4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c011e209131cea319d0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c011e209131cea319d0.md>)

[Implementation](<https://github.com/xiye17/InterpCalib>)

Recorded evaluation configuration: {"Approach":"LIMECAL"}. This record does not assert independent reproduction.

### SHAPCAL

Accuracy: 71.8. Result ID: 698eb9f2ffc3901de0a021d5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c011e209131cea319d0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c011e209131cea319d0.md>)

[Implementation](<https://github.com/xiye17/InterpCalib>)

Recorded evaluation configuration: {"Approach":"SHAPCAL"}. This record does not assert independent reproduction.

### BOWPROP

Accuracy: 71.2. Result ID: 698eb9f2ffc3901de0a021d3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c011e209131cea319d0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c011e209131cea319d0.md>)

[Implementation](<https://github.com/xiye17/InterpCalib>)

Recorded evaluation configuration: {"Approach":"BOWPROP"}. This record does not assert independent reproduction.

### KAMATH

Accuracy: 70.6. Result ID: 698eb9f2ffc3901de0a021d2.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c011e209131cea319d0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67b61e209131cea17ff7.md>)

Recorded evaluation configuration: {"Approach":"KAMATH"}. This record does not assert independent reproduction.

### MAXPROB

Accuracy: 67. Result ID: 698eb9f2ffc3901de0a021d1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6c011e209131cea319d0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa638b1e209131ce9ff6d1.md>)

Recorded evaluation configuration: {"Approach":"MAXPROB"}. This record does not assert independent reproduction.

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