# Logical Refinement of Natural Language Explanations on QASC

Source: [Logical Refinement of Natural Language Explanations on QASC](<https://www.sota2.com/research/sota/logical-refinement-of-natural-language-explanations-on-qasc>)

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

Data updated: 2026-02-26T05:47:13.145Z

Evaluation of model performance on refining natural language explanations using the QASC dataset, focusing on the quality improvement achieved through iterative refinement steps.

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 | Logical Refinement of Natural Language Explanations |
| Dataset (including any stated split) | QASC |
| Metric | Initial Score |
| Unit | PERCENT |
| 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

- [Initial Score](<https://www.sota2.com/research/sota/logical-refinement-of-natural-language-explanations-on-qasc.md?metric=Initial%20Score>): Higher is better; unit: PERCENT.

- [Final Score](<https://www.sota2.com/research/sota/logical-refinement-of-natural-language-explanations-on-qasc.md?metric=Final%20Score>): Higher is better; unit: PERCENT.

- [Number of Iterations](<https://www.sota2.com/research/sota/logical-refinement-of-natural-language-explanations-on-qasc.md?metric=Number%20of%20Iterations>): Lower is better; unit: SCALAR.

- [Number of Calls](<https://www.sota2.com/research/sota/logical-refinement-of-natural-language-explanations-on-qasc.md?metric=Number%20of%20Calls>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Initial Score | Result date | Report paper |
| --- | --- | --- | --- |
| Faithful-Refiner | 17 | Not recorded | 68fa78e71e209131cea7b1e8 |
| Faithful-Refiner | 12 | Not recorded | 68fa78e71e209131cea7b1e8 |
| Faithful-Refiner | 11 | Not recorded | 68fa78e71e209131cea7b1e8 |
| Faithful-Refiner | 10 | Not recorded | 68fa78e71e209131cea7b1e8 |
| Explanation-Refiner | 4 | Not recorded | 68fa78e71e209131cea7b1e8 |
| Explanation-Refiner | 4 | Not recorded | 68fa78e71e209131cea7b1e8 |
| Explanation-Refiner | 4 | Not recorded | 68fa78e71e209131cea7b1e8 |
| Explanation-Refiner | 3 | Not recorded | 68fa78e71e209131cea7b1e8 |

### Faithful-Refiner

Initial Score: 17. Result ID: 6995ac7a3badfaffe6ab2865.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78e71e209131cea7b1e8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78e71e209131cea7b1e8.md>)

[Implementation](<https://github.com/neurosymbolicai/faithful_and_robust_nli_refinement>)

Recorded evaluation configuration: {"LLM":"Deepseek-V3"}. This record does not assert independent reproduction.

### Faithful-Refiner

Initial Score: 12. Result ID: 6995ac7a3badfaffe6ab2863.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78e71e209131cea7b1e8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78e71e209131cea7b1e8.md>)

[Implementation](<https://github.com/neurosymbolicai/faithful_and_robust_nli_refinement>)

Recorded evaluation configuration: {"LLM":"GPT-4o-mini"}. This record does not assert independent reproduction.

### Faithful-Refiner

Initial Score: 11. Result ID: 6995ac7a3badfaffe6ab2862.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78e71e209131cea7b1e8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78e71e209131cea7b1e8.md>)

[Implementation](<https://github.com/neurosymbolicai/faithful_and_robust_nli_refinement>)

Recorded evaluation configuration: {"LLM":"Llama3.1-70b"}. This record does not assert independent reproduction.

### Faithful-Refiner

Initial Score: 10. Result ID: 6995ac7a3badfaffe6ab2864.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78e71e209131cea7b1e8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78e71e209131cea7b1e8.md>)

[Implementation](<https://github.com/neurosymbolicai/faithful_and_robust_nli_refinement>)

Recorded evaluation configuration: {"LLM":"GPT-4o"}. This record does not assert independent reproduction.

### Explanation-Refiner

Initial Score: 4. Result ID: 6995ac7a3badfaffe6ab285e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78e71e209131cea7b1e8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa752f1e209131cea66324.md>)

Recorded evaluation configuration: {"LLM":"Llama3.1-70b"}. This record does not assert independent reproduction.

### Explanation-Refiner

Initial Score: 4. Result ID: 6995ac7a3badfaffe6ab2860.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78e71e209131cea7b1e8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa752f1e209131cea66324.md>)

Recorded evaluation configuration: {"LLM":"GPT-4o"}. This record does not assert independent reproduction.

### Explanation-Refiner

Initial Score: 4. Result ID: 6995ac7a3badfaffe6ab2861.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78e71e209131cea7b1e8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa752f1e209131cea66324.md>)

Recorded evaluation configuration: {"LLM":"Deepseek-V3"}. This record does not assert independent reproduction.

### Explanation-Refiner

Initial Score: 3. Result ID: 6995ac7a3badfaffe6ab285f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78e71e209131cea7b1e8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa752f1e209131cea66324.md>)

Recorded evaluation configuration: {"LLM":"GPT-4o-mini"}. This record does not assert independent reproduction.

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