# Parts Mental Model Accuracy on ParRoT++

Source: [Parts Mental Model Accuracy on ParRoT++](<https://www.sota2.com/research/sota/parts-mental-model-accuracy-on-parrot-plus-plus>)

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

Data updated: 2026-02-26T05:37:39.266Z

Evaluation of a model's accuracy in understanding and predicting parts based on the ParRoT++ dataset, focusing on mental model performance metrics.

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 | Parts Mental Model Accuracy |
| Dataset (including any stated split) | ParRoT++ |
| Metric | Base LM Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 2 of 2 |
| 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

- [Base LM Accuracy](<https://www.sota2.com/research/sota/parts-mental-model-accuracy-on-parrot-plus-plus.md?metric=Base%20LM%20Accuracy>): Higher is better; unit: PERCENT.

- [ParRoT-Con Accuracy](<https://www.sota2.com/research/sota/parts-mental-model-accuracy-on-parrot-plus-plus.md?metric=ParRoT-Con%20Accuracy>): Higher is better; unit: PERCENT.

- [Improvement (%)](<https://www.sota2.com/research/sota/parts-mental-model-accuracy-on-parrot-plus-plus.md?metric=Improvement%20(%25)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Base LM Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Macaw-11B | 60.04 | Not recorded | 68fa6fbc1e209131cea471b4 |
| GPT-3 (text-davinci-003) | 55.51 | Not recorded | 68fa6fbc1e209131cea471b4 |

### Macaw-11B

Base LM Accuracy: 60.04. Result ID: 6990fc9a3a6c31484904cf7f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6fbc1e209131cea471b4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ab91e209131cea2a13d.md>)

Recorded evaluation configuration: {"# params":"11B"}. This record does not assert independent reproduction.

### GPT-3 (text-davinci-003)

Base LM Accuracy: 55.51. Result ID: 6990fc9a3a6c31484904cf7e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6fbc1e209131cea471b4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa680c1e209131cea19d53.md>)

[Implementation](<https://github.com/openai/gpt-3/blob/master/overlap_frequency.md>)

Recorded evaluation configuration: {"# params":"175B"}. This record does not assert independent reproduction.

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