# Embodied Decision-Grounded QA on Microsoft PdM

Source: [Embodied Decision-Grounded QA on Microsoft PdM](<https://www.sota2.com/research/sota/embodied-decision-grounded-qa-on-microsoft-pdm>)

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

Data updated: 2026-04-28T08:30:44.242Z

Evaluation of model performance on Embodied Decision-Grounded Question Answering using the Microsoft PdM dataset, focusing on structural correctness, provenance, and claim accuracy.

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 | Embodied Decision-Grounded QA |
| Dataset (including any stated split) | Microsoft PdM |
| Metric | Structural OK |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 10 of 10 |
| 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

- [Structural OK](<https://www.sota2.com/research/sota/embodied-decision-grounded-qa-on-microsoft-pdm.md?metric=Structural%20OK>): Higher is better; unit: PERCENT.

- [Provenance OK](<https://www.sota2.com/research/sota/embodied-decision-grounded-qa-on-microsoft-pdm.md?metric=Provenance%20OK>): Higher is better; unit: PERCENT.

- [Label Consistency](<https://www.sota2.com/research/sota/embodied-decision-grounded-qa-on-microsoft-pdm.md?metric=Label%20Consistency>): Higher is better; unit: PERCENT.

- [CF Accuracy](<https://www.sota2.com/research/sota/embodied-decision-grounded-qa-on-microsoft-pdm.md?metric=CF%20Accuracy>): Higher is better; unit: PERCENT.

- [Entailment Pass Rate](<https://www.sota2.com/research/sota/embodied-decision-grounded-qa-on-microsoft-pdm.md?metric=Entailment%20Pass%20Rate>): Higher is better; unit: PERCENT.

- [Claim Precision](<https://www.sota2.com/research/sota/embodied-decision-grounded-qa-on-microsoft-pdm.md?metric=Claim%20Precision>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Structural OK | Result date | Report paper |
| --- | --- | --- | --- |
| Full IndustryAssetEQA | 90 | Not recorded | 69f03f4355a3c32af89fa22f |
| Full IndustryAssetEQA | 88 | Not recorded | 69f03f4355a3c32af89fa22f |
| + Provenance-Enforced | 82 | Not recorded | 69f03f4355a3c32af89fa22f |
| + Provenance-Enforced | 80 | Not recorded | 69f03f4355a3c32af89fa22f |
| LLM + Episodic + KG | 55 | Not recorded | 69f03f4355a3c32af89fa22f |
| LLM + Episodic | 54 | Not recorded | 69f03f4355a3c32af89fa22f |
| LLM + Episodic | 52 | Not recorded | 69f03f4355a3c32af89fa22f |
| LLM + Episodic + KG | 52 | Not recorded | 69f03f4355a3c32af89fa22f |
| LLM-only QA | 42 | Not recorded | 69f03f4355a3c32af89fa22f |
| LLM-only QA | 39 | Not recorded | 69f03f4355a3c32af89fa22f |

### Full IndustryAssetEQA

Structural OK: 90. Result ID: 69f06fe447f34ce924d654ad.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f4355a3c32af89fa22f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f4355a3c32af89fa22f.md>)

Recorded evaluation configuration: {"Model":"Claude Sonnet 4"}. This record does not assert independent reproduction.

### Full IndustryAssetEQA

Structural OK: 88. Result ID: 69f06fe447f34ce924d654a8.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f4355a3c32af89fa22f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f4355a3c32af89fa22f.md>)

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

### + Provenance-Enforced

Structural OK: 82. Result ID: 69f06fe447f34ce924d654a7.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f4355a3c32af89fa22f.md>)

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

### + Provenance-Enforced

Structural OK: 80. Result ID: 69f06fe447f34ce924d654ac.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f4355a3c32af89fa22f.md>)

Recorded evaluation configuration: {"Model":"Claude Sonnet 4"}. This record does not assert independent reproduction.

### LLM + Episodic + KG

Structural OK: 55. Result ID: 69f06fe447f34ce924d654ab.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f4355a3c32af89fa22f.md>)

Recorded evaluation configuration: {"Model":"Claude Sonnet 4"}. This record does not assert independent reproduction.

### LLM + Episodic

Structural OK: 54. Result ID: 69f06fe447f34ce924d654aa.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f4355a3c32af89fa22f.md>)

Recorded evaluation configuration: {"Model":"Claude Sonnet 4"}. This record does not assert independent reproduction.

### LLM + Episodic

Structural OK: 52. Result ID: 69f06fe447f34ce924d654a5.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f4355a3c32af89fa22f.md>)

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

### LLM + Episodic + KG

Structural OK: 52. Result ID: 69f06fe447f34ce924d654a6.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f4355a3c32af89fa22f.md>)

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

### LLM-only QA

Structural OK: 42. Result ID: 69f06fe447f34ce924d654a4.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f4355a3c32af89fa22f.md>)

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

### LLM-only QA

Structural OK: 39. Result ID: 69f06fe447f34ce924d654a9.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f4355a3c32af89fa22f.md>)

Recorded evaluation configuration: {"Model":"Claude Sonnet 4"}. This record does not assert independent reproduction.

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