# Question-Type Diversity Alignment on Metacog Taxonomy

Source: [Question-Type Diversity Alignment on Metacog Taxonomy](<https://www.sota2.com/research/sota/question-type-diversity-alignment-on-metacog-taxonomy>)

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

Data updated: 2026-03-06T07:48:31.477Z

Evaluation of model performance on the Metacog Taxonomy dataset, specifically assessing alignment based on Question-Type Diversity, measured using Jensen Shannon Divergence.

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-Type Diversity Alignment |
| Dataset (including any stated split) | Metacog Taxonomy |
| Metric | Jensen Shannon Divergence |
| Unit | SCALAR |
| Direction | Lower 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

- [Jensen Shannon Divergence](<https://www.sota2.com/research/sota/question-type-diversity-alignment-on-metacog-taxonomy.md?metric=Jensen%20Shannon%20Divergence>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Jensen Shannon Divergence | Result date | Report paper |
| --- | --- | --- | --- |
| gemini-2.5-pro | 0.059 | Not recorded | 69aa5f672d603e6de3e4172b |
| gemini-2.5-pro | 0.072 | Not recorded | 69aa5f672d603e6de3e4172b |
| Llama-3.3-70B-Instruct | 0.08 | Not recorded | 69aa5f672d603e6de3e4172b |
| gpt4o | 0.084 | Not recorded | 69aa5f672d603e6de3e4172b |
| gpt4o | 0.089 | Not recorded | 69aa5f672d603e6de3e4172b |
| gpt-oss-120b | 0.129 | Not recorded | 69aa5f672d603e6de3e4172b |
| Qwen3-32B | 0.149 | Not recorded | 69aa5f672d603e6de3e4172b |
| gpt-oss-120b | 0.18 | Not recorded | 69aa5f672d603e6de3e4172b |

### gemini-2.5-pro

Jensen Shannon Divergence: 0.059. Result ID: 69aa86296dbb7075b9540553.

[Reporting paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

Recorded evaluation configuration: {"Simulator Variant":"AGENT"}. This record does not assert independent reproduction.

### gemini-2.5-pro

Jensen Shannon Divergence: 0.072. Result ID: 69aa86296dbb7075b9540554.

[Reporting paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

Recorded evaluation configuration: {"Simulator Variant":"SCOTUS\_DEFAULT"}. This record does not assert independent reproduction.

### Llama-3.3-70B-Instruct

Jensen Shannon Divergence: 0.08. Result ID: 69aa86296dbb7075b9540551.

[Reporting paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

Recorded evaluation configuration: {"Simulator Variant":"SCOTUS\_DEFAULT"}. This record does not assert independent reproduction.

### gpt4o

Jensen Shannon Divergence: 0.084. Result ID: 69aa86296dbb7075b9540557.

[Reporting paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

Recorded evaluation configuration: {"Simulator Variant":"AGENT"}. This record does not assert independent reproduction.

### gpt4o

Jensen Shannon Divergence: 0.089. Result ID: 69aa86296dbb7075b9540558.

[Reporting paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

Recorded evaluation configuration: {"Simulator Variant":"SCOTUS\_DEFAULT"}. This record does not assert independent reproduction.

### gpt-oss-120b

Jensen Shannon Divergence: 0.129. Result ID: 69aa86296dbb7075b9540555.

[Reporting paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

Recorded evaluation configuration: {"Simulator Variant":"AGENT"}. This record does not assert independent reproduction.

### Qwen3-32B

Jensen Shannon Divergence: 0.149. Result ID: 69aa86296dbb7075b9540552.

[Reporting paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

Recorded evaluation configuration: {"Simulator Variant":"SCOTUS\_DEFAULT"}. This record does not assert independent reproduction.

### gpt-oss-120b

Jensen Shannon Divergence: 0.18. Result ID: 69aa86296dbb7075b9540556.

[Reporting paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69aa5f672d603e6de3e4172b.md>)

Recorded evaluation configuration: {"Simulator Variant":"SCOTUS\_DEFAULT"}. This record does not assert independent reproduction.

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