# Role-Playing Ability on Role-playing

Source: [Role-Playing Ability on Role-playing](<https://www.sota2.com/research/sota/role-playing-ability-on-role-playing>)

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

Data updated: 2026-05-19T12:42:56.782Z

Evaluation of Role-Playing Ability using custom metrics (CUS, RAW, SPE) on the 'Role-playing' dataset.

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 | Role-Playing Ability |
| Dataset (including any stated split) | Role-playing |
| Metric | CUS Score |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 12 of 12 |
| 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

- [CUS Score](<https://www.sota2.com/research/sota/role-playing-ability-on-role-playing.md?metric=CUS%20Score>): Higher is better; unit: SCALAR.

- [RAW Score](<https://www.sota2.com/research/sota/role-playing-ability-on-role-playing.md?metric=RAW%20Score>): Higher is better; unit: SCALAR.

- [SPE Score](<https://www.sota2.com/research/sota/role-playing-ability-on-role-playing.md?metric=SPE%20Score>): Higher is better; unit: SCALAR.

- [Average Score](<https://www.sota2.com/research/sota/role-playing-ability-on-role-playing.md?metric=Average%20Score>): Higher is better; unit: SCALAR.

## Recorded results

| Method | CUS Score | Result date | Report paper |
| --- | --- | --- | --- |
| PPR-GDE | 4.99 | Not recorded | 6a0bef08dbc836b87d82c9ae |
| PPR-GDE | 4.97 | Not recorded | 6a0bef08dbc836b87d82c9ae |
| PPR-GDE | 4.94 | Not recorded | 6a0bef08dbc836b87d82c9ae |
| GRPO | 4.88 | Not recorded | 6a0bef08dbc836b87d82c9ae |
| PPO | 4.83 | Not recorded | 6a0bef08dbc836b87d82c9ae |
| GRPO | 4.81 | Not recorded | 6a0bef08dbc836b87d82c9ae |
| GRPO | 4.8 | Not recorded | 6a0bef08dbc836b87d82c9ae |
| PPO | 4.77 | Not recorded | 6a0bef08dbc836b87d82c9ae |
| PPO | 4.71 | Not recorded | 6a0bef08dbc836b87d82c9ae |
| w/o RL | 4.04 | Not recorded | 6a0bef08dbc836b87d82c9ae |
| w/o RL | 3.38 | Not recorded | 6a0bef08dbc836b87d82c9ae |
| w/o RL | 3.21 | Not recorded | 6a0bef08dbc836b87d82c9ae |

### PPR-GDE

CUS Score: 4.99. Result ID: 6a0c5aaefbc373592c03e1bc.

[Reporting paper](<https://www.sota2.com/research/paper/6a0bef08dbc836b87d82c9ae.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0bef08dbc836b87d82c9ae.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-3B-Instruct"}. This record does not assert independent reproduction.

### PPR-GDE

CUS Score: 4.97. Result ID: 6a0c5aaefbc373592c03e1c4.

[Reporting paper](<https://www.sota2.com/research/paper/6a0bef08dbc836b87d82c9ae.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0bef08dbc836b87d82c9ae.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-0.6B"}. This record does not assert independent reproduction.

### PPR-GDE

CUS Score: 4.94. Result ID: 6a0c5aaefbc373592c03e1c0.

[Reporting paper](<https://www.sota2.com/research/paper/6a0bef08dbc836b87d82c9ae.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0bef08dbc836b87d82c9ae.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-1.5B-Instruct"}. This record does not assert independent reproduction.

### GRPO

CUS Score: 4.88. Result ID: 6a0c5aaefbc373592c03e1bb.

[Reporting paper](<https://www.sota2.com/research/paper/6a0bef08dbc836b87d82c9ae.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73101e209131cea5a395.md>)

[Implementation](<https://github.com/deepseek-ai/DeepSeek-Math>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-3B-Instruct"}. This record does not assert independent reproduction.

### PPO

CUS Score: 4.83. Result ID: 6a0c5aaefbc373592c03e1c2.

[Reporting paper](<https://www.sota2.com/research/paper/6a0bef08dbc836b87d82c9ae.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63a01e209131ce9ffdaa.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-0.6B"}. This record does not assert independent reproduction.

### GRPO

CUS Score: 4.81. Result ID: 6a0c5aaefbc373592c03e1bf.

[Reporting paper](<https://www.sota2.com/research/paper/6a0bef08dbc836b87d82c9ae.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73101e209131cea5a395.md>)

[Implementation](<https://github.com/deepseek-ai/DeepSeek-Math>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-1.5B-Instruct"}. This record does not assert independent reproduction.

### GRPO

CUS Score: 4.8. Result ID: 6a0c5aaefbc373592c03e1c3.

[Reporting paper](<https://www.sota2.com/research/paper/6a0bef08dbc836b87d82c9ae.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73101e209131cea5a395.md>)

[Implementation](<https://github.com/deepseek-ai/DeepSeek-Math>)

Recorded evaluation configuration: {"Backbone":"Qwen3-0.6B"}. This record does not assert independent reproduction.

### PPO

CUS Score: 4.77. Result ID: 6a0c5aaefbc373592c03e1be.

[Reporting paper](<https://www.sota2.com/research/paper/6a0bef08dbc836b87d82c9ae.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63a01e209131ce9ffdaa.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-1.5B-Instruct"}. This record does not assert independent reproduction.

### PPO

CUS Score: 4.71. Result ID: 6a0c5aaefbc373592c03e1ba.

[Reporting paper](<https://www.sota2.com/research/paper/6a0bef08dbc836b87d82c9ae.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63a01e209131ce9ffdaa.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-3B-Instruct"}. This record does not assert independent reproduction.

### w/o RL

CUS Score: 4.04. Result ID: 6a0c5aaefbc373592c03e1b9.

[Reporting paper](<https://www.sota2.com/research/paper/6a0bef08dbc836b87d82c9ae.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-3B-Instruct"}. This record does not assert independent reproduction.

### w/o RL

CUS Score: 3.38. Result ID: 6a0c5aaefbc373592c03e1c1.

[Reporting paper](<https://www.sota2.com/research/paper/6a0bef08dbc836b87d82c9ae.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-0.6B"}. This record does not assert independent reproduction.

### w/o RL

CUS Score: 3.21. Result ID: 6a0c5aaefbc373592c03e1bd.

[Reporting paper](<https://www.sota2.com/research/paper/6a0bef08dbc836b87d82c9ae.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-1.5B-Instruct"}. This record does not assert independent reproduction.

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