# Language Generation on Synthetic Data (test)

Source: [Language Generation on Synthetic Data (test)](<https://www.sota2.com/research/sota/language-generation-on-synthetic-data-test>)

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

Data updated: 2026-03-12T07:13:28.925Z

Evaluation of language generation models on the test split of the Synthetic Data dataset, assessed using ROUGE metrics and Cosine Similarity.

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 | Language Generation |
| Dataset (including any stated split) | Synthetic Data (test) |
| Metric | ROUGE-1 |
| Unit | PERCENT |
| 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

- [ROUGE-1](<https://www.sota2.com/research/sota/language-generation-on-synthetic-data-test.md?metric=ROUGE-1>): Higher is better; unit: PERCENT.

- [ROUGE-2](<https://www.sota2.com/research/sota/language-generation-on-synthetic-data-test.md?metric=ROUGE-2>): Higher is better; unit: PERCENT.

- [ROUGE-L](<https://www.sota2.com/research/sota/language-generation-on-synthetic-data-test.md?metric=ROUGE-L>): Higher is better; unit: PERCENT.

- [Cosine Similarity](<https://www.sota2.com/research/sota/language-generation-on-synthetic-data-test.md?metric=Cosine%20Similarity>): Higher is better; unit: SCALAR.

## Recorded results

| Method | ROUGE-1 | Result date | Report paper |
| --- | --- | --- | --- |
| P-GRPO | 62.67 | Not recorded | 69b24862d030084c22e2f706 |
| P-GRPO | 62.67 | Not recorded | 69b24862d030084c22e2f706 |
| P-GRPO | 61.33 | Not recorded | 69b24862d030084c22e2f706 |
| P-GRPO | 61.33 | Not recorded | 69b24862d030084c22e2f706 |
| GRPO | 60.75 | Not recorded | 69b24862d030084c22e2f706 |
| GRPO | 60.75 | Not recorded | 69b24862d030084c22e2f706 |
| GRPO | 60.42 | Not recorded | 69b24862d030084c22e2f706 |
| GRPO | 60.42 | Not recorded | 69b24862d030084c22e2f706 |
| GDPO | 56.63 | Not recorded | 69b24862d030084c22e2f706 |
| GDPO | 56.63 | Not recorded | 69b24862d030084c22e2f706 |
| GDPO | 52.97 | Not recorded | 69b24862d030084c22e2f706 |
| GDPO | 52.97 | Not recorded | 69b24862d030084c22e2f706 |

### P-GRPO

ROUGE-1: 62.67. Result ID: 69b265c06dbb7075b95c80dc.

[Reporting paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

Recorded evaluation configuration: {"Model":"Qwen3-8B"}. This record does not assert independent reproduction.

### P-GRPO

ROUGE-1: 62.67. Result ID: 69b265c06dbb7075b95c80eb.

[Reporting paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-8B"}. This record does not assert independent reproduction.

### P-GRPO

ROUGE-1: 61.33. Result ID: 69b265c06dbb7075b95c80d9.

[Reporting paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

Recorded evaluation configuration: {"Model":"Gemma-2B"}. This record does not assert independent reproduction.

### P-GRPO

ROUGE-1: 61.33. Result ID: 69b265c06dbb7075b95c80ee.

[Reporting paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

Recorded evaluation configuration: {"Backbone":"Gemma-2B"}. This record does not assert independent reproduction.

### GRPO

ROUGE-1: 60.75. Result ID: 69b265c06dbb7075b95c80db.

[Reporting paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

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

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

Recorded evaluation configuration: {"Model":"Qwen3-8B"}. This record does not assert independent reproduction.

### GRPO

ROUGE-1: 60.75. Result ID: 69b265c06dbb7075b95c80ea.

[Reporting paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

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

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

Recorded evaluation configuration: {"Backbone":"Qwen3-8B"}. This record does not assert independent reproduction.

### GRPO

ROUGE-1: 60.42. Result ID: 69b265c06dbb7075b95c80d8.

[Reporting paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

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

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

Recorded evaluation configuration: {"Model":"Gemma-2B"}. This record does not assert independent reproduction.

### GRPO

ROUGE-1: 60.42. Result ID: 69b265c06dbb7075b95c80ed.

[Reporting paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

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

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

Recorded evaluation configuration: {"Backbone":"Gemma-2B"}. This record does not assert independent reproduction.

### GDPO

ROUGE-1: 56.63. Result ID: 69b265c06dbb7075b95c80da.

[Reporting paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78841e209131cea78c7e.md>)

Recorded evaluation configuration: {"Model":"Qwen3-8B"}. This record does not assert independent reproduction.

### GDPO

ROUGE-1: 56.63. Result ID: 69b265c06dbb7075b95c80e9.

[Reporting paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78841e209131cea78c7e.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-8B"}. This record does not assert independent reproduction.

### GDPO

ROUGE-1: 52.97. Result ID: 69b265c06dbb7075b95c80d7.

[Reporting paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78841e209131cea78c7e.md>)

Recorded evaluation configuration: {"Model":"Gemma-2B"}. This record does not assert independent reproduction.

### GDPO

ROUGE-1: 52.97. Result ID: 69b265c06dbb7075b95c80ec.

[Reporting paper](<https://www.sota2.com/research/paper/69b24862d030084c22e2f706.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78841e209131cea78c7e.md>)

Recorded evaluation configuration: {"Backbone":"Gemma-2B"}. This record does not assert independent reproduction.

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