# Constrained Text Generation on CommonGen 500 randomly sampled data (test)

Source: [Constrained Text Generation on CommonGen 500 randomly sampled data (test)](<https://www.sota2.com/research/sota/constrained-text-generation-on-commongen-500-randomly-sampled-data-test>)

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

Data updated: 2026-02-26T05:37:18.225Z

Evaluation of models performing Constrained Text Generation specifically on a 500-sample subset of the CommonGen test set, measured using BLEU-4, CIDEr, and SPICE scores.

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 | Constrained Text Generation |
| Dataset (including any stated split) | CommonGen 500 randomly sampled data (test) |
| Metric | BLEU-4 |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 9 of 9 |
| 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

- [BLEU-4](<https://www.sota2.com/research/sota/constrained-text-generation-on-commongen-500-randomly-sampled-data-test.md?metric=BLEU-4>): Higher is better; unit: PERCENT.

- [CIDEr](<https://www.sota2.com/research/sota/constrained-text-generation-on-commongen-500-randomly-sampled-data-test.md?metric=CIDEr>): Higher is better; unit: PERCENT.

- [SPICE](<https://www.sota2.com/research/sota/constrained-text-generation-on-commongen-500-randomly-sampled-data-test.md?metric=SPICE>): Higher is better; unit: PERCENT.

## Recorded results

| Method | BLEU-4 | Result date | Report paper |
| --- | --- | --- | --- |
| MORE | 32.53 | Not recorded | 68fa73a51e209131cea5da67 |
| MORE | 31.81 | Not recorded | 68fa73a51e209131cea5da67 |
| GPT-4 | 30 | Not recorded | 68fa73a51e209131cea5da67 |
| GPT-4 | 28.53 | Not recorded | 68fa73a51e209131cea5da67 |
| GPT-3.5 | 28.35 | Not recorded | 68fa73a51e209131cea5da67 |
| GPT-4 | 27.87 | Not recorded | 68fa73a51e209131cea5da67 |
| GPT-3.5 | 25.54 | Not recorded | 68fa73a51e209131cea5da67 |
| GPT-3.5 | 22.97 | Not recorded | 68fa73a51e209131cea5da67 |
| GPT-4 | 19.86 | Not recorded | 68fa73a51e209131cea5da67 |

### MORE

BLEU-4: 32.53. Result ID: 6992fc5b3a6c31484912dec1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73a51e209131cea5da67.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73a51e209131cea5da67.md>)

Implementation

Recorded evaluation configuration: {"backbone":"OPT-2.7b","retrieval augmentation":"3 images, 3 texts"}. This record does not assert independent reproduction.

### MORE

BLEU-4: 31.81. Result ID: 6992fc5b3a6c31484912dec0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73a51e209131cea5da67.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73a51e209131cea5da67.md>)

Implementation

Recorded evaluation configuration: {"backbone":"OPT-2.7b","retrieval augmentation":"1 image, 1 text"}. This record does not assert independent reproduction.

### GPT-4

BLEU-4: 30. Result ID: 6992fc5b3a6c31484912debb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73a51e209131cea5da67.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa727a1e209131cea56a1e.md>)

[Implementation](<https://github.com/openai/evals>)

Recorded evaluation configuration: {"n-shot":"3-shot"}. This record does not assert independent reproduction.

### GPT-4

BLEU-4: 28.53. Result ID: 6992fc5b3a6c31484912deb9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73a51e209131cea5da67.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa727a1e209131cea56a1e.md>)

[Implementation](<https://github.com/openai/evals>)

Recorded evaluation configuration: {"n-shot":"0-shot"}. This record does not assert independent reproduction.

### GPT-3.5

BLEU-4: 28.35. Result ID: 6992fc5b3a6c31484912debf.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73a51e209131cea5da67.md>)

Recorded evaluation configuration: {"n-shot":"3-shot"}. This record does not assert independent reproduction.

### GPT-4

BLEU-4: 27.87. Result ID: 6992fc5b3a6c31484912deba.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73a51e209131cea5da67.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa727a1e209131cea56a1e.md>)

[Implementation](<https://github.com/openai/evals>)

Recorded evaluation configuration: {"n-shot":"0-shot","length constraint":true}. This record does not assert independent reproduction.

### GPT-3.5

BLEU-4: 25.54. Result ID: 6992fc5b3a6c31484912debe.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73a51e209131cea5da67.md>)

Recorded evaluation configuration: {"n-shot":"0-shot","length constraint":true}. This record does not assert independent reproduction.

### GPT-3.5

BLEU-4: 22.97. Result ID: 6992fc5b3a6c31484912debd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73a51e209131cea5da67.md>)

Recorded evaluation configuration: {"n-shot":"0-shot"}. This record does not assert independent reproduction.

### GPT-4

BLEU-4: 19.86. Result ID: 6992fc5b3a6c31484912debc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73a51e209131cea5da67.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa727a1e209131cea56a1e.md>)

[Implementation](<https://github.com/openai/evals>)

Recorded evaluation configuration: {"n-shot":"0-shot","retrieval augmentation":"1 image, 1 text"}. This record does not assert independent reproduction.

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