# Legal Document Generation on Legal Document Drafting Dataset (test)

Source: [Legal Document Generation on Legal Document Drafting Dataset (test)](<https://www.sota2.com/research/sota/legal-document-generation-on-legal-document-drafting-dataset-test>)

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

Data updated: 2026-03-26T13:57:32.538Z

Evaluation of models performing Legal Document Generation on the test split of the Legal Document Drafting Dataset, assessed using metrics including BLEU, BERTScore, Factual Accuracy, and G-Eval.

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 | Legal Document Generation |
| Dataset (including any stated split) | Legal Document Drafting Dataset (test) |
| Metric | RL Score |
| Unit | SCALAR |
| Direction | Higher 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

- [RL Score](<https://www.sota2.com/research/sota/legal-document-generation-on-legal-document-drafting-dataset-test.md?metric=RL%20Score>): Higher is better; unit: SCALAR.

- [BLEU](<https://www.sota2.com/research/sota/legal-document-generation-on-legal-document-drafting-dataset-test.md?metric=BLEU>): Higher is better; unit: SCALAR.

- [METEOR](<https://www.sota2.com/research/sota/legal-document-generation-on-legal-document-drafting-dataset-test.md?metric=METEOR>): Higher is better; unit: SCALAR.

- [BERTScore](<https://www.sota2.com/research/sota/legal-document-generation-on-legal-document-drafting-dataset-test.md?metric=BERTScore>): Higher is better; unit: SCALAR.

- [BLANC](<https://www.sota2.com/research/sota/legal-document-generation-on-legal-document-drafting-dataset-test.md?metric=BLANC>): Higher is better; unit: SCALAR.

- [G-Eval Score](<https://www.sota2.com/research/sota/legal-document-generation-on-legal-document-drafting-dataset-test.md?metric=G-Eval%20Score>): Higher is better; unit: SCALAR.

- [Factual Accuracy](<https://www.sota2.com/research/sota/legal-document-generation-on-legal-document-drafting-dataset-test.md?metric=Factual%20Accuracy>): Higher is better; unit: PERCENT.

- [Completeness & Comprehension](<https://www.sota2.com/research/sota/legal-document-generation-on-legal-document-drafting-dataset-test.md?metric=Completeness%20%26%20Comprehension>): Higher is better; unit: SCALAR.

## Recorded results

| Method | RL Score | Result date | Report paper |
| --- | --- | --- | --- |
| Wrapper (Over LLaMA-3.1-8B) | 0.15 | Not recorded | 68fa77c41e209131cea75331 |
| Wrapper (Over Gemma-3-12B) | 0.15 | Not recorded | 68fa77c41e209131cea75331 |
| GPT-4o | 0.14 | Not recorded | 68fa77c41e209131cea75331 |
| Gemma-3-12B-It SFT | 0.11 | Not recorded | 68fa77c41e209131cea75331 |
| Qwen3-14B | 0.09 | Not recorded | 68fa77c41e209131cea75331 |
| LLaMA-3.1-8B-Instruct | 0.09 | Not recorded | 68fa77c41e209131cea75331 |
| Gemma-3-12B-It | 0.09 | Not recorded | 68fa77c41e209131cea75331 |
| LLaMA-3.1-8B-Instruct SFT | 0.08 | Not recorded | 68fa77c41e209131cea75331 |

### Wrapper (Over LLaMA-3.1-8B)

RL Score: 0.15. Result ID: 69c53a8547f34ce924aef046.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c41e209131cea75331.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c41e209131cea75331.md>)

Recorded evaluation configuration: {"Base Model":"LLaMA-3.1-8B","Prompting Strategy":"Wrapper"}. This record does not assert independent reproduction.

### Wrapper (Over Gemma-3-12B)

RL Score: 0.15. Result ID: 69c53a8547f34ce924aef049.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c41e209131cea75331.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c41e209131cea75331.md>)

Recorded evaluation configuration: {"Base Model":"Gemma-3-12B","Prompting Strategy":"Wrapper"}. This record does not assert independent reproduction.

### GPT-4o

RL Score: 0.14. Result ID: 69c53a8547f34ce924aef04a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c41e209131cea75331.md>)

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

### Gemma-3-12B-It SFT

RL Score: 0.11. Result ID: 69c53a8547f34ce924aef048.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c41e209131cea75331.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa779f1e209131cea746cd.md>)

Recorded evaluation configuration: {"Base Model":"Gemma-3-12B","Prompting Strategy":"Instruct","Fine-tuning Strategy":"SFT"}. This record does not assert independent reproduction.

### Qwen3-14B

RL Score: 0.09. Result ID: 69c53a8547f34ce924aef043.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c41e209131cea75331.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa788e1e209131cea79116.md>)

Recorded evaluation configuration: {"Base Model":"Qwen3-14B"}. This record does not assert independent reproduction.

### LLaMA-3.1-8B-Instruct

RL Score: 0.09. Result ID: 69c53a8547f34ce924aef044.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c41e209131cea75331.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75d11e209131cea6a028.md>)

[Implementation](<https://github.com/meta-llama/llama-agentic-system>)

Recorded evaluation configuration: {"Base Model":"LLaMA-3.1-8B","Prompting Strategy":"Instruct"}. This record does not assert independent reproduction.

### Gemma-3-12B-It

RL Score: 0.09. Result ID: 69c53a8547f34ce924aef047.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c41e209131cea75331.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa779f1e209131cea746cd.md>)

Recorded evaluation configuration: {"Base Model":"Gemma-3-12B","Prompting Strategy":"Instruct"}. This record does not assert independent reproduction.

### LLaMA-3.1-8B-Instruct SFT

RL Score: 0.08. Result ID: 69c53a8547f34ce924aef045.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c41e209131cea75331.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75d11e209131cea6a028.md>)

[Implementation](<https://github.com/meta-llama/llama-agentic-system>)

Recorded evaluation configuration: {"Base Model":"LLaMA-3.1-8B","Prompting Strategy":"Instruct","Fine-tuning Strategy":"SFT"}. This record does not assert independent reproduction.

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