# Structured tax-form completion on TaxCalcBench TY24 (train)

Source: [Structured tax-form completion on TaxCalcBench TY24 (train)](<https://www.sota2.com/research/sota/structured-tax-form-completion-on-taxcalcbench-ty24-train>)

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

Data updated: 2026-05-12T08:00:27.566Z

Evaluation of model performance on the training split of the TaxCalcBench TY24 dataset for structured tax-form completion tasks.

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 | Structured tax-form completion |
| Dataset (including any stated split) | TaxCalcBench TY24 (train) |
| Metric | Train-A Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 10 of 10 |
| 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

- [Train-A Score](<https://www.sota2.com/research/sota/structured-tax-form-completion-on-taxcalcbench-ty24-train.md?metric=Train-A%20Score>): Higher is better; unit: PERCENT.

- [Train-B Score](<https://www.sota2.com/research/sota/structured-tax-form-completion-on-taxcalcbench-ty24-train.md?metric=Train-B%20Score>): Higher is better; unit: PERCENT.

- [Train-C Score](<https://www.sota2.com/research/sota/structured-tax-form-completion-on-taxcalcbench-ty24-train.md?metric=Train-C%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Train-A Score | Result date | Report paper |
| --- | --- | --- | --- |
| DPA-GRPO | 90 | Not recorded | 6a02b3f8dbc836b87d82459d |
| Zero-shot | 84 | Not recorded | 6a02b3f8dbc836b87d82459d |
| GRPO | 84 | Not recorded | 6a02b3f8dbc836b87d82459d |
| DAPO | 84 | Not recorded | 6a02b3f8dbc836b87d82459d |
| GSPO | 84 | Not recorded | 6a02b3f8dbc836b87d82459d |
| GRPO | 74 | Not recorded | 6a02b3f8dbc836b87d82459d |
| DAPO | 74 | Not recorded | 6a02b3f8dbc836b87d82459d |
| GSPO | 74 | Not recorded | 6a02b3f8dbc836b87d82459d |
| DPA-GRPO | 74 | Not recorded | 6a02b3f8dbc836b87d82459d |
| Zero-shot | 53 | Not recorded | 6a02b3f8dbc836b87d82459d |

### DPA-GRPO

Train-A Score: 90. Result ID: 6a02dce347f34ce924e56e11.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b3f8dbc836b87d82459d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a02b3f8dbc836b87d82459d.md>)

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

### Zero-shot

Train-A Score: 84. Result ID: 6a02dce347f34ce924e56e09.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b3f8dbc836b87d82459d.md>)

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

### GRPO

Train-A Score: 84. Result ID: 6a02dce347f34ce924e56e0d.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b3f8dbc836b87d82459d.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.

### DAPO

Train-A Score: 84. Result ID: 6a02dce347f34ce924e56e0e.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b3f8dbc836b87d82459d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78a81e209131cea79ac6.md>)

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

### GSPO

Train-A Score: 84. Result ID: 6a02dce347f34ce924e56e0f.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b3f8dbc836b87d82459d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a481e209131cea8231d.md>)

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

### GRPO

Train-A Score: 74. Result ID: 6a02dce347f34ce924e56e0a.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b3f8dbc836b87d82459d.md>)

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

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

Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction.

### DAPO

Train-A Score: 74. Result ID: 6a02dce347f34ce924e56e0b.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b3f8dbc836b87d82459d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78a81e209131cea79ac6.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction.

### GSPO

Train-A Score: 74. Result ID: 6a02dce347f34ce924e56e0c.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b3f8dbc836b87d82459d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a481e209131cea8231d.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction.

### DPA-GRPO

Train-A Score: 74. Result ID: 6a02dce347f34ce924e56e10.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b3f8dbc836b87d82459d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a02b3f8dbc836b87d82459d.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction.

### Zero-shot

Train-A Score: 53. Result ID: 6a02dce347f34ce924e56e08.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b3f8dbc836b87d82459d.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction.

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