# Unified understanding-and-generation on DPG-Bench

Source: [Unified understanding-and-generation on DPG-Bench](<https://www.sota2.com/research/sota/unified-understanding-and-generation-on-dpg-bench>)

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

Data updated: 2026-06-09T08:23:30.227Z

Evaluation of models on DPG-Bench focusing on unified understanding and generation capabilities, reporting efficiency metrics (KV Size, Attn. TFLOPs) alongside compositional fidelity scores (Entity, Relation, Attribute, Overall).

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 | Unified understanding-and-generation |
| Dataset (including any stated split) | DPG-Bench |
| Metric | KV Size (GB) |
| Unit | SIZE |
| Direction | Lower is better |
| Results loaded | 6 of 6 |
| 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

- [KV Size (GB)](<https://www.sota2.com/research/sota/unified-understanding-and-generation-on-dpg-bench.md?metric=KV%20Size%20(GB)>): Lower is better; unit: SIZE.

- [Attn. TFLOPs](<https://www.sota2.com/research/sota/unified-understanding-and-generation-on-dpg-bench.md?metric=Attn.%20TFLOPs>): Lower is better; unit: SCALAR.

- [Entity Score](<https://www.sota2.com/research/sota/unified-understanding-and-generation-on-dpg-bench.md?metric=Entity%20Score>): Higher is better; unit: PERCENT.

- [Relation Score](<https://www.sota2.com/research/sota/unified-understanding-and-generation-on-dpg-bench.md?metric=Relation%20Score>): Higher is better; unit: PERCENT.

- [Attribute Score](<https://www.sota2.com/research/sota/unified-understanding-and-generation-on-dpg-bench.md?metric=Attribute%20Score>): Higher is better; unit: PERCENT.

- [Overall Score](<https://www.sota2.com/research/sota/unified-understanding-and-generation-on-dpg-bench.md?metric=Overall%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | KV Size (GB) | Result date | Report paper |
| --- | --- | --- | --- |
| HACK++ | 0.07 | Not recorded | 6a279e4890966ade919bf932 |
| HACK++ | 0.12 | Not recorded | 6a279e4890966ade919bf932 |
| HACK | 0.26 | Not recorded | 6a279e4890966ade919bf932 |
| HACK | 0.37 | Not recorded | 6a279e4890966ade919bf932 |
| Full | 0.72 | Not recorded | 6a279e4890966ade919bf932 |
| Full | 1.23 | Not recorded | 6a279e4890966ade919bf932 |

### HACK++

KV Size (GB): 0.07. Result ID: 6a27ccccfbc373592c247f2e.

[Reporting paper](<https://www.sota2.com/research/paper/6a279e4890966ade919bf932.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a279e4890966ade919bf932.md>)

Recorded evaluation configuration: {"Backbone":"OneCAT-3B \[33\]","ηa":"30%","ηc":"10%","Batch size":"1"}. This record does not assert independent reproduction.

### HACK++

KV Size (GB): 0.12. Result ID: 6a27ccccfbc373592c247f2b.

[Reporting paper](<https://www.sota2.com/research/paper/6a279e4890966ade919bf932.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a279e4890966ade919bf932.md>)

Recorded evaluation configuration: {"Backbone":"VARGPT-v1.1 \[32\]","ηa":"30%","ηc":"10%","Batch size":"1"}. This record does not assert independent reproduction.

### HACK

KV Size (GB): 0.26. Result ID: 6a27ccccfbc373592c247f2d.

[Reporting paper](<https://www.sota2.com/research/paper/6a279e4890966ade919bf932.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77ec1e209131cea760c8.md>)

Recorded evaluation configuration: {"Backbone":"OneCAT-3B \[33\]","ηa":"30%","ηc":"30%","Batch size":"1"}. This record does not assert independent reproduction.

### HACK

KV Size (GB): 0.37. Result ID: 6a27ccccfbc373592c247f2a.

[Reporting paper](<https://www.sota2.com/research/paper/6a279e4890966ade919bf932.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77ec1e209131cea760c8.md>)

Recorded evaluation configuration: {"Backbone":"VARGPT-v1.1 \[32\]","ηa":"30%","ηc":"30%","Batch size":"1"}. This record does not assert independent reproduction.

### Full

KV Size (GB): 0.72. Result ID: 6a27ccccfbc373592c247f2c.

[Reporting paper](<https://www.sota2.com/research/paper/6a279e4890966ade919bf932.md>)

Recorded evaluation configuration: {"Backbone":"OneCAT-3B \[33\]","ηa":"100%","ηc":"100%","Batch size":"1"}. This record does not assert independent reproduction.

### Full

KV Size (GB): 1.23. Result ID: 6a27ccccfbc373592c247f29.

[Reporting paper](<https://www.sota2.com/research/paper/6a279e4890966ade919bf932.md>)

Recorded evaluation configuration: {"Backbone":"VARGPT-v1.1 \[32\]","ηa":"100%","ηc":"100%","Batch size":"1"}. This record does not assert independent reproduction.

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