# Dialogue on Dialogue (test)

Source: [Dialogue on Dialogue (test)](<https://www.sota2.com/research/sota/dialogue-on-dialogue-test>)

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

Data updated: 2026-02-26T05:45:22.343Z

Evaluation of dialogue system performance on the test split of the Dialogue dataset, measured using human-centric metrics including Fluency, Accuracy, Toxicity, and Helpfulness.

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 | Dialogue |
| Dataset (including any stated split) | Dialogue (test) |
| Metric | Fluency |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| 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

- [Fluency](<https://www.sota2.com/research/sota/dialogue-on-dialogue-test.md?metric=Fluency>): Higher is better; unit: SCALAR.

- [Accuracy](<https://www.sota2.com/research/sota/dialogue-on-dialogue-test.md?metric=Accuracy>): Higher is better; unit: SCALAR.

- [Toxicity](<https://www.sota2.com/research/sota/dialogue-on-dialogue-test.md?metric=Toxicity>): Lower is better; unit: SCALAR.

- [Helpfulness](<https://www.sota2.com/research/sota/dialogue-on-dialogue-test.md?metric=Helpfulness>): Higher is better; unit: SCALAR.

- [Overall Score](<https://www.sota2.com/research/sota/dialogue-on-dialogue-test.md?metric=Overall%20Score>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Fluency | Result date | Report paper |
| --- | --- | --- | --- |
| SFT | 8.84 | Not recorded | 68fa73b51e209131cea5e0b2 |
| HBAT | 8.8 | Not recorded | 68fa73b51e209131cea5e0b2 |
| HBAT | 8.79 | Not recorded | 68fa73b51e209131cea5e0b2 |
| DPO | 8.67 | Not recorded | 68fa73b51e209131cea5e0b2 |
| RLHF | 8.39 | Not recorded | 68fa73b51e209131cea5e0b2 |

### SFT

Fluency: 8.84. Result ID: 69931a113a6c31484913b8d6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73b51e209131cea5e0b2.md>)

Recorded evaluation configuration: {"Backbone":"LLaMA2-13B"}. This record does not assert independent reproduction.

### HBAT

Fluency: 8.8. Result ID: 69931a113a6c31484913b8d8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73b51e209131cea5e0b2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73b51e209131cea5e0b2.md>)

[Implementation](<https://github.com/wangclnlp/DeepSpeed-Chat-Extension/tree/main/examples/hybrid_alignment_training>)

Recorded evaluation configuration: {"Backbone":"LLaMA2-13B","Optimization":"PPO"}. This record does not assert independent reproduction.

### HBAT

Fluency: 8.79. Result ID: 69931a113a6c31484913b8da.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73b51e209131cea5e0b2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73b51e209131cea5e0b2.md>)

[Implementation](<https://github.com/wangclnlp/DeepSpeed-Chat-Extension/tree/main/examples/hybrid_alignment_training>)

Recorded evaluation configuration: {"Backbone":"LLaMA2-13B","Optimization":"DPO"}. This record does not assert independent reproduction.

### DPO

Fluency: 8.67. Result ID: 69931a113a6c31484913b8d9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73b51e209131cea5e0b2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74511e209131cea6105d.md>)

Recorded evaluation configuration: {"Backbone":"LLaMA2-13B","Optimization":"DPO"}. This record does not assert independent reproduction.

### RLHF

Fluency: 8.39. Result ID: 69931a113a6c31484913b8d7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73b51e209131cea5e0b2.md>)

Recorded evaluation configuration: {"Backbone":"LLaMA2-13B","Optimization":"PPO"}. This record does not assert independent reproduction.

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