# Unsupervised Multimodal Machine Translation on Multi30K En-De and De-En (test)

Source: [Unsupervised Multimodal Machine Translation on Multi30K En-De and De-En (test)](<https://www.sota2.com/research/sota/unsupervised-multimodal-machine-translation-on-multi30k-en-de-and-de-en-test>)

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

Data updated: 2026-02-26T05:36:29.320Z

Evaluation of Unsupervised Multimodal Machine Translation performance on the combined English-to-German and German-to-English test splits of the Multi30K dataset, using metrics like Avg. BLEU, Completeness, Ambiguity, and Fluency.

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 | Unsupervised Multimodal Machine Translation |
| Dataset (including any stated split) | Multi30K En-De and De-En (test) |
| Metric | Avg. BLEU |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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

- [Avg. BLEU](<https://www.sota2.com/research/sota/unsupervised-multimodal-machine-translation-on-multi30k-en-de-and-de-en-test.md?metric=Avg.%20BLEU>): Higher is better; unit: PERCENT.

- [Completeness](<https://www.sota2.com/research/sota/unsupervised-multimodal-machine-translation-on-multi30k-en-de-and-de-en-test.md?metric=Completeness>): Higher is better; unit: SCALAR.

- [Ambiguity](<https://www.sota2.com/research/sota/unsupervised-multimodal-machine-translation-on-multi30k-en-de-and-de-en-test.md?metric=Ambiguity>): Lower is better; unit: SCALAR.

- [Fluency](<https://www.sota2.com/research/sota/unsupervised-multimodal-machine-translation-on-multi30k-en-de-and-de-en-test.md?metric=Fluency>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Avg. BLEU | Result date | Report paper |
| --- | --- | --- | --- |
| Ours | 39.3 | Not recorded | 68fa6f7c1e209131cea45e36 |
| w/o SG | 35.7 | Not recorded | 68fa6f7c1e209131cea45e36 |
| PVP\* (PR) | 35 | Not recorded | 68fa6f7c1e209131cea45e36 |
| PVP\* (SR) | 33.2 | Not recorded | 68fa6f7c1e209131cea45e36 |

### Ours

Avg. BLEU: 39.3. Result ID: 6990c4253a6c314849031aee.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f7c1e209131cea45e36.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f7c1e209131cea45e36.md>)

[Implementation](<https://github.com/scofield7419/UMMT-VSH>)

Recorded evaluation configuration: {"scene\_graph\_usage":"true"}. This record does not assert independent reproduction.

### w/o SG

Avg. BLEU: 35.7. Result ID: 6990c4253a6c314849031aef.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f7c1e209131cea45e36.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f7c1e209131cea45e36.md>)

[Implementation](<https://github.com/scofield7419/UMMT-VSH>)

Recorded evaluation configuration: {"scene\_graph\_usage":"false"}. This record does not assert independent reproduction.

### PVP\* (PR)

Avg. BLEU: 35. Result ID: 6990c4253a6c314849031aed.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f7c1e209131cea45e36.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c141e209131cea32003.md>)

Recorded evaluation configuration: {"visual\_hallucination":"phrase-level retrieval (PR)"}. This record does not assert independent reproduction.

### PVP\* (SR)

Avg. BLEU: 33.2. Result ID: 6990c4253a6c314849031aec.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f7c1e209131cea45e36.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c141e209131cea32003.md>)

Recorded evaluation configuration: {"visual\_hallucination":"sentence-level retrieval (SR)"}. This record does not assert independent reproduction.

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