# Navigation Instruction Generation on Unseen routes 700 instances (test)

Source: [Navigation Instruction Generation on Unseen routes 700 instances (test)](<https://www.sota2.com/research/sota/navigation-instruction-generation-on-unseen-routes-700-instances-test>)

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

Data updated: 2026-02-26T05:47:15.441Z

Evaluation of performance for generating navigation instructions based on unseen routes derived from the 700-instance dataset, measured using generation quality metrics like BLEU, length, landmark accuracy, and path similarity.

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 | Navigation Instruction Generation |
| Dataset (including any stated split) | Unseen routes 700 instances (test) |
| Metric | Instruction Length |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 2 of 2 |
| 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

- [Instruction Length](<https://www.sota2.com/research/sota/navigation-instruction-generation-on-unseen-routes-700-instances-test.md?metric=Instruction%20Length>): Higher is better; unit: SCALAR.

- [BLEU Score](<https://www.sota2.com/research/sota/navigation-instruction-generation-on-unseen-routes-700-instances-test.md?metric=BLEU%20Score>): Higher is better; unit: SCALAR.

- [Landmark Accuracy](<https://www.sota2.com/research/sota/navigation-instruction-generation-on-unseen-routes-700-instances-test.md?metric=Landmark%20Accuracy>): Higher is better; unit: SCALAR.

- [SDTW Similarity](<https://www.sota2.com/research/sota/navigation-instruction-generation-on-unseen-routes-700-instances-test.md?metric=SDTW%20Similarity>): Higher is better; unit: SCALAR.

- [Success Rate](<https://www.sota2.com/research/sota/navigation-instruction-generation-on-unseen-routes-700-instances-test.md?metric=Success%20Rate>): Higher is better; unit: PERCENT.

- [Instruction Success Rate](<https://www.sota2.com/research/sota/navigation-instruction-generation-on-unseen-routes-700-instances-test.md?metric=Instruction%20Success%20Rate>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Instruction Length | Result date | Report paper |
| --- | --- | --- | --- |
| reference | 54.2 | Not recorded | 68fa6a011e209131cea2552c |
| g2t+pretrain | 53.2 | Not recorded | 68fa6a011e209131cea2552c |

### reference

Instruction Length: 54.2. Result ID: 698dfcddffc3901de099f621.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a011e209131cea2552c.md>)

Recorded evaluation configuration: {"type":"human reference"}. This record does not assert independent reproduction.

### g2t+pretrain

Instruction Length: 53.2. Result ID: 698dfcddffc3901de099f622.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6a011e209131cea2552c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a011e209131cea2552c.md>)

[Implementation](<https://www.sota2.com/www.cl.uni-heidelberg.de/statnlpgroup/map2seq/>)

Recorded evaluation configuration: {"pretraining":true}. This record does not assert independent reproduction.

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