# Long-Horizon Vision-Language Navigation on LH-VLN 3-4 Subtasks

Source: [Long-Horizon Vision-Language Navigation on LH-VLN 3-4 Subtasks](<https://www.sota2.com/research/sota/long-horizon-vision-language-navigation-on-lh-vln-3-4-subtasks>)

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

Data updated: 2026-02-26T06:03:04.627Z

Evaluation of Long-Horizon Vision-Language Navigation performance on the LH-VLN 3-4 Subtasks dataset, measured using metrics including Success Rate (SR), Navigation Error (NE), Intermediate Success Rate (ISR), Constraint Success Rate (CSR), and Correct Goal Time (CGT).

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 | Long-Horizon Vision-Language Navigation |
| Dataset (including any stated split) | LH-VLN 3-4 Subtasks |
| Metric | SR |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 9 of 9 |
| 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

- [SR](<https://www.sota2.com/research/sota/long-horizon-vision-language-navigation-on-lh-vln-3-4-subtasks.md?metric=SR>): Higher is better; unit: PERCENT.

- [NE](<https://www.sota2.com/research/sota/long-horizon-vision-language-navigation-on-lh-vln-3-4-subtasks.md?metric=NE>): Lower is better; unit: SCALAR.

- [ISR](<https://www.sota2.com/research/sota/long-horizon-vision-language-navigation-on-lh-vln-3-4-subtasks.md?metric=ISR>): Higher is better; unit: PERCENT.

- [CSR](<https://www.sota2.com/research/sota/long-horizon-vision-language-navigation-on-lh-vln-3-4-subtasks.md?metric=CSR>): Higher is better; unit: PERCENT.

- [CGT](<https://www.sota2.com/research/sota/long-horizon-vision-language-navigation-on-lh-vln-3-4-subtasks.md?metric=CGT>): Lower is better; unit: TIME.

## Recorded results

| Method | SR | Result date | Report paper |
| --- | --- | --- | --- |
| Random | 0 | Not recorded | 68fa77831e209131cea73bb6 |
| InstructNav | 0 | Not recorded | 68fa77831e209131cea73bb6 |
| GLM-4v prompt | 0 | Not recorded | 68fa77831e209131cea73bb6 |
| NaviLLM | 0 | Not recorded | 68fa77831e209131cea73bb6 |
| NaviLLM | 0 | Not recorded | 68fa77831e209131cea73bb6 |
| GPT-4 + NaviLLM | 0 | Not recorded | 68fa77831e209131cea73bb6 |
| MGDM | 0 | Not recorded | 68fa77831e209131cea73bb6 |
| MGDM | 0 | Not recorded | 68fa77831e209131cea73bb6 |
| MGDM | 0 | Not recorded | 68fa77831e209131cea73bb6 |

### Random

SR: 0. Result ID: 69952b1e3badfaffe6113dd3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77831e209131cea73bb6.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### InstructNav

SR: 0. Result ID: 69952b1e3badfaffe6113dd4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77831e209131cea73bb6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73831e209131cea5cd89.md>)

Recorded evaluation configuration: {"Type":"Zero-shot"}. This record does not assert independent reproduction.

### GLM-4v prompt

SR: 0. Result ID: 69952b1e3badfaffe6113dd5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77831e209131cea73bb6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74521e209131cea610b6.md>)

Recorded evaluation configuration: {"Type":"Zero-shot"}. This record does not assert independent reproduction.

### NaviLLM

SR: 0. Result ID: 69952b1e3badfaffe6113dd6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77831e209131cea73bb6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72be1e209131cea58557.md>)

[Implementation](<https://github.com/zd11024/NaviLLM>)

Recorded evaluation configuration: {"Type":"Pretrain"}. This record does not assert independent reproduction.

### NaviLLM

SR: 0. Result ID: 69952b1e3badfaffe6113dd7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77831e209131cea73bb6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72be1e209131cea58557.md>)

[Implementation](<https://github.com/zd11024/NaviLLM>)

Recorded evaluation configuration: {"Type":"Finetuned"}. This record does not assert independent reproduction.

### GPT-4 + NaviLLM

SR: 0. Result ID: 69952b1e3badfaffe6113dd8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77831e209131cea73bb6.md>)

Recorded evaluation configuration: {"Type":"Pretrain"}. This record does not assert independent reproduction.

### MGDM

SR: 0. Result ID: 69952b1e3badfaffe6113dd9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77831e209131cea73bb6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77831e209131cea73bb6.md>)

[Implementation](<https://hcplab-sysu.github.io/LH-VLN>)

Recorded evaluation configuration: {"Type":"Finetuned","LLM":"Llama3","Training Strategy":"Two Stage"}. This record does not assert independent reproduction.

### MGDM

SR: 0. Result ID: 69952b1e3badfaffe6113dda.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77831e209131cea73bb6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77831e209131cea73bb6.md>)

[Implementation](<https://hcplab-sysu.github.io/LH-VLN>)

Recorded evaluation configuration: {"Type":"Finetuned","LLM":"Vicuna","Training Strategy":"Alternate"}. This record does not assert independent reproduction.

### MGDM

SR: 0. Result ID: 69952b1e3badfaffe6113ddb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77831e209131cea73bb6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77831e209131cea73bb6.md>)

[Implementation](<https://hcplab-sysu.github.io/LH-VLN>)

Recorded evaluation configuration: {"Type":"Finetuned","LLM":"Vicuna","Training Strategy":"Two Stage"}. This record does not assert independent reproduction.

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