voxreality/navigation_intructions_v2

---
license: apache-2.0
task_categories:
- text-generation
language:
- en
---

## Dataset Description

The Navigation Dataset is part of the European VOXReality project, aimed at fine-tuning models to generate precise navigation instructions for users. This dataset includes pairs of user inputs, knowledge bases, and corresponding generated outputs, focusing on providing detailed navigation instructions within a defined space.

## Dataset Structure

The dataset consists of three primary fields:

1. **input**: The user's input query, asking for directions to various locations.
2. **knowledge**: The knowledge base string length related to the input.
3. **output**: The generated navigation instructions, providing step-by-step guidance to reach the desired location.

### Example

Below is an example of a data entry in the dataset:

```python
Input: "Can you show me the way to the social space?"
Knowledge: "start, turn right, turn left, crossing yellow sphere left, crossing booth 2 right, arrive booth 1, turn right, crossing booth 1 left, arrive booth 4, finish"
Output: "To find the social space, follow these steps:\n- Start by turning right.\n- Turn left.\n- Cross the yellow sphere on your left.\n- Cross booth 2 on your right.\n- Continue until you arrive at booth 1.\n- Turn right.\n- Cross booth 1 on your left.\n\nYou should now be at booth 4."
```