mjf-su/PhysicalAI-ADE-US

---
pretty_name: PhysicalAI-US-ADE
language:
- en
license: mit
task_categories:
- text-generation
tags:
- autonomous-driving
- trajectory-prediction
- evaluation
- waypoint-prediction
- physical-ai
- nvidia
- us-driving
size_categories:
- 100K<n<1M
---

# PhysicalAI-US-ADE

## Dataset Summary

**PhysicalAI-US-ADE** contains per-sample evaluation outputs for autonomous driving waypoint prediction on the **US subset** of the **PhysicalAI NVIDIA dataset**.

This dataset stores inference-time predictions and evaluation statistics for models evaluated on the dataset, organized by model name at the top level. Each model directory contains sample-level records for that model’s predictions against ground truth.

The current release includes evaluation results for:

- `base-AV-VLA` — corresponding to the model [`mjf-su/base-AV-VLA`](https://huggingface.co/mjf-su/base-AV-VLA)

The underlying source dataset used for evaluation is:

- [`tom-jerry-123/Physical-AI-AV-US`](https://huggingface.co/datasets/tom-jerry-123/Physical-AI-AV-US)

## What this dataset contains

Each JSONL record corresponds to a single evaluated sample and includes:

- sample identifiers:
- `scene_id`
- `chunk_name`
- `sample_idx`
- `timestamp_us`
- model outputs:
- `prediction_waypoints`
- reference targets:
- `ground_truth_waypoints`
- evaluation metrics:
- `ADE`
- `ADE_normalized`
- `ADE_combined`
- parsing / validity flags:
- `waypoint_count_match`
- `parse_fail`

## Directory structure

At the top level, each directory corresponds to one evaluated model.

Example:

```text
PhysicalAI-US-ADE/
├── base-AV-VLA/
│ ├── shard_00000.jsonl
│ ├── shard_00001.jsonl
│ ├── ...
│ └── CoT-targets.jsonl