Tera-AI/STRIDE

Dataset

STRIDE

5

31 commits

2 linked in READMEs

updated Aug 11, 2025

See the code

README

STRIDE

We develop STRIDE (Spatio-Temporal Road Image Dataset for Exploration), that consists of approximately 82B tokens which were arranged into a total of 6M visual "sentences" or token sequences. The sequences are generated from a relatively small set of 131k panoramic images, along with their metadata and openly available highway system data. Our approach enables a 27x increase in information, allowing for generative world model training. In essence, STRIDE turns real street-view panoramic observations into a navigable, interactive environment suitable for free-play across space and time.

For ease of use and data exploration, we prepared this sample of 10k detokenized paths, which amounts to about 100k projected panoramic images along with its corresponding metadata.

Breakdown

Dataset (STRIDE)

The full tokenized dataset is made available through two downloadable files in a public GCS bucket:

gsutil -m cp gs://tera-tardis/STRIDE-1/training.jsonl . # ~327GB
gsutil -m cp gs://tera-tardis/STRIDE-1/testing.jsonl . # ~9GB

In the future, the fully detokenized dataset will be made available. Should you need it, feel free to contact the authors.

San Mateo Coverage Map

image/png

Above is the 70km^2 area we selected for putting together the Google StreetView data using openly available road data. Each directly connected component of the graph is represented with a distinct color, for ease of visualization.

Queens Coverage Map

image/png

Above is an additional Queens, NY 27km^2 area. Each directly connected component of the graph is represented with a distinct color, for ease of visualization.

Code (TARDIS)

The code used for training of the model is available on GitHub.

Checkpoints (TARDIS)

The checkpoint/state used for evaluation of the model was saved in MessagePack format and is made available through this downloadable file:

gsutil -m cp gs://tera-tardis/STRIDE-1/checkpoint.msgpack . # ~10GB

Should you need other checkpoints, feel free to contact the authors.

Project Website

The project website is available here.

Contacts

Paper

TARDIS STRIDE: A Spatio-Temporal Road Image Dataset for Exploration and Autonomy

autonomous-driving
road-images
spatio-temporal

Contributors

VctorAHernndez

25 commits

nielsr

3 commits

hcarrion

2 commits

ayushzenith

1 commits

Tera-AI/STRIDE

Dataset

STRIDE

5

31 commits

2 linked in READMEs

updated Aug 11, 2025

See the code

README

STRIDE

We develop STRIDE (Spatio-Temporal Road Image Dataset for Exploration), that consists of approximately 82B tokens which were arranged into a total of 6M visual "sentences" or token sequences. The sequences are generated from a relatively small set of 131k panoramic images, along with their metadata and openly available highway system data. Our approach enables a 27x increase in information, allowing for generative world model training. In essence, STRIDE turns real street-view panoramic observations into a navigable, interactive environment suitable for free-play across space and time.

For ease of use and data exploration, we prepared this sample of 10k detokenized paths, which amounts to about 100k projected panoramic images along with its corresponding metadata.

Breakdown

Dataset (STRIDE)

The full tokenized dataset is made available through two downloadable files in a public GCS bucket:

gsutil -m cp gs://tera-tardis/STRIDE-1/training.jsonl . # ~327GB
gsutil -m cp gs://tera-tardis/STRIDE-1/testing.jsonl . # ~9GB

In the future, the fully detokenized dataset will be made available. Should you need it, feel free to contact the authors.

San Mateo Coverage Map

image/png

Above is the 70km^2 area we selected for putting together the Google StreetView data using openly available road data. Each directly connected component of the graph is represented with a distinct color, for ease of visualization.

Queens Coverage Map

image/png

Above is an additional Queens, NY 27km^2 area. Each directly connected component of the graph is represented with a distinct color, for ease of visualization.

Code (TARDIS)

The code used for training of the model is available on GitHub.

Checkpoints (TARDIS)

The checkpoint/state used for evaluation of the model was saved in MessagePack format and is made available through this downloadable file:

gsutil -m cp gs://tera-tardis/STRIDE-1/checkpoint.msgpack . # ~10GB

Should you need other checkpoints, feel free to contact the authors.

Project Website

The project website is available here.

Contacts

Paper

TARDIS STRIDE: A Spatio-Temporal Road Image Dataset for Exploration and Autonomy

autonomous-driving
road-images
spatio-temporal

Contributors

VctorAHernndez

25 commits

nielsr

3 commits

hcarrion

2 commits

ayushzenith

1 commits