onedean commited on
Commit
a399faa
·
verified ·
1 Parent(s): bc2052a

Add dataset card

Browse files
Files changed (1) hide show
  1. README.md +23 -20
README.md CHANGED
@@ -1,32 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  # UrbanFM Datasets
2
 
3
- The full dataset (~10 GB) is hosted on Hugging Face:
4
 
5
- **https://huggingface.co/datasets/Onedean/UrbanFM-datasets**
6
 
7
- ## Download
 
 
 
8
 
9
- Install the Hugging Face CLI:
10
 
11
  ```bash
12
  pip install huggingface_hub
13
- ```
14
-
15
- Download the entire datasets folder into the project root:
16
-
17
- ```bash
18
  huggingface-cli download Onedean/UrbanFM-datasets --repo-type dataset --local-dir ./datasets
19
  ```
20
 
21
- Or use the provided script from the project root:
22
-
23
- ```bash
24
- bash scripts/download_datasets.sh
25
- ```
26
-
27
- ## Contents
28
 
29
- - `eval_datasets/` — evaluation datasets
30
- - `full_pretrain_datasets/` — full pre-training datasets
31
- - `pretrain_datasets/` — pre-training datasets
32
- - `clean.ipynb`, `spatial_index_generation.py`, `data_index_process.sh` — data processing scripts
 
1
+ ---
2
+ license: mit
3
+ task_categories:
4
+ - time-series-forecasting
5
+ language:
6
+ - en
7
+ tags:
8
+ - spatio-temporal
9
+ - urban-computing
10
+ - foundation-model
11
+ size_categories:
12
+ - 1B<n<10B
13
+ ---
14
+
15
  # UrbanFM Datasets
16
 
17
+ Datasets for [UrbanFM: Scaling Urban Spatio-Temporal Foundation Models](https://github.com/Onedean/UrbanFM).
18
 
19
+ ## Structure
20
 
21
+ - `eval_datasets/` — evaluation datasets (traffic flow, speed, occupancy, etc.)
22
+ - `full_pretrain_datasets/` — full-scale pre-training datasets
23
+ - `pretrain_datasets/` — pre-training datasets
24
+ - `clean.ipynb`, `spatial_index_generation.py`, `data_index_process.sh` — data processing utilities
25
 
26
+ ## Download
27
 
28
  ```bash
29
  pip install huggingface_hub
 
 
 
 
 
30
  huggingface-cli download Onedean/UrbanFM-datasets --repo-type dataset --local-dir ./datasets
31
  ```
32
 
33
+ ## Citation
 
 
 
 
 
 
34
 
35
+ If you use these datasets, please cite the UrbanFM paper and star the [GitHub repository](https://github.com/Onedean/UrbanFM).