Update README.md
Browse files
README.md
CHANGED
@@ -26,7 +26,34 @@ Simply use Git (or TortoiseGit):
|
|
26 |
git clone https://huggingface.co/datasets/sraimund/MapPool/
|
27 |
```
|
28 |
|
29 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
30 |
|
31 |
## How can the parquet files be read?
|
32 |
|
|
|
26 |
git clone https://huggingface.co/datasets/sraimund/MapPool/
|
27 |
```
|
28 |
|
29 |
+
Alternatively use the HuggingFace API:
|
30 |
+
```python
|
31 |
+
import json
|
32 |
+
import os
|
33 |
+
from huggingface_hub import hf_hub_download
|
34 |
+
|
35 |
+
download_folder = "<your-download-folder>"
|
36 |
+
repo_id = "sraimund/MapPool"
|
37 |
+
|
38 |
+
# this file is given at the root of this repository
|
39 |
+
with open("file_list.json") as f:
|
40 |
+
file_list = json.load(f)
|
41 |
+
|
42 |
+
for part, files in file_list.items():
|
43 |
+
for file in files:
|
44 |
+
file_path = f"{download_folder}/{part}/{file}.parquet"
|
45 |
+
|
46 |
+
if os.path.exists(file_path):
|
47 |
+
continue
|
48 |
+
|
49 |
+
hf_hub_download(repo_type="dataset",
|
50 |
+
repo_id=repo_id,
|
51 |
+
filename=f"{part}/{file}.parquet",
|
52 |
+
local_dir=download_folder,
|
53 |
+
token=read_token)
|
54 |
+
```
|
55 |
+
|
56 |
+
About 225 GB of space are required. The amount doubles when using Git since the files are duplicated in the .git folder.
|
57 |
|
58 |
## How can the parquet files be read?
|
59 |
|