Datasets:
File size: 2,404 Bytes
84a242a | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 | # Upload this prepared folder
Upload **the contents of MolAS_HF** to the root of a Hugging Face **dataset** repository, keeping the directory layout. The dataset card, mixed-license metadata, graph payloads, small inputs, result archives, source mirror, and checksums are already prepared. No scientific runs or data conversions are needed before upload.
## Recommended upload
The destination dataset repository is https://huggingface.co/datasets/SeanWan05/MolAS-Active-Learning. Use the current official Hub client for a resumable upload of this approximately 15.5 GB folder:
```bash
python3 -m pip install --upgrade huggingface_hub
hf auth login
cd /path/to/MolAS_HF
python3 scripts/verify_hf_package.py
hf upload SeanWan05/MolAS-Active-Learning . . --repo-type dataset
```
Authentication belongs in the normal client login, never in this folder. The current Hub client handles large-folder uploading; rerun the same command if interrupted. See https://huggingface.co/docs/huggingface_hub/guides/upload.
The optional helper `python3 scripts/upload_to_hub.py --repo-id SeanWan05/MolAS-Active-Learning` prints a plan only. Add `--upload` to perform the upload using an already authenticated client and an existing dataset repository. It does not create or change repository visibility.
The browser uploader can also upload these files, but the Hub client is preferable for thousands of graph files and resumability. Include the root README.md, LICENSE files, and hidden .gitattributes/.gitignore files. The HF .gitignore deliberately **does not exclude data/external/**; the GitHub repository's ignore rules must not replace it.
## After the upload completes
Open the public dataset URL without signing in and check its Files tab, dataset-card metadata, and all nine Viewer configurations. Record the actual dataset commit/revision and, if desired, tag the complete dataset as v1.0.0. Link that real public dataset URL and the existing GitHub code URL in the camera-ready manuscript; add the dataset URL to the GitHub README. Never substitute a planned/nonexistent URL for the actual publication.
The package has fewer than 100,000 files, fewer than 10,000 entries in any folder, and no individual file near the Hub's large-file recommendations. Storage/account eligibility is evaluated by Hugging Face when uploading; this local preparation does not claim that an upload has already happened.
|