James Edmunds commited on
Commit ·
57df620
1
Parent(s): 9567ff8
Stage 1 HF Integration. Added upload_embeddings.py
Browse files- scripts/upload_embeddings.py +36 -0
scripts/upload_embeddings.py
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""Upload embeddings to HuggingFace"""
|
| 2 |
+
import sys
|
| 3 |
+
from pathlib import Path
|
| 4 |
+
from huggingface_hub import HfApi
|
| 5 |
+
from dotenv import load_dotenv
|
| 6 |
+
|
| 7 |
+
# Add parent directory to path
|
| 8 |
+
sys.path.append(str(Path(__file__).parent.parent))
|
| 9 |
+
from config.settings import Settings
|
| 10 |
+
|
| 11 |
+
|
| 12 |
+
def main():
|
| 13 |
+
"""Upload embeddings directory to HuggingFace dataset"""
|
| 14 |
+
print("Starting upload process...")
|
| 15 |
+
|
| 16 |
+
# Print size info
|
| 17 |
+
total_size = sum(
|
| 18 |
+
f.stat().st_size for f in Settings.EMBEDDINGS_DIR.glob('**/*') if f.is_file()
|
| 19 |
+
)
|
| 20 |
+
print(f"Found embeddings: {total_size / 1024 / 1024:.2f} MB")
|
| 21 |
+
|
| 22 |
+
api = HfApi(token=Settings.HF_TOKEN)
|
| 23 |
+
|
| 24 |
+
print(f"Uploading to {Settings.HF_DATASET}...")
|
| 25 |
+
api.upload_folder(
|
| 26 |
+
folder_path=str(Settings.EMBEDDINGS_DIR),
|
| 27 |
+
repo_id=Settings.HF_DATASET,
|
| 28 |
+
repo_type="dataset"
|
| 29 |
+
)
|
| 30 |
+
|
| 31 |
+
print("Upload complete!")
|
| 32 |
+
|
| 33 |
+
|
| 34 |
+
if __name__ == "__main__":
|
| 35 |
+
load_dotenv()
|
| 36 |
+
main()
|