Spaces:
Sleeping
Sleeping
added dumpy
Browse files
dumpy.py
ADDED
|
@@ -0,0 +1,52 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import json
|
| 2 |
+
import logging
|
| 3 |
+
import os
|
| 4 |
+
|
| 5 |
+
import argilla as rg
|
| 6 |
+
from huggingface_hub import HfApi
|
| 7 |
+
|
| 8 |
+
logger = logging.getLogger(__name__)
|
| 9 |
+
logger.setLevel(logging.INFO)
|
| 10 |
+
|
| 11 |
+
if __name__ == "__main__":
|
| 12 |
+
logger.info("*** Initializing Argilla session ***")
|
| 13 |
+
rg.init(
|
| 14 |
+
api_url=os.getenv("ARGILLA_API_URL"),
|
| 15 |
+
api_key=os.getenv("ARGILLA_API_KEY"),
|
| 16 |
+
extra_headers={"Authorization": f"Bearer {os.getenv('HF_TOKEN')}"},
|
| 17 |
+
)
|
| 18 |
+
|
| 19 |
+
logger.info("*** Fetching dataset from Argilla ***")
|
| 20 |
+
dataset = rg.FeedbackDataset.from_argilla(
|
| 21 |
+
os.getenv("SOURCE_DATASET"),
|
| 22 |
+
workspace=os.getenv("SOURCE_WORKSPACE"),
|
| 23 |
+
)
|
| 24 |
+
logger.info("*** Filtering records by `response_status` ***")
|
| 25 |
+
dataset = dataset.filter_by(response_status=["submitted"]) # type: ignore
|
| 26 |
+
|
| 27 |
+
logger.info("*** Calculating users and annotation count ***")
|
| 28 |
+
output = {}
|
| 29 |
+
for record in dataset.records:
|
| 30 |
+
for response in record.responses:
|
| 31 |
+
if response.user_id not in output:
|
| 32 |
+
output[response.user_id] = 0
|
| 33 |
+
output[response.user_id] += 1
|
| 34 |
+
|
| 35 |
+
for key in list(output.keys()):
|
| 36 |
+
output[rg.User.from_id(key).username] = output.pop(key)
|
| 37 |
+
|
| 38 |
+
logger.info("*** Users and annotation count successfully calculated! ***")
|
| 39 |
+
|
| 40 |
+
logger.info("*** Dumping Python dict into `stats.json` ***")
|
| 41 |
+
with open("stats.json", "w") as file:
|
| 42 |
+
json.dump(output, file, indent=4)
|
| 43 |
+
|
| 44 |
+
logger.info("*** Uploading `stats.json` to Hugging Face Hub ***")
|
| 45 |
+
api = HfApi(token=os.getenv("HF_TOKEN"))
|
| 46 |
+
api.upload_file(
|
| 47 |
+
path_or_fileobj="stats.json",
|
| 48 |
+
path_in_repo="stats.json",
|
| 49 |
+
repo_id="DIBT/prompt-collective-dashboard",
|
| 50 |
+
repo_type="space",
|
| 51 |
+
)
|
| 52 |
+
logger.info("*** `stats.json` successfully uploaded to Hugging Face Hub! ***")
|