fix null serialization for empty hub stats fields

add arxiv papers export to hub stats job script

fix: relax pinned dependency versions for Python 3.13 compat

fix: write parquet with 50k row groups instead of single giant row group

add a sleep of 30 sec after each

Upload daily_papers_sync.py with huggingface_hub

Fix: Collect all data first then write parquet with unified schema

Fix: Cast parquet tables to unified schema for type consistency

Fix: Use HF_TOKEN from environment for HF Jobs compatibility

Update hub-stats.py

Update hub-stats.py

Create README.md

Update hub-stats.py

Update hub-stats.py

Update hub-stats.py

Update hub-stats.py

Create hub-stats.py

Delete classification.py

Delete 100-rows.py

Delete dedupe.py

Update 100-rows.py

Update 100-rows.py

Rename classification to classification.py

Create classification

Update 100-rows.py

Update 100-rows.py

Update 100-rows.py

Create 100-rows.py

Update dedupe.py

Create dedupe.py

Delete dedupe.py

Update dedupe.py

Create dedupe.py

initial commit