Add arxiv dataset to RelBench #295
Conversation
- Added `arxiv.py` for dataset processing - Registered dataset in `__init__.py` - Updated `hashes.json` with dataset hash - Dataset source: "https://www.dropbox.com/scl/fi/tjj6r1fqikt4j0rz4qomu/db.zip?rlkey=1ykfkp8pj3hu6n4utz8g9dkx2&st=azmm56dc&dl=1" - SHA256: ff9e03e467e28df959d08c79c453db1f31b525f07ff3c0e0b5e571e732acc63f
- Added `arxiv.py` for dataset processing - Registered dataset in `__init__.py` - Updated `hashes.json` with dataset hash
|
Thanks for the contribution @PKUTHM ! @kexinhuang12345 can you please upload this database and let me know so that I can merge it? Thanks |
|
Hi @PKUTHM, to maintain uniformity with RelBench, we would like to have some tasks defined on this database before we can merge the PR. Can you please contribute some associated tasks: at least one each of entity classification, entity regression and recommendation? You can refer to existing tasks for inspiration and use the example scripts to run RDL on your tasks. We have found that adding new tasks is fairly light coding work, but designing the task to be not to easy but still predictable reuires some iteration of design choices such as the task timedelta, etc. Thanks! And please let me know the expected timeline you have in mind for when you will be able to get this done? |
|
We will happily acknowledge your contribution on the RelBench website and in upcoming releases! |
Add arxiv dataset to RelBench