Datasets:
mteb
/

Modalities:
Text
Formats:
json
Languages:
English
Libraries:
Datasets
pandas

Not actually Climate FEVER

#1
by sasha HF staff - opened

Just flagging that this seems to be (a copy of) the original FEVER dataset, not the climate-relevant one (which can be found here: https://huggingface.co/datasets/tdiggelm/climate_fever/).

That's fine of course, but just changing the name or making this clearer would be helpful :)

Massive Text Embedding Benchmark org
Massive Text Embedding Benchmark org

The queries are the same as in the one you are mentioning no? https://huggingface.co/datasets/mteb/climate-fever/viewer/queries

I think it is just the corpus which is the same as FEVER (Wikipedia) but the queries are different.

Indeed, you're right! It's just unclear what the different splits of the dataset represent (including the 'default' one) -- maybe a README would clear this up?
Given that this version of the dataset has much more downloads than the original ClimateFEVER, it make it easier to understand for folks :)

πŸ‘‹ πŸ«‚

Massive Text Embedding Benchmark org

I have added an issue on the readme's here: https://github.com/embeddings-benchmark/mteb/issues/1892

We would be more than happy to welcome a PR on this, but currently it is not our main priority (documentation of MTEB tasks generally reside within the package). We would however to happy to maintain it once added.

Sign up or log in to comment