Sean MacAvaney

Awesome! What's the corpus download process like? We can handle the case like we do for other licensed datasets: provide instructions in the software, and ask them to link the...

As far as the structure goes -- can you clarify if the dataset is a typical clueweb22 split, or a special subset for ikat? If the former, we have a...

I realized that we already have an agreement for cw22, so I can request a copy and check :)

Hey @cadurosar -- I'm curious about the "Reranker Type" label in the table above. Are you using these all as re-rankers? If so, is that due to pooling bias or...

No worries, thanks for the clarification!

@iain-mackie can you look into this? @Karoljv in the meantime, here's a one-time transfer using [magic-wormhole](https://github.com/magic-wormhole/magic-wormhole): ```bash pip install magic-wormhole wormhole receive 5-paragon-quota ```

It looks like the license was added in https://github.com/cvangysel/pytrec_eval/commit/b8358d417dc9dbfee1af42e93c383b028105ec9f. cc: @cvangysel

Hey @mrdrozdov -- [trec_eval itself checks for duplicate documents and raises an error](https://github.com/usnistgov/trec_eval/blob/main/form_res_rels.c#L129) if it finds any. So I'm not sure diverging from this behavior in the python wrapper would...

A few things I want to update with fresh eyes: 1. I think `ComposedPipeline` should first call `compile` on each consistent transformer. 2. If the result of a fuse_left or...

Right, what I was thinking is that it needs to expand them out in the case where compile gives another ComposedPipeline. Otherwise the example won't work. I think something like...