Linking to paperetl/paperai pipelines

[jcalifornia]

Hi, trying to link to embeddings outputted using the paperai example. Noticing that in the Extractor.batchsearch function that the results of self.similarity.batchsearch([self.tokenize(x) for x in queries], self.context) are tuples rather than dictionaries.

I noticed the following comment on line 267 of txtai.pipeline.txt.extractor.py

# Assume embeddings content is enabled and results are dictionaries

Does this mean we need to make a modification to the paperai.Index step? Thanks in advance

[jcalifornia]

OK looks like I need to re-run paperai.Index with a config yaml where content: True is enabled... closing this issue for now.