Promptedgraphs

Custom tracking of apis
python
spacy_usage = Usage(model='spacy')
spacy_usage.start()
ents_spacy = await extract_entities_spacy....
spacy_usage.end()
print(spacy_usage)

Can compare api costs and durations across multiple models


Enables the generation of structured data. We can brainstorm thousands of samples of arbitrary datatypes.

The way to display the results has improved with a cleaner wrapper of the spacy.displacy functionality

Added the ability for nested data structures to be extracted from text and returned in Pydantic DataModels

Implemented a clean way to extract typed objects (via Pydantic) from text.

`data = extract_data(text=msg, output_type=list[UserIntent])`

Initial release
only has entity extraction implemented