In Prodigy v1.14.0 we switched to radicli as the main dependency for our command-line interfaces. This change is fully backward compatible so all custom recipes written with plac syntax will work as before.
radiclibrings many DX improvements such as using type hints for argument parsing including support for custom types as well as custom CLI errors. We recommend checking the radicli documentation for a complete overview of benefits.
In this release we have also updated our dependencies. The users can now work with Pydantic v2 (<3.0), FastAPI <0.103.0 and spaCy-LLM <0.6.0. spacy-llm upgrade is especially exciting as it comes with the support for chain-of-though prompting. These prompts allow you to attach reasoning to your few-shot examples, which might boost the performance of your pipeline and it can now be used inside Prodigy for NER and spancat.
We have added a dedicated section to our guide on Large Language Models.
Finally, we have deprecates some of the older loaders and helpers i.e:
Redditdataset loaderread_jsonlwrite_jsonlread_jsonb64_uri_to_bytespretty_print_nerpretty_print_tc
See here for the full changelog: Changelog · Prodigy · An annotation tool for AI, Machine Learning & NLP
Finally, a reminder that you can get 10% off a new personal license, but also off any upgrade. Make sure you use PRODIGY10 as a discount code at checkout. The code is valid all through September!
Looking forward to hearing what you think 