Peter Baumgartner

It's not a hard dependency, but https://github.com/explosion/spaCy/pull/10994 is probably a prior PR we want before this because it fixes displaCy for spans. @svlandeg this is somewhat related, but maybe we...

@svlandeg This might be overkill and the beginning of a longer discussion about using `TypedDict` in more places across the repo (i.e. matcher rules), but I've gone ahead and pushed...

We're scoping this one down to providing documentation on the dispaCy data structures and parse functions.

Alright, this PR now just updates the documentation on using displacy's `parse` functions. This happens in two spots: 1. `usage/visualizers` - Updated to make reference to the now documenated `parse_{deps,ents,spans}`...

@fabianvf I've been working on a rewrite to clean things up, add functionality, and likely release to pypi. You seem to be the most active on this package, would you...

This is ready for another review. Here's the current output: ``` ℹ 2709 lemmatizer trees generated from training data. ℹ 565 lemmatizer trees generated from dev data. ⚠ 105 lemmatizer...

I ran the example with timing - on one submission, it completed in 5s, but on another captured below it took 20m. ``` %%time resp = client.chat.completions.create( model="mistral:instruct", messages=[ {...

Some more local testing, it responsds in a normal (

@jxnl I can take a look. Could you point me in the right place in the codebase to start? I've been digging through the codebase and I'm not totally clear...

@jxnl Thanks! So `instructor` isn't doing anything with `ollama`'s [JSON Output mode](https://github.com/ollama/ollama/blob/main/docs/api.md#request-json-mode)? It's just done through prompting?