# Vak version 0.5.0 released!

**URL:** <https://forum.vocalpy.org/t/vak-version-0-5-0-released/35>\
**Category:** Announcements\
**Tags:** vak\
**Created:** [June 30, 2022, 2:49am UTC](https://forum.vocalpy.org/t/vak-version-0-5-0-released/35 "2022-06-30T02:49:59Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![nicholdav](https://yyz2.discourse-cdn.com/flex032/user_avatar/forum.vocalpy.org/nicholdav/32/5_2.png) [@nicholdav](https://forum.vocalpy.org/u/nicholdav)\
**Post date:** [June 30, 2022, 2:49am UTC](https://forum.vocalpy.org/t/vak-version-0-5-0-released/35/1 "2022-06-30T02:49:59Z")

</div>

Hi all,

Just a heads up that `vak` version 0.5.0 is released:

> <https://github.com/vocalpy/vak/blob/main/doc/CHANGELOG.md#050----2022-06-25>

You should be able to `pip install` or get it through the `conda-forge` channel now.

Big addition here is the ability to continue training models, contributed by @kaiyaprovost and based on work in [their pre-print](https://www.biorxiv.org/content/10.1101/2022.02.24.481827v3). Thank you Kaiya! 🎉

This is also the first time I’ve worked with someone as a maintainer on a sizable code contribution to a project 🥲  
Gory detail in the pull request here if you want to read:

> <https://github.com/vocalpy/vak/pull/504>
>
> ENH: Add "train\_checkpoint module" based off of "train" module and "predict" mod…ule. Copied main code of "train" module while updating core, cli, config. Added code from "predict" to import an existing checkpoint from a previous run of a model, rather than initializing a new model without any training.

There are also some doc fixes – thanks to @YMK123 for helping catch issues with how we explained the expected naming conventions for annotation files:

> **[How do I use my own vocal annotation format?](https://vak.readthedocs.io/en/latest/howto/howto_user_annot.html#explanation-of-when-you-can-use-the-simple-seq-format)**
>
> To load annotation formats, vak uses another Python package called crowsetta ( https://crowsetta.readthedocs.io/en/latest/). It has built-in support for common formats such as the .TextGrid files g...

Based on other issues I think this still might be a point of confusion, see e.g. [DOC: Better document naming convention · Issue #524 · vocalpy/vak · GitHub](https://github.com/vocalpy/vak/issues/524)  
The long term plan is to make it easier to build datasets by letting people specify whatever convention they want  
[ENH: add `pair` function with `key` argument · Issue #2 · vocalpy/vocles · GitHub](https://github.com/vocalpy/vocles/issues/2)  
(I know the wording in this issue is only clear to me at the moment so you just have to take my word that’s the plan 😛 )

Thanks all for contributing and please feel free to give feedback when you test it out!