Request for Ukrainian pretrained model artifacts (tree, phones.txt) for custom HCLG graph

Hello Vosk team,

I’m working on building custom HCLG decoding graphs for Ukrainian using the pretrained Vosk acoustic model.

However, the public model release doesn’t include the context tree or phones.txt files, which are necessary to rebuild the decoding graph without retraining the AM.

Would it be possible to get access to these artifacts (or guidance on how to generate compatible resources) so we can extend the LM/graph side safely?

Thank you for your amazing work!