Long text is sometimes truncated as a result of processing

[tseylerd]

Happens because we process text in chunks and if it's not split by sentences, the last sentence in each chunk might be skipped by AI model if it's not finished sentence.

[tseylerd]

I improved this in 1.1 by splitting sequence in chunks by sentence end characters.

However, if the text itself contains unfinished sentences, it might not help. Need to add such examples in training data and retrain the adapter.