Max Nedelchev

Results 10 comments of Max Nedelchev

My variant. The model has been taught on russian dataset (common voice) ![image](https://user-images.githubusercontent.com/193212/66982392-a4bfda80-f0be-11e9-81e2-807ef14278ef.png)

@tkgmomosheep have you overcome your issue? I run into it too, but It happened after I replaced huge dataset with particular dataset with 180 files in it

yes, it helped. Reduced batch size from 32 to 16 and it started to work

look at `hyperparams.py` at the end of it just set new value `B = 16`

` fname, _, text = line.strip().split("|")` this is key line. It expects that line have format like `filename1|text 1 2 3|text one two three` if you have only two column...

Look at error again ``` fname, _, text = line.strip().split("|") ValueError: not enough values to unpack (expected 3, got 1) ``` it says that split results in array of one...

Build log for 2.7.2 ``` hecking for ruby... /usr/bin/ruby tool/config.guess already exists tool/config.sub already exists checking build system type... x86_64-pc-linux-gnu checking host system type... x86_64-pc-linux-gnu checking target system type... x86_64-pc-linux-gnu...

Such method of setting webhook I took from here https://github.com/solyaris/BOTServer/blob/master/lib/webhook.rb#L48 I found solution: `curl -F "url=https:///webhooks/telegram" -F "certificate=@

> Fixed the issue was not specifying sample size > > new model.KaldiRecognizer(48000); this helped me too. Thanks man