David Huggins-Daines
David Huggins-Daines
This example code, as seen in https://github.com/katspaugh/wavesurfer.js/blob/master/src/plugin/spectrogram/index.js#L42, but also in all the other plugins, does not work: ```javascript * // es6 * import SpectrogramPlugin from 'wavesurfer.spectrogram.js'; * * // commonjs...
This behaviour (seen at https://github.com/huggingface/tokenizers/blob/master/tokenizers/src/decoders/wordpiece.rs#L40) is quite unexpected (even though I realize tokenization is not invertible): ``` In [34]: tok = transformers.AutoTokenizer.from_pretrained('distilbert-base-uncased') In [35]: tok.convert_tokens_to_string(tok.convert_ids_to_tokens(tok("I do not like green eggs...
I maintain that silently removing frames from the input in sphinx_fe is a super bad idea, but it seems that all of the public models out there were trained with...
It's difficult to understand which version of the Python module is the most functional and least buggy, but a good place to start would be whatever is used by https://github.com/Uberi/speech_recognition...
Merge with 0.8 branch if possible as well as suggestions from #125
There are not any unit or regression tests or CI for FIXED_POINT at the moment.
Same problem as #1938 and #1939, though it isn't due to Python 3.10 but rather the most recent versions of pip and setuptools in Python 3.8, I think. With pip...
The register_m2m() function is hard coded to use the Category class. Is there any other way to do this?
Hi! I've already added a few missing functions to the SWIG API (in particular, the FSG API was useless without the ability to set the final state index!). However they...
In theory, semi-continuous or PTM models are supposed to be fast! But training them is incredibly slow, especially the initial flat start. This is most likely due to some redundant...