Damien Daspit
Damien Daspit
First off, I just want to say that I appreciate the work that you have done on Thot. The incremental training and interactive machine translation features are invaluable. I certainly...
That sounds good to me. I might be able to work on it at some point and submit it as a pull request. Thank you for keeping it as a...
After further research, I believe that it is intended that `PretrainedConfig.decoder_start_token_id` should be used to insert the lang code at the beginning of the sentence during fine tuning of NLLB...
The "research" I was referring to was entirely about how to properly use HuggingFace Transformers to prefix the target sentences with the lang code during decoding. I was not aware...
Awesome, thanks for fixing this issue.
Unfortunately, I am not aware of any comparisons between the approaches. The primary cognate identification approach that Cog implements is called the [Blair method](http://surveywiki.info/index.php/Blair_Method). I have always wanted to include...
Here is a suggestion I received from a user regarding this issue: > I was just thinking about the Cog feature to define segments as similar. > Presently this is...
Thanks for all of the great feedback. This is a similar issue to problems you are having with editing words. The expectation is that the user will be importing their...
Makes sense. I will put it on my to do list.
Could it have something to do with syllabification? Tone letters do affect syllabification. Cog uses them as syllable breaks. The tone diacritics do not have this effect.