sdadas / polish-nlp-resources Goto Github PK

When i wanted to unpack zipped files, I got an error: error: invalid compressed data to inflate. I got this error only with these embeddings:

• fasttext_500_3_polish.zip
• fasttext_800_3_polish.zip
• word2vec_800_3_polish.zip

also got CRC error with file:

• fasttext_500_3_polish.bin.wv.vectors_vocab.npy

I used zip, 7z on linux and winrar on Windows to unpack so I think these files are packed incorrectly.

When I write word2vec.similar_by_word("moglbym") , it works fine
When I write word2vec.similar_by_word("mógłbym") I get then message delivered by keyedvectors.py

"word 'mógłbym' not in vocabulary"

How do you manage accents and non english letters ?

It is difficult to load these models on laptop. Thanks to the fact, that models were saved to standalone files, it is much faster to .load() model with mmap='r' argument via memory-mapping.

Great work and extensive resource!
Can you add the following resource which might be of interest of ASR community?
https://github.com/goodmike31/pl-asr-speech-data-survey

Could you please provide links to GDrive files that enable getting them by wget/curl, or host them somewhere else?

Google Drive makes programmatic downloading of uploaded files using wget/curl a huge pain in the ass, and if you overcame that, I'm sure more people would use your library.