|
|
|
|
|
by davidzweig
1284 days ago
|
|
Instead of GPT, you can take sentences from Tatoeba, and analyse and index by vocabulary. 'PhrasePump' does this already. We glued together an NLP pipeline (Stanza etc.) to capture the lemmas in sentences. https://www.languagereactor.com/phrasepump |
|