She said it was amazingly beautiful and would have been really moved by it had it not been written by an AI. I counter-argued that human curation of AI generated text was a perfectly valid expression of love. I lost :-|.<p>The model generates results that are formatted like love letters with reasonably structured phrases. Verbatim reproduction of training content seems very low. But the sentences are often nonsensical and there's not much context preservation apart from some repetition.<p>However it does generate some quite interesting original phrases:
" Mon âme a moins de conversations que le désir "
" C'est une joie bien trop violente pour moi "
" vous [êtes] comme l'eau d'un pur amour "<p>These phrases can be cherry-picked based on the feeling you're trying to convey and used to compose unique letters influenced by the best french writers in history.<p>Technically: an OpenAI GPT-2 model (117/124M parameters) trained on around 1Gb of french classical litterature was fine-tuned on 300Kb of french love poetry and love songs.<p>We used GPT-2-Simple <a href="https://github.com/minimaxir/gpt-2-simple" rel="nofollow">https://github.com/minimaxir/gpt-2-simple</a> and GPT-2-Cloud-Run <a href="https://github.com/minimaxir/gpt-2-cloud-run" rel="nofollow">https://github.com/minimaxir/gpt-2-cloud-run</a> by Max Woolf.<p>The base french model (GPT2-French) was trained by William Jacques: <a href="https://github.com/aquadzn/gpt2-french" rel="nofollow">https://github.com/aquadzn/gpt2-french</a>. We used his "romans" model which is trained using text from <a href="https://www.bibebook.com/download" rel="nofollow">https://www.bibebook.com/download</a>, which looks like a compilation of project Gutenberg french novels. His model is trained over 5000 steps.<p>Fine-tuning was done using Max Woolf's Colab notebook: <a href="https://colab.research.google.com/drive/1VLG8e7YSEwypxU-noRNhsv5dW4NfTGce" rel="nofollow">https://colab.research.google.com/drive/1VLG8e7YSEwypxU-noRN...</a> and lasted only 600 steps (less than 30 minutes). Less steps and the content wasn't great, more and it started to reproduce training content verbatim.