3 ms·
You really need to say how the text is being generated, even in broad strokes. The "trained on a large corpora of text available in the public domain" + a bit
by minimaxir 7y ago
You really need to say how the text is being generated, even in broad strokes.
The "trained on a large corpora of text available in the public domain" + a bit of generation testing implies it's a GPT-2-based model.
- applecrazy 7y agoBased on the responses to prompts in this thread, it does seem to be GPT-2 based.