7 ms·
And yet, ChatGPT can generate these strings. Somehow despite using the wrong loss function it still seems to work by simply absorbing more training data. http
by 33a 3y ago
And yet, ChatGPT can generate these strings. Somehow despite using the wrong loss function it still seems to work by simply absorbing more training data.
https://chat.openai.com/share/82509815-d418-43bb-95a3-348bd5c700a4 https://chat.openai.com/share/82509815-d418-43bb-95a3-348bd5...
It can also recognize them, albeit it tends to cheat by shelling out to python (which makes sense, since it tends to lose count on large strings just like a human...)
https://chat.openai.com/share/b106ca5f-409a-43db-bc02-21da86ab40ce https://chat.openai.com/share/b106ca5f-409a-43db-bc02-21da86...
- toxik 3y agoArguably the skill to generate a program to do this is a higher-order skill and much more impressive.
- yorwba 3y agoChatGPT incorrectly put a space between the as and bs, but if we let that slide, there's still the issue that the best trained model in the article got 77.3% of the first 1500 strings correct, i.e. even if ChatGPT performed exactly the same, you'd expect it to get a single example correct more often than not.
- canjobear 3y agoThe difference is that you told it the language you wanted to recognize. In the paper, they are trying to learn the language from example strings alone.