3 ms·
With some brief experimentation ChatGPT also fails this test.
by User23 2y ago
With some brief experimentation ChatGPT also fails this test.
- lostmsu 2y agoIt might make sense: any kind of fine-tuning of LLMs usually reduces generalization capabilities, and instruction-tuning is a kind of fine-tuning.