2 ms·
It might make sense: any kind of fine-tuning of LLMs usually reduces generalization capabilities, and instruction-tuning is a kind of fine-tuning.
by lostmsu 2y ago
It might make sense: any kind of fine-tuning of LLMs usually reduces generalization capabilities, and instruction-tuning is a kind of fine-tuning.