5fa1a76
1
2
LLMs that have been fine-tuned on instruction datasets, generally perform well on such "zero-shot" tasks.
