Home › Claude › aifoundations › many_shot_in_context_learning › The implication for prompt design when shots are…
The implication for prompt design when shots are abundant is
AQuantity matters AND order still matters
BOrder ceases to matter once enough shots are added
CHand-written exemplars become unnecessary always
DModel-generated rationales always beat human-written
Answer & Solution
Correct answer: A. Quantity matters AND order still matters
Both quantity (more usually helps, with task-specific peaks) and order (still affects accuracy in many-shot) remain live design choices. Practitioners must validate both rather than assume scale resolves them.
Related questions
An architectural risk of many-shot ICL exposed by the XSum result isWhy is many-shot ICL practically more attractive than fine-tuning?Long-context scaling laws (next-token loss)Many-shot ICL can overcomeMany-shot ICL can perform comparably toThe order of examples in many-shot promptsAbstractive summarization on XSumMachine translation many-shot results