ReAct Prompting

How does ReAct prompting differ from self$-$consistency?

ReAct prompting involves recursively decomposing a complex problem i

nto sub$-$problems, whereas self$-$consistency involves sampling multiple,

diverse reasoning paths through a few$-$shot chain$-$of$-$thought technique.

ReAct prompting involves fact$-$checking the accuracy of the model's pre

dictions, whereas self$-$consistency involves optimising the model's hyper

parameters.

ReAct prompting involves using interleaved 'thoughts$-$actions$-$observatio

ns$\text{'}$ to mimic human thinking, whereas self$-$consistency involves sampling

multiple, diverse reasoning paths to choose the most consistent answer.

ReAct prompting involves using interleaved 'thoughts$-$actions$-$observatio

ns$\text{'}$ to mimic human thinking, whereas self$-$consistency involves recursive

ly decomposing a complex problem into sub$-$problems.