Description

The documentation for sklearn.manifold.TSNE on the perplexity parameter has the following sentences:

Consider selecting a value between 5 and 50. The choice is not extremely critical since t-SNE is quite insensitive to this parameter.

Given personal experimentation with the parameter, sklearn docs examining how perplexity impacts shape, and the article here, isn't it misleading to claim that t-SNE is insensitive to the perplexity parameter value? Should it instead be changed to something like:

Consider selecting a value between 5 and 50. While the choice is not extremely critical, results have been shown to change considerably across values.

?