Pipe
Closed this issue · 0 comments
koaning commented
I notice that very often I end up writing code that is super similar. Especially when you're comparing two language backends. Suppose I have a spaCy backend and a conveRT backend.
p1 = lang_sp[text].transform(Umap(2)).plot_interactive("umap_0", "umap_1", annot=False).properties(height=200, width=200)
p2 = lang_rt[text].transform(Umap(2)).plot_interactive("umap_0", "umap_1", annot=False).properties(height=200, width=200)
p1 | p2
It would be much nicer if I could do;
def make_plot(embset):
return (embset
.transform(Umap(2))
.plot_interactive("umap_0", "umap_1", annot=False)
.properties(height=200, width=200))
p1 = lang_sp[text].pipe(make_plot)
p2 = lang_rt[text].pipe(make_plot)
p1 | p2
A proper .pipe()
should allow for less repetition.