question on random result

[HPicatto]

Hi I've data simmilar to the axiv one, I've reduce text to embedding and then reduce that array using UMAP. but the result is only approximatly the same given same input.
for example, in this case the text was only the word "html" the embedding produce identical arrays, but umap produced never the same values:

src_url_surt	extracted_text	umap_1	umap_2	embedding
"au,net,daedalus)/about"	"html"	7.922148	21.642248	[-0.054749, 0.011139, … 0.021515]
"au,org,galaxyfoundation)/"	"html"	8.211867	21.60751	[-0.054749, 0.011139, … 0.021515]
"au,org,makerheals)/"	"html"	8.094625	21.699774	[-0.054749, 0.011139, … 0.021515]
"au,org,thehut)/"	"html"	8.280423	21.611532	[-0.054749, 0.011139, … 0.021515]
"cat,geieg,reserves:8800)/geieg"	"html"	7.953833	21.644217	[-0.054749, 0.011139, … 0.021515]
…	…	…	…	…
"com,chillyfillmores)/"	"html"	8.031129	21.586952	[-0.054749, 0.011139, … 0.021515]
"com,chinasystems)/"	"html"	7.906438	21.706869	[-0.054749, 0.011139, … 0.021515]
"com,chinaums,open)/saas-web"	"html"	7.969978	21.430769	[-0.054749, 0.011139, … 0.021515]
"com,chinayawei)/"	"html"	8.02914	21.780499	[-0.054749, 0.011139, … 0.021515]
"com,pixiteapps)/"	"html"	8.171796	21.647501	[-0.054749, 0.011139, … 0.021515]

is there some kind of stochastic optimization in this packege?