Chamorro (ch) subword embeddings

Vocab size vocab model 25 dim 50 dim 100 dim 200 dim 300 dim
1000 vocab model txt | bin
bokeh | umap | matrix
txt | bin
bokeh | umap | matrix
txt | bin
bokeh | umap | matrix
txt | bin
bokeh | umap | matrix
txt | bin
bokeh | umap | matrix
3000 vocab model txt | bin
bokeh | umap | matrix
txt | bin
bokeh | umap | matrix
txt | bin
bokeh | umap | matrix
txt | bin
bokeh | umap | matrix
txt | bin
bokeh | umap | matrix
5000 vocab model txt | bin
bokeh | umap | matrix
txt | bin
bokeh | umap | matrix
txt | bin
bokeh | umap | matrix
txt | bin
bokeh | umap | matrix
txt | bin
bokeh | umap | matrix

Embedding matrix plots

Training corpus sample, encoded with different BPE vocabulary sizes

Vocab sizechwiki sample
original * rogers, robert f (0000). destiny's landfall: a history of guam: university of hawai'i press. isbn 0-0000-0000-0
mafa'tinas kalendario ni meggai na sibilisasion yan sosiedat siha, (gi mayot i biahi) ginen otru na kalendario ya matulaika para nisisidat i taotao si
mientras afamao i nina'siña-ña gi åtmoniu gi lina'la'-ña, ti gus matungo' gue' kumu tituge' musika. sa' gus umusa gue' fotman baroque yan i kontrapunt
1000 ▁* ▁ro g ers , ▁ro ber t ▁f ▁(0000 ). ▁d est in y ' s ▁l and f all : ▁a ▁history ▁of ▁guam : ▁un ivers ity ▁of ▁ha w ai ' i ▁pr ess . ▁is b n ▁0 - 0000 - 0000 - 0
▁mafa ' tinas ▁kalendario ▁ni ▁meggai ▁na ▁si b il is asion ▁yan ▁s os ied at ▁siha , ▁( g i ▁ma y ot ▁i ▁b ia hi ) ▁ginen ▁otru ▁na ▁kalendario ▁ya ▁mat ula ika ▁para ▁n is is idat ▁i ▁taotao ▁si
▁m ient r as ▁af am ao ▁i ▁n ina ' s i ña - ña ▁gi ▁ åt m on i u ▁gi ▁l ina ' la '- ña , ▁ti ▁g us ▁mat ungo ' ▁gue ' ▁kum u ▁tit ug e ' ▁musika . ▁sa ' ▁g us ▁um usa ▁gue ' ▁f ot m an ▁b ar o que ▁yan ▁i ▁kont r ap un t
3000 ▁* ▁rogers , ▁robert ▁f ▁(0000). ▁dest iny ' s ▁landfall : ▁a ▁history ▁of ▁guam : ▁un iversity ▁of ▁hawai ' i ▁press . ▁isbn ▁0-0000 -0000-0
▁mafa ' tinas ▁kalendario ▁ni ▁meggai ▁na ▁si b ilis asion ▁yan ▁sosiedat ▁siha , ▁( gi ▁mayot ▁i ▁biahi ) ▁ginen ▁otru ▁na ▁kalendario ▁ya ▁mat ulaika ▁para ▁n is is idat ▁i ▁taotao ▁si
▁mient ras ▁af am ao ▁i ▁nina ' si ña - ña ▁gi ▁åtmoniu ▁gi ▁lina ' la '- ña , ▁ti ▁gus ▁matungo ' ▁gue ' ▁kumu ▁tituge ' ▁musika . ▁sa ' ▁gus ▁um usa ▁gue ' ▁fotman ▁baroque ▁yan ▁i ▁kont rap un t
5000 ▁* ▁rogers , ▁robert ▁f ▁(0000). ▁destiny ' s ▁landfall : ▁a ▁history ▁of ▁guam : ▁university ▁of ▁hawai ' i ▁press . ▁isbn ▁0-0000-0000-0
▁mafa ' tinas ▁kalendario ▁ni ▁meggai ▁na ▁si bilis asion ▁yan ▁sosiedat ▁siha , ▁( gi ▁mayot ▁i ▁biahi ) ▁ginen ▁otru ▁na ▁kalendario ▁ya ▁matulaika ▁para ▁nis is idat ▁i ▁taotao ▁si
▁mientras ▁af am ao ▁i ▁nina ' si ña - ña ▁gi ▁åtmoniu ▁gi ▁lina ' la '- ña , ▁ti ▁gus ▁matungo ' ▁gue ' ▁kumu ▁tituge ' ▁musika . ▁sa ' ▁gus ▁um usa ▁gue ' ▁fotman ▁baroque ▁yan ▁i ▁kont rap un t