Kinyarwanda (rw) 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 sizerwwiki sample
original ijambo noheli rikomoka ku ijambo ry’igifaransa “noël”, naryo ryakomotse mu kilatini “natalis” bivuze “amavuka”. mu cyongereza noheli ikaba “christmas”
(izina mu giturukiya : ''kastamonu '' ) n’umujyi wa turukiya, n’umurwa mukuru w’intara y’kastamonu. abaturage 00,000.
000px kirusiya (''русский язык'') ni ururimi rwa uburugc gcg cv'gh bujnksiya na belarusi. itegekongenga iso 000-0 rus.
1000 ▁i ja mbo ▁no he li ▁ri ko mo ka ▁ku ▁i ja mbo ▁ry ’ igi faransa ▁ “ n o ë l ” , ▁na ryo ▁rya ko mo tse ▁mu ▁ki la t ini ▁ “ na ta li s ” ▁bi vu ze ▁ “ ama vu ka ” . ▁mu ▁cy ongereza ▁no he li ▁i kaba ▁ “ ch ri st ma s ”
▁( izina ▁mu ▁gi turukiya ▁: ▁'' ka s ta m on u ▁'' ▁) ▁n ’ umu jyi ▁wa ▁tu rukiya , ▁n ’ umu rwa ▁mukuru ▁w ’ in tara ▁y ’ ka s ta m on u . ▁abaturage ▁00 ,000 .
▁000 p x ▁ki ru si ya ▁('' р у с с к и й ▁ я з ы к '') ▁ni ▁ururimi ▁rwa ▁ubu ru g c ▁g c g ▁c v ' g h ▁bu j n k si ya ▁na ▁be la ru si . ▁i tege kon genga ▁iso ▁000 - 0 ▁ru s .
3000 ▁ijambo ▁no he li ▁ri komoka ▁ku ▁ijambo ▁ry ’ igi faransa ▁“ no ë l ” , ▁na ryo ▁rya komotse ▁mu ▁kilatini ▁“ na ta lis ” ▁bi vuze ▁“ ama vuka ” . ▁mu ▁cyongereza ▁no he li ▁ikaba ▁“ ch ri st mas ”
▁( izina ▁mu ▁giturukiya ▁: ▁'' ka sta mon u ▁'' ▁) ▁n ’ umujyi ▁wa ▁turukiya , ▁n ’ umurwa ▁mukuru ▁w ’ intara ▁y ’ ka sta mon u . ▁abaturage ▁00,000 .
▁000 px ▁kiru siya ▁('' р у с с к и й ▁ я з ы к '') ▁ni ▁ururimi ▁rwa ▁ubu ru g c ▁g c g ▁c v ' gh ▁bu jn k siya ▁na ▁be la rusi . ▁itegekongenga ▁iso ▁000-0 ▁ru s .
5000 ▁ijambo ▁no heli ▁ri komoka ▁ku ▁ijambo ▁ry ’ igifaransa ▁“ no ë l ” , ▁na ryo ▁ryakomotse ▁mu ▁kilatini ▁“ na talis ” ▁bi vuze ▁“ ama vuka ”. ▁mu ▁cyongereza ▁no heli ▁ikaba ▁“ ch ri st mas ”
▁( izina ▁mu ▁giturukiya ▁: ▁'' ka sta mon u ▁'' ▁) ▁n ’ umujyi ▁wa ▁turukiya , ▁n ’ umurwa ▁mukuru ▁w ’ intara ▁y ’ ka sta mon u . ▁abaturage ▁00,000 .
▁000 px ▁kirusiya ▁('' р у с с ки й ▁я з ы к '') ▁ni ▁ururimi ▁rwa ▁ubu ru g c ▁g c g ▁c v ' gh ▁bu jn k siya ▁na ▁be la rusi . ▁itegekongenga ▁iso ▁000-0 ▁ru s .