Kara-Kalpak (kaa) 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 |
| 10000 | 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 |
Training corpus sample, encoded with different BPE vocabulary sizes
| Vocab size | kaawiki sample |
|---|---|
| original | * 0000: magia * 0000: peligro * 0000: pies descalzos, suenos blancos * 0000: version en italiano * 0000: pies descalzos * 0000: donde estan los ladron transnatsional korporatsiyalar (tnk) bmsh ekspertlerinin' pikirinshe "du'nya ekonomikası dvigateli" bolıp tabıladı. bunday etip bmshnın transnatsional file:khotanlik ulama in 0000, muhammad amin bughra wearing black in foreground.jpg |кaрa чaпaн кедимди м. эмин бугрa(0000) |
| 1000 | ▁* ▁0000 : ▁mag i a ▁* ▁0000 : ▁p e lig ro ▁* ▁0000 : ▁p i es ▁de s c al z os , ▁su en os ▁b lan c os ▁* ▁0000 : ▁v er sion ▁en ▁ it ali an o ▁* ▁0000 : ▁p i es ▁de s c al z os ▁* ▁0000 : ▁d on de ▁es tan ▁l os ▁ la d r on ▁t ran s n at sion al ▁kor por atsiy alar ▁( t n k ) ▁b m sh ▁ek s per t lerinin ' ▁pikir in she ▁" d u ' nya ▁ekonomik ası ▁d v ig at e li " ▁bolıp ▁tab ıladı . ▁bun day ▁et ip ▁b m sh nın ▁t ran s n at sion al ▁fi le : k h o tan lik ▁ul ama ▁in ▁0000 , ▁mu h am ma d ▁a m in ▁b ug h ra ▁w e ar ing ▁b la c k ▁in ▁for eg ro un d . jpg ▁ | к a р a ▁ ч a п a н ▁ к е д и м д и ▁ м . ▁ э м и н ▁ б у г р a ( 0000 ) |
| 3000 | ▁* ▁0000: ▁mag ia ▁* ▁0000: ▁pe lig ro ▁* ▁0000: ▁pi es ▁de s c al z os , ▁su en os ▁b lan c os ▁* ▁0000: ▁ver sion ▁en ▁it ali an o ▁* ▁0000: ▁pi es ▁de s c al z os ▁* ▁0000: ▁d on de ▁es tan ▁l os ▁la dr on ▁transnatsional ▁korpor atsiyalar ▁( t n k ) ▁b m sh ▁eks per t lerinin ' ▁pikir inshe ▁" du ' nya ▁ekonomik ası ▁dvigat eli " ▁bolıp ▁tabıladı . ▁bunday ▁etip ▁b m sh nın ▁transnatsional ▁file : k h o tan lik ▁ul ama ▁in ▁0000, ▁muh am ma d ▁a min ▁bug h ra ▁w e ar ing ▁b la ck ▁in ▁for eg ro und . jpg ▁| к a р a ▁ ч a п a н ▁к е д и м д и ▁м . ▁ э м и н ▁ б у г р a ( 0000) |
| 5000 | ▁* ▁0000: ▁mag ia ▁* ▁0000: ▁pe lig ro ▁* ▁0000: ▁pi es ▁des c al z os , ▁su en os ▁b lan c os ▁* ▁0000: ▁ver sion ▁en ▁it ali ano ▁* ▁0000: ▁pi es ▁des c al z os ▁* ▁0000: ▁don de ▁es tan ▁l os ▁la dr on ▁transnatsional ▁korporatsiyalar ▁( t nk ) ▁bmsh ▁eks per t lerinin ' ▁pikirinshe ▁" du ' nya ▁ekonomik ası ▁dvigat eli " ▁bolıp ▁tabıladı . ▁bunday ▁etip ▁bmsh nın ▁transnatsional ▁file : k ho tan lik ▁ul ama ▁in ▁0000, ▁muh am ma d ▁a min ▁bug h ra ▁w e ar ing ▁b la ck ▁in ▁for eg ro und . jpg ▁| к a р a ▁ ч a п a н ▁к е д и м д и ▁м . ▁ э м ин ▁б у г р a ( 0000) |
| 10000 | ▁* ▁0000: ▁mag ia ▁* ▁0000: ▁pe lig ro ▁* ▁0000: ▁pi es ▁des cal z os , ▁su enos ▁b lan c os ▁* ▁0000: ▁ver sion ▁en ▁itali ano ▁* ▁0000: ▁pi es ▁des cal z os ▁* ▁0000: ▁don de ▁es tan ▁los ▁la dr on ▁transnatsional ▁korporatsiyalar ▁( t nk ) ▁bmsh ▁eksper t lerinin ' ▁pikirinshe ▁" du ' nya ▁ekonomikası ▁dvigateli " ▁bolıp ▁tabıladı . ▁bunday ▁etip ▁bmsh nın ▁transnatsional ▁file : k ho tan lik ▁ulama ▁in ▁0000, ▁muhammad ▁a min ▁bug h ra ▁we ar ing ▁b lack ▁in ▁for eg ro und . jpg ▁| к a р a ▁ч a п a н ▁к е ди м ди ▁м . ▁э м ин ▁б у г р a ( 0000) |