Nauru (na) 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 sizenawiki sample
original adolf hitler - temoniba t djermani - ''führer und reichskanzler des deutschen volkes'', 0000-0000 * pudu 00.0, 0000, ian braunau am inn, oteriya-ungar
pagani zonda c00 s pagani zonda c00 s ''(bet pagani zonda s)'' wono superauto atsiu pagani (itari). wei eobweni wono 0000.
* bujumbura (000,000 engame) * gitega (00,000 engame) * muyinga (00,000 engame) * ngozi (00,000 engame) * ruyigi (00,000 engame)
1000 ▁ad ol f ▁hitler ▁- ▁t em on ib a ▁t ▁djermani ▁- ▁'' f ü h r er ▁ und ▁r ei ch sk an z ler ▁d es ▁de ut sch en ▁v ol k es '', ▁0000 -0000 ▁* ▁pudu ▁00.0, ▁0000, ▁ian ▁b ra un au ▁am ▁in n , ▁o ter iya - ung ar
▁pagani ▁zonda ▁c 00 ▁s ▁pagani ▁zonda ▁c 00 ▁s ▁'' ( bet ▁pagani ▁zonda ▁s )'' ▁wono ▁s up era uto ▁atsiu ▁pagani ▁( itari ). ▁wei ▁eobweni ▁wono ▁0000.
▁* ▁bu j um bur a ▁(000 ,000 ▁engame ) ▁* ▁g it eg a ▁(00 ,000 ▁engame ) ▁* ▁m u y ing a ▁(00 ,000 ▁engame ) ▁* ▁n go zi ▁(00 ,000 ▁engame ) ▁* ▁r u y ig i ▁(00 ,000 ▁engame )
3000 ▁adolf ▁hitler ▁- ▁temoniba ▁t ▁djermani ▁- ▁'' f ü hr er ▁ und ▁r ei ch sk an z ler ▁des ▁de utsch en ▁vol k es '', ▁0000-0000 ▁* ▁pudu ▁00.0, ▁0000, ▁ian ▁braun au ▁am ▁in n , ▁oter iya - ung ar
▁pagani ▁zonda ▁c 00 ▁s ▁pagani ▁zonda ▁c 00 ▁s ▁''( bet ▁pagani ▁zonda ▁s )'' ▁wono ▁superauto ▁atsiu ▁pagani ▁( itari ). ▁wei ▁eobweni ▁wono ▁0000.
▁* ▁bu j um bur a ▁(000 ,000 ▁engame ) ▁* ▁g it ega ▁(00,000 ▁engame ) ▁* ▁mu y inga ▁(00,000 ▁engame ) ▁* ▁n go zi ▁(00,000 ▁engame ) ▁* ▁ru y ig i ▁(00,000 ▁engame )
5000 ▁adolf ▁hitler ▁- ▁temoniba ▁t ▁djermani ▁- ▁'' f ü hr er ▁und ▁r ei ch sk an z ler ▁des ▁de utsch en ▁vol kes '', ▁0000-0000 ▁* ▁pudu ▁00.0, ▁0000, ▁ian ▁braun au ▁am ▁in n , ▁oteriya - ung ar
▁pagani ▁zonda ▁c 00 ▁s ▁pagani ▁zonda ▁c 00 ▁s ▁''( bet ▁pagani ▁zonda ▁s )'' ▁wono ▁superauto ▁atsiu ▁pagani ▁( itari ). ▁wei ▁eobweni ▁wono ▁0000.
▁* ▁bu j um bur a ▁(000,000 ▁engame ) ▁* ▁g it ega ▁(00,000 ▁engame ) ▁* ▁mu y inga ▁(00,000 ▁engame ) ▁* ▁n go zi ▁(00,000 ▁engame ) ▁* ▁ru y igi ▁(00,000 ▁engame )