Cornish (kw) 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 sizekwwiki sample
original {| cellpadding="0" cellspacing="0" border="0" style="border: #cccccc solid 0px; border-collapse: collapse;"
yn 0000 y feu drehevys an kottha rann a'n chi a rowan; mes tan yn 0000 a dhistruyas an brassa darn. an perghenn, an 0ves isyurl clifden, a ros an chi
angwilla - aruba - bermuda - bonaire - curaçao - ynysow faroe (enesow faroe) - greunland - gwadeloup - jibraltar - ynysow kayman (enys kayman) - ynys
1000 ▁{| ▁cellpadding ="0" ▁cellspacing ="0" ▁border ="0" ▁style =" b or der : ▁# c c c c c c ▁solid ▁0 px ; ▁border - collapse : ▁collapse ;"
▁yn ▁0000 ▁y ▁f eu ▁dre hev ys ▁an ▁k ot tha ▁rann ▁a ' n ▁ch i ▁a ▁r ow an ; ▁mes ▁t an ▁yn ▁0000 ▁a ▁dh ist r u yas ▁an ▁brassa ▁d arn . ▁an ▁p ergh enn , ▁an ▁0 ves ▁is y ur l ▁c li f d en , ▁a ▁ro s ▁an ▁ch i
▁an gw il la ▁- ▁ar ub a ▁- ▁ber m u da ▁- ▁b on a ire ▁- ▁c u ra ç a o ▁- ▁ynys ow ▁f ar o e ▁( en es ow ▁f ar o e ) ▁- ▁gre un land ▁- ▁gw ad el ou p ▁- ▁j ib ra l t ar ▁- ▁ynys ow ▁k ay man ▁( en ys ▁k ay man ) ▁- ▁ynys
3000 ▁{| ▁cellpadding ="0" ▁cellspacing ="0" ▁border ="0" ▁style =" b or der : ▁# c c c c c c ▁solid ▁0 px ; ▁border - collapse : ▁collapse ;"
▁yn ▁0000 ▁y ▁feu ▁drehevys ▁an ▁kottha ▁rann ▁a ' n ▁chi ▁a ▁row an ; ▁mes ▁tan ▁yn ▁0000 ▁a ▁dh ist ru yas ▁an ▁brassa ▁d arn . ▁an ▁p ergh enn , ▁an ▁0 ves ▁is y url ▁c li f den , ▁a ▁ros ▁an ▁chi
▁an gw il la ▁- ▁ar uba ▁- ▁ber m u da ▁- ▁bon a ire ▁- ▁c u ra ç a o ▁- ▁ynysow ▁far o e ▁( en esow ▁far o e ) ▁- ▁gre un land ▁- ▁gw adel ou p ▁- ▁j ib ral t ar ▁- ▁ynysow ▁kay man ▁( enys ▁kay man ) ▁- ▁ynys
5000 ▁{| ▁cellpadding ="0" ▁cellspacing ="0" ▁border ="0" ▁style =" bor der : ▁# cc cc cc ▁solid ▁0 px ; ▁border - collapse : ▁collapse ;"
▁yn ▁0000 ▁y ▁feu ▁drehevys ▁an ▁kottha ▁rann ▁a ' n ▁chi ▁a ▁row an ; ▁mes ▁tan ▁yn ▁0000 ▁a ▁dh istru yas ▁an ▁brassa ▁darn . ▁an ▁pergh enn , ▁an ▁0 ves ▁is y url ▁cli f den , ▁a ▁ros ▁an ▁chi
▁an gw illa ▁- ▁ar uba ▁- ▁ber mu da ▁- ▁bon a ire ▁- ▁c ura ç a o ▁- ▁ynysow ▁far oe ▁( en esow ▁far oe ) ▁- ▁greunland ▁- ▁gw adel oup ▁- ▁j ib ral tar ▁- ▁ynysow ▁kay man ▁( enys ▁kay man ) ▁- ▁ynys
10000 ▁{| ▁cellpadding ="0" ▁cellspacing ="0" ▁border ="0" ▁style =" bor der : ▁# cc cc cc ▁solid ▁0 px ; ▁border - collapse : ▁collapse ;"
▁yn ▁0000 ▁y ▁feu ▁drehevys ▁an ▁kottha ▁rann ▁a ' n ▁chi ▁a ▁row an ; ▁mes ▁tan ▁yn ▁0000 ▁a ▁dh istru yas ▁an ▁brassa ▁darn . ▁an ▁pergh enn , ▁an ▁0 ves ▁is yurl ▁cli f den , ▁a ▁ros ▁an ▁chi
▁an gw illa ▁- ▁aruba ▁- ▁ber mu da ▁- ▁bon aire ▁- ▁c ura ç ao ▁- ▁ynysow ▁faroe ▁( enesow ▁faroe ) ▁- ▁greunland ▁- ▁gwadeloup ▁- ▁jib raltar ▁- ▁ynysow ▁kayman ▁( enys ▁kayman ) ▁- ▁ynys