Atikamekw (atj) 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 sizeatjwiki sample
original micta notcimiw awesis ka miretc, micta piresiw ka micta kinwaskokanetc, ka kinwaskokotetc, ka kinwaskokowetc, rarewak ka nta ici pa nanto mitcisotc.
00 otatakon pisimw ka nitawikitc ote itekera la tuque ka 0000 tatopiponikarik. edmond ottawa ki icinikasoriwa otatama. maggie dubé tca otcotcoma. ki m
enkote e isparik mitcim e kotcipiritcikatek. nipi takon neta enko tca e tci ickwatek mitcim minawatc wepinikatew. okitc omitciwocik actew oskon. nawec
1000 ▁micta ▁notc im iw ▁awesis ▁ka ▁mir etc , ▁micta ▁piresiw ▁ka ▁micta ▁kin was kok an etc , ▁ka ▁kin was ko ko tetc , ▁ka ▁kin was ko ko wetc , ▁rar ew ak ▁ka ▁nta ▁ici ▁pa ▁nan to ▁mitc isotc .
▁00 ▁ota takon ▁pisimw ▁ka ▁nitaw ikitc ▁ote ▁itekera ▁la ▁t u que ▁ka ▁0000 ▁tato p ipon ikarik . ▁e d m on d ▁ottawa ▁ki ▁icinikasoriwa ▁ota ta ma . ▁ma g g ie ▁du bé ▁tca ▁otc otc o ma . ▁ki ▁m
▁enko te ▁e ▁is parik ▁mitcim ▁e ▁kotc ip ir itcikatek . ▁nipi ▁takon ▁neta ▁enko ▁tca ▁e ▁tc i ▁ickwa tek ▁mitcim ▁minawatc ▁we pin ikatew . ▁okitc ▁om itc iw oc ik ▁actew ▁os kon . ▁na wec
3000 ▁micta ▁notcimiw ▁awesis ▁ka ▁miretc , ▁micta ▁piresiw ▁ka ▁micta ▁kinwas kok an etc , ▁ka ▁kinwasko ko tetc , ▁ka ▁kinwasko ko wetc , ▁rarewak ▁ka ▁nta ▁ici ▁pa ▁nanto ▁mitcisotc .
▁00 ▁otatakon ▁pisimw ▁ka ▁nitaw ikitc ▁ote ▁itekera ▁la ▁tuque ▁ka ▁0000 ▁tatopipon ikarik . ▁e d mon d ▁ottawa ▁ki ▁icinikasoriwa ▁otatama . ▁maggie ▁dubé ▁tca ▁otcotc oma . ▁ki ▁m
▁enko te ▁e ▁isparik ▁mitcim ▁e ▁kotc ip ir itcikatek . ▁nipi ▁takon ▁neta ▁enko ▁tca ▁e ▁tci ▁ickwa tek ▁mitcim ▁minawatc ▁we pin ikatew . ▁okitc ▁omitc iw ocik ▁actew ▁oskon . ▁na wec
5000 ▁micta ▁notcimiw ▁awesis ▁ka ▁miretc , ▁micta ▁piresiw ▁ka ▁micta ▁kinwaskok anetc , ▁ka ▁kinwaskoko tetc , ▁ka ▁kinwaskoko wetc , ▁rarewak ▁ka ▁nta ▁ici ▁pa ▁nanto ▁mitcisotc .
▁00 ▁otatakon ▁pisimw ▁ka ▁nitaw ikitc ▁ote ▁itekera ▁la ▁tuque ▁ka ▁0000 ▁tatopipon ikarik . ▁edmond ▁ottawa ▁ki ▁icinikasoriwa ▁otatama . ▁maggie ▁dubé ▁tca ▁otcotcoma . ▁ki ▁m
▁enko te ▁e ▁isparik ▁mitcim ▁e ▁kotc ipir itcikatek . ▁nipi ▁takon ▁neta ▁enko ▁tca ▁e ▁tci ▁ickwa tek ▁mitcim ▁minawatc ▁we pin ikatew . ▁okitc ▁omitc iwocik ▁actew ▁oskon . ▁nawec