GPT2 surprisals are very wrong

cpllab / lm-zoo

Easy black-box access to state-of-the-art language models

https://cpllab.github.io/lm-zoo/

MIT License

14 stars 6 forks source link

GPT2 surprisals are very wrong #31

Closed hans closed 4 years ago

hans commented 4 years ago

sentence_id     token_id        token   surprisal
1       1       This    0.000000
1       2       Ġis     -0.000000
1       3       Ġa      94.981335
1       4       Ġtest   101.192016
1       5       Ġsentence       72.894987
1       6       .       102.022926