voidful/albert_chinese_xxlarge

The model is a albert_chinese_xxlarge model from Google's GitHubconverted by huggingface's script . Since sentencepiece is not used in the modelyou have to call BertTokenizer instead of AlbertTokenizer !!! We can eval it using an example on MaskedLM . The result is 0.995713472366333 . The model was created using a model from a model on Google's Github . We can now use the model to test our results using a new version of the model . We have to use a new model to try and find out what we want to do with the new model . The new model is called AlbertForMaskedLM,