ValueError: 77% of the tokens in adata.var[feature_name] are not in vocab.

[siyao-zhu]

I am confused that when following the instructions to Build Training Cell Corpus from Cellxgene Census and in last step, build the scb. I got errors ValueError: 77% of the tokens in adata.var[feature_name] are not in vocab. no matter how I change the vocab json. Does anyone face the same issue?

[subercui]

Did you check whether your adata.var[feature_name] matches the existing vocab, as mentioned here #107 (comment) ? Or do you mean you specifically want to use your own vocab?