text encoder training ignored on free colab

[dalmackay]

i know the text encoder is what was crashing ram for the free users and i noticed it still tries to do it but now just skips it if it fails. Will the text encoder be optimized in the future to work for free users? and in the meantime will the models turn out okay without the text encoding? will character or style training suffer the most from lack of text encoder?

[TheLastBen]

there was a problem with the dependencies, now it's fixed