Classification metrics can't handle a mix of binary and continuous targets

All your commented-out models are not classifiers but regression models, for which accuracy is meaningless.

You get the error because these regression models do not produce binary outcomes, but continuous (float) numbers (as all regression models do); so, when scikit-learn attempts to calculate the accuracy by comparing a binary number (true label) with a float (predicted value), it not unexpectedly gives an error. And this cause is clearly hinted at the error message itself:

Classification metrics can't handle a mix of binary and continuous target

Notice also that the accepted (and highly upvoted...!) answer in the question suggested at the first comment as a possible duplicate of yours is wrong; there, as here, the root cause is the use of accuracy in a LinearRegression model, which, as already said, is meaningless.