fix: clean up inadvertent change in tf_t5
This was the beginnings of an attempt to address the test failure on this layer, and instead I backed out of making this layer keras-serializable at all ... so it was a mistake to commit this.
Showing
Please register or sign in to comment