TensorFlow.jl
TensorFlow.jl copied to clipboard
TF2 and incongruence with TF Dev summit message?
I was watching your dev summit talk and you mentioned that the design was mostly eager, but in the examples on this repo it seems you make use of TF1 functional/graph mode extensively?
Also, the last commit was almost a year ago. Is this direction a dead end and is there something more relevant we should look at?
Overall, if not in active development, it might not be a bad idea to declare the project as such in the main readme.md?
I was watching your dev summit talk and you mentioned that the design was mostly eager, but in the examples on this repo it seems you make use of TF1 functional/graph mode extensively?
None of the examples have been updated to use eager mode. Eager mode works.
is there something more relevant we should look at?
Flux is really great, I'm writing mostly Flux code these days.