Treinamento distribuído com o Keras avaliações

NACER E. · Revisado há almost 2 years

GIRISH KUMAR S. · Revisado há almost 2 years

Ahmad A. · Revisado há almost 2 years

Harsha V. · Revisado há almost 2 years

Dmytro K. · Revisado há almost 2 years

Sebastián A. · Revisado há almost 2 years

Leticia G. · Revisado há almost 2 years

Leticia G. · Revisado há almost 2 years

Scot J. · Revisado há almost 2 years

Solomon B. · Revisado há almost 2 years

James Z. · Revisado há almost 2 years

VU T. · Revisado há almost 2 years

Himal D. · Revisado há almost 2 years

David T. · Revisado há almost 2 years

Leticia G. · Revisado há almost 2 years

Leticia G. · Revisado há almost 2 years

Sándor Tamás K. · Revisado há almost 2 years

Srihari S. · Revisado há almost 2 years

Darshan P. · Revisado há almost 2 years

Sneha J. · Revisado há almost 2 years

Hadrián P. · Revisado há almost 2 years

Crystal E. · Revisado há almost 2 years

Kriti V. · Revisado há almost 2 years

Ruben R. · Revisado há almost 2 years

Several of the steps (such as saving in "tf" format) had not been shown before and were ambiguous, so I had to look several times at the solution. The output is cluttered with "Cleanup called...". It was nice to see that the distributed and undistributed versions of model.evaluate() give the same result. It would be even nicer to compare the speed of distributed vs. undistributed model.fit() when adding some GPU.

Matthias G. · Revisado há almost 2 years