use batching api

src/backend/tensorflow/model.jl

@@ -17,14 +17,21 @@ function tf(model)
         [gradients(output, input)])
 end
 
-batch(x) = Batch((x,))
+batchone(x) = Batch((x,))
+
+function batch(xs)
+  dims = ndims(xs)-1
+  T = Array{eltype(xs),dims}
+  B = Array{eltype(xs),dims+1}
+  Batch{T,B}(xs)
+end
 
 function (m::Model)(args::Batch...)
   @assert length(args) == length(m.inputs)
-  run(m.session, m.outputs[1], Dict(zip(m.inputs, args)))
+  batch(run(m.session, m.outputs[1], Dict(zip(m.inputs, args))))
 end
 
-(m::Model)(args...) = m(map(batch, args)...)
+(m::Model)(args...) = first(m(map(batchone, args)...))
 
 function Flux.back!(m::Model, Δ, args...)
   @assert length(args) == length(m.inputs)
@@ -49,7 +56,7 @@ function Flux.train!(m::Model, train, test=[]; epoch = 1, η = 0.1,
     info("Epoch $e\n")
     @progress for (x, y) in train
       y, cur_loss, _ = run(m.session, vcat(m.outputs[1], Loss, minimize_op),
-                           Dict(m.inputs[1]=>batch(x), Y=>batch(y)))
+                           Dict(m.inputs[1]=>batchone(x), Y=>batchone(y)))
       if i % 5000 == 0
         @show y
         @show accuracy(m, test)