Merge pull request apache#50 from kasiabozek/acc_callback

accuracy in callback apache#49

src/callback.jl

@@ -110,13 +110,13 @@ end
 function every_n_epoch(callback :: Function, n :: Int; call_on_0 :: Bool = false)
   EpochCallback(n, call_on_0, callback)
 end
-function Base.call(cb :: EpochCallback, model :: Any, state :: OptimizationState)
+function Base.call{T<:Real}(cb :: EpochCallback, model :: Any, state :: OptimizationState, metric :: Vector{Tuple{Base.Symbol, T}})
   if state.curr_epoch == 0
     if cb.call_on_0
-      cb.callback(model, state)
+      cb.callback(model, state, metric)
     end
   elseif state.curr_epoch % cb.frequency == 0
-    cb.callback(model, state)
+    cb.callback(model, state, metric)
   end
 end
 
@@ -136,7 +136,7 @@ end
 =#
 function do_checkpoint(prefix::AbstractString; frequency::Int=1, save_epoch_0=false)
   mkpath(dirname(prefix))
-  every_n_epoch(frequency, call_on_0=save_epoch_0) do model, state
+  every_n_epoch(frequency, call_on_0=save_epoch_0) do model, state, metric
     save_checkpoint(model, prefix, state)
   end
 end

src/model.jl

@@ -270,13 +270,14 @@ end
   callbacks   :: Vector{AbstractCallback} = AbstractCallback[],
 )
 
-function _invoke_callbacks(self::FeedForward, callbacks::Vector{AbstractCallback},
-                           state::OptimizationState, type_filter::Type)
+function _invoke_callbacks{T<:Real}(self::FeedForward, callbacks::Vector{AbstractCallback},
+                                    state::OptimizationState, type_filter::Type;
+                                    metric::Vector{Tuple{Base.Symbol, T}} = Vector{Tuple{Base.Symbol, Real}}())
   map(callbacks) do cb
     if isa(cb, type_filter)
       if type_filter == AbstractEpochCallback
         # epoch callback have extra access to the model object
-        cb(self, state)
+        cb(self, state, metric)
       else
         cb(state)
       end
@@ -465,9 +466,10 @@ function fit(self :: FeedForward, optimizer :: AbstractOptimizer, data :: Abstra
     end # end of one epoch
 
     time_stop = time()
+    metric = get(opts.eval_metric)
     info(format("== Epoch {1:0>3d} ==========", i_epoch))
     info("## Training summary")
-    for (name, value) in get(opts.eval_metric)
+    for (name, value) in metric
       info(format("{1:>18s} = {2:.4f}", string(name), value))
     end
     info(format("{1:>18s} = {2:.4f} seconds", "time", time_stop-time_start))
@@ -514,7 +516,7 @@ function fit(self :: FeedForward, optimizer :: AbstractOptimizer, data :: Abstra
         copy!(self.aux_params[name], aux_avg)
       end
     end
-    _invoke_callbacks(self, opts.callbacks, op_state, AbstractEpochCallback)
+    _invoke_callbacks(self, opts.callbacks, op_state, AbstractEpochCallback; metric=metric)
   end # end of all epochs
 end
 
@@ -573,4 +575,3 @@ function load_checkpoint(self :: FeedForward, prefix :: AbstractString, epoch ::
   self.aux_params = aux_params
   return self
 end
-