give meaning to batchsize=0

Author: mcabbott

src/eachobs.jl

@@ -56,7 +56,8 @@ The original data is preserved in the `data` field of the DataLoader.
 - `data`: The data to be iterated over. The data type has to be supported by
   [`numobs`](@ref) and [`getobs`](@ref).
 - `batchsize`: If less than 0, iterates over individual observations.
-  Otherwise, each iteration (except possibly the last) yields a mini-batch
+  If 0, then one mini-batch containing all `numobs(x)` observations.
+  If larger than 0, each iteration (except possibly the last) yields a mini-batch
   containing `batchsize` observations. Default `1`.
 - `buffer`: If `buffer=true` and supported by the type of `data`,
   a buffer will be allocated and reused for memory efficiency.
@@ -149,6 +150,7 @@ function DataLoader(
     if !(collate ∈ (Val(nothing), Val(true), Val(false)))
         throw(ArgumentError("`collate` must be one of `nothing`, `true` or `false`."))
     end
+    batchsize = batchsize == 0 ? numobs(data) : batchsize
     return DataLoader(data, batchsize, buffer, partial, shuffle, parallel, collate, rng)
 end