* modify existing autodiff unit tests to test fprop cache

* cleanup

* fix compile error introduced with bad merge

* remove invalid negative/negative backwards power test

* bprop for avg pool

remove debug statements + formatting

* fix CPU test failures

* numeric tests

* use make_shared; unprotect c-tor

* fix convlution reference script

* convolution backprop

* cleanup

* fix build warnings

* Missing include

* fix build warning part 2

* move numeric_compare to its own header
code review feedback

* fix build warnings 3

* fix build warnings 4

* clang-format

* cast to avoid implicit cast warning

* wip

* wip

* remove get_vector from runtime::TensorView class as it was for unit test only

* cleanup

* move writting vector to runtime::TensorView to the unit test dir

* merge fix

* PR review change

* update from PR comment

* update changes file

* Function can have multiple results
Remove external use of ValueType, TupleType, Tuple
Remove many external uses of Output and Input

* corresponding CPU backend changes

* Update master changes.

* Remove type arg from Function, add changes.md

* Merge changes.

* Move bodies to .cpp, add brief doc

* Merge CPU changes.

* Remove xla includes from non-xla files

* Remove xla from tests

* First part of xla tuple support

* change fprop_cache to assume multi-output bprop functions

* New wrappers for handling tuples with XLA

* Review comments

* remove old xla files

* fix merge errors

* hand edit models to use multi output instead of tuples

* remove ngvm

* remove NGVM from cmake

* remove ParameterizedConstant
* use simpler element Type definition
* Move TraitedType to NGVM directory

* WIP generalized dot

* Add some multi-axis 3D, 4D, and 5D tests

* Add test on some 'pretty big' tensors

* Reworked dot to have less flexible axis-pairing behavior

* Backprop for dot... and a fix for a dumb bug in CoordinateTransform

* Forgot to commit some stuff in merge

* Disable tests that currently don't work on CPU

* Fix temporarily disabled test that should pass on NGVM and INTERPRETER but wasn't due to new axis-selection convention for dot

* Remove obsolete ScalarTensorProduct kernel/instruction

* Review comment

* s/n_dot_axes/dot_axis_count/

* s/dot_axis_count/reduction_axes_count/

* Adapt CPU emitter dot fallback to new kernel

* New Interpreter backend

* PR review comments

* More RP fixes

* oops

* make autodiff tests backend aware

* wip

* wip

* more ops

* wip

* fix merge error

* merge fixes

* Autodiff for abs

* Formatting, more tests for abs

* Cos autodiff; also a clarifying comment in the abs test

* Forgot cos.cpp

* Sin autodiff

* Again, forgot to add sin.cpp  :/

* Tan autodiff

* Minor formatting tweak

* Commit partial work on select backprop so I can run valgrind on the server  :/

* Fix boolean thingy so it works on Linux

* Autodiff for ceiling, convert (untested), floor, sign. Fix unit test for tan. Implement ceiling, floor in VM.

* Fix bug in abs/sign unit test ranges

* Add multiplicative inverse and square-root ops (needed for hyperbolic trig autodiff)

* Better formula for sqrt adjoints

* Autodiff for hyperbolic trig ops

* Forgot to add cpp files for hyperbolics

* Remove inv (don't need it after all); also formatting (oops)

* fix bug with Convert autodiff

* Autodiff for concat

* Restore the accidentally-commented-out unit test for abs

* Formatting

* Fix 'unordered_map.at' exception when Adjoints::Adjoints visit a node that has never been add_delta'd; remove workarounds for that bug

* Fix erroneous 'not implemented' docstring for op::Not

* Autodiff for sum

* Checking in broken support for replace_slice so I can test it with valgrind on the dev box

* Fix unit test bug with tensor initialization; formatting

* Implement replace-slice in CPU backend

* Autodiff for Slice and ReplaceSlice

* Tweak docs for ReplaceSlice

* Remove no-longer-needed cast of arg list to runtime::TensorView

* ParameterizedTensorView and ElementType cleanup

* change element to_type() to from()

Multiply and Add numeric/autodiff unit tests pass