release 0.2.0.6

- fix bugs in quantization
- add evaluating tool for quantization
- add ADMM support in quantization
- fix lock in thread pool
- fix fusing for deconv
- fix reshape converting from ONNX to MNN
- turn off blob size checking by default

beta 0.2.0.5

- CPU
	- add support for DepthToSpace & SpaceToDepth ops
- OpenGL
	- add Android demo
	- add half / float runtime option
	- add support for ROIPooling, Squeeze
	- fix bugs in conv im2col
- OpenCL
	- fix Concat, Eltwise, Reshape bugs
- Tools
	- add KL threshold method in quantization tool
	- support optimization for graph with multiple rnn

beta 0.2.0.4

    - bug fix for quantization tool
    - bug fix/performance update for thread pool
    - bug fix for converters
    - tutorial/doc update
    - more op support

beta 0.2.0.3

- add quantization tool & cpu impl & demo/exec
- add thread pool
- add tests
- fix onnx converter tensor name mismatch
- optimize cpu performance with SSE for windows

beta 0.2.0.2

- CPU
  - add padding support
  - fix bug in permute when channel % 4 != 0
  - fix bug in exp with extreme value
- OpenCL
  - add protecting logics
- OpenGL
  - add protecting logics
  - support NCHW format in Squeeze and Reshape
- Converter
  - add ShuffleChannel support for Caffe
  - add Clip/Transpose/Unary/Pad supports for ONNX

beta 0.2.0.1

- support both armv7/arm64 in podspec (pod version >= 1.5.0 required)
- refactor neg axis support
- fix memory overlap in de-conv
- fix CONVOLUTION_TILED_NUMBER spell error
- fix few warnings
- add binary / interp / permute / relu / reshape / softmax support and optimize conv for OpenGL backend
- add clean in nmake build script

beta 0.2.0.0

- replace FreeImage with stb_image
- warn unicode error in Windows compiling
- separate clang/gcc build script for android
- add default values in fbs
- optimize CPU conv / conv depthwise / deconv / deconv depthwise / lstm / sigmoid
- add sub support in eltwise
- add reciprocal / log1p / log in unary
- add zero like / select / set diff 1d
- add batch support for permute
- add training codes
- fix metal error in dynamic separate storage type handling

beta 0.1.1.6

- add support for windows
- fix bugs in converting dropout
- fix bugs in post treat

beta 0.1.1.5

- cpu & gpu
	- add ceil mode in pool
	- fix softmax with neg axis
- cpu
	- add unsqueeze op
	- optimize lstm
- gpu
	- add 5x5 winograd in metal
	- add batch support for winograd in opencl
- onnx
	- add concat / gather / shape / squeeze / unsqueeze
	- fix data type support in constant

beta 0.1.1.4

- refactor OpenGL, test relative codes
- optimize LRN
- fix doc link error