fc109ba: On June 26, we updated our code on Github. This new version introduces some significant changes which might affect older versions of Ristretto:
- Ristretto network description files:
precisionfield: the quantization modes are
- Ristretto layers:
DeconvolutionRistrettoLayerfor applications like semantic segmentation.
- Quantize layer inputs and layer outputs (previously: only quantize layer outputs).
DataRistrettoLayersince images can be quantized in first convolutional layer.
- Layers without multiplications:
INTEGER_POWER_OF_2_WEIGHTS, the layer activations are now in dynamic fixed point.
- Quantization of layer activations
- In older Ristretto versions, only the layer outputs were quantized. With this new version, layer inputs and layer outputs are in reduced word width format.
- For dynamic fixed point approximation, we introduce 2 new protobuffer fields:
bw_layer_infor the word width of layer inputs
fl_layer_infor the fractional length of layer inputs
- Ristretto Demo on SqueezeNet:
- We renamed the models/SqueezeNet/demo folder (new: models/SqueezeNet/RistrettoDemo)
- We renamed the demo shell scripts in examples/ristretto/
- Ristretto Documentation:
- We removed the docs/ristretto/ folder. All documentation is available on this web page.
- Refactoring in code base:
- We improved the naming of class functions and variables for clarification:
- “fixed point” -> “dynamic fixed point”
- “mini floating point” -> “minifloat”
- “power-2-weights” -> “integer power of 2 weights”
- Updated documentation
- Merge with upstream: This Ristretto version is merged with official Caffe commit