concrete.ml.quantization.quantized_module_passes.md
Last updated
Was this helpful?
Last updated
Was this helpful?
concrete.ml.quantization.quantized_module_passesOptimization passes for QuantizedModules.
PowerOfTwoScalingRoundPBSAdapterDetect neural network patterns that can be optimized with round PBS.
__init__property num_ignored_valid_patterns
Get the number of optimizable patterns that were ignored.
Patterns could be ignored since a number of rounding bits was set manually through the compilation function.
Returns:
result (int): number of patterns that could be optimized but were not
compute_op_predecessorsCompute the predecessors for each QuantizedOp in a QuantizedModule.
Stores, for each quantized op, a list of quantized ops that produce its inputs. Currently only the first input of the operations is considered as it is, usually, the encrypted input.
Returns:
result (PredecessorsType): a dictionary containing a hierarchy of op predecessors
detect_patternsDetect the patterns that can be optimized with roundPBS in the QuantizedModule.
Args:
predecessors (PredecessorsType): Module predecessor operation list
Returns:
result (PatternDict): list of optimizable patterns
match_path_patternDetermine if a pattern has the structure that makes it viable for roundPBS.
Args:
predecessors (PredecessorsType): Module predecessor operation list
nodes_in_path (List[QuantizedOp]): list of quantized ops in the pattern
input_producer_of_path (Optional[QuantizedOp]): operation that produces the input
Returns:
result (bool): whether the pattern can be optimized
processAnalyze an ONNX graph and detect Gemm/Conv patterns that can use RoundPBS.
We want to detect a gemm/conv node whose weights/bias are Brevitas QAT, and whose input is produced by a Brevitas QAT node that is applied on the output of another Gemm/conv node. Optionally a Relu can be placed before this input quantization node.
Nothing will be done if rounding is already specified.
Returns:
result (PatternDict): a dictionary containing for each Conv/Gemm node for which round PBS can be applied based on power-of-two scaling factors
process_patternsConfigure the rounding bits of roundPBS for the optimizable operations.
Args:
valid_paths (PatternDict): list of optimizable patterns
Returns:
result (PatternDict): list of patterns actually optimized with roundPBS