Skip to main content

hetorch.passes.builtin.lowering

Operation Lowering Pass: Convert generic torch operations to HE-specific operations

This pass converts generic PyTorch operations (like torch.mul, torch.add) into HE-specific operations (cmult/pmult, cadd/padd) based on operand types.

Classes​

OperationLoweringPass​

Lower generic torch operations to HE-specific operations.

This pass distinguishes between:

  • Ciphertext-Ciphertext operations (cmult, cadd)
  • Ciphertext-Plaintext operations (pmult, padd)

The pass uses heuristics to determine operand types:

  • Constants (float/int literals) are plaintext
  • Outputs of HE operations are ciphertexts
  • Inputs (placeholders) are ciphertexts
  • Module outputs (Linear, Conv, etc.) are ciphertexts

Attributes: None

Methods:

__init__(self)​

Initialize OperationLoweringPass

analyze_cost(self, graph_module: torch.fx.graph_module.GraphModule, context: hetorch.compiler.context.CompilationContext) -\> hetorch.backend.cost_model.CostAnalysis​

Analyze cost impact of this pass

Args: graph_module: Graph module to analyze context: Compilation context

Returns: Cost analysis result

transform(self, graph_module: torch.fx.graph_module.GraphModule, context: hetorch.compiler.context.CompilationContext) -\> torch.fx.graph_module.GraphModule​

Lower generic operations to HE-specific operations.

Args: graph_module: Input graph module context: Compilation context

Returns: Transformed graph module with HE-specific operations

validate(self, graph_module: torch.fx.graph_module.GraphModule, context: hetorch.compiler.context.CompilationContext) -\> bool​

Validate that the pass can be applied.

Args: graph_module: Graph module to validate context: Compilation context

Returns: True if validation passes