zeekay
e037e8f7ea
The distribution called itself torus-ml while shipping a package directory named concrete, so the name a user installs and the name they import disagreed — and most of the tree already said torus.ml, which could not resolve against src/concrete. Moved src/concrete to src/torus and pointed the include at it. The dependency was Zama's concrete-python from their index. It is torus-fhe from ours, which is what luxfhe/torus already ships: that frontend declares torus-fhe and ships a torus/fhe module, so every 'from torus.fhe' in this tree was already naming our compiler and only the dependency lagged. Two distributions share one namespace, which is the shape the upstream this forked from used and the reason the split works: torus-fhe gives torus.fhe, torus-ml gives torus.ml. Swept every other spelling — identifiers, enum members, workflow variables, filenames, notebooks. LICENSE and NOTICE keep the upstream attribution a fork owes; the figures carry the old name in JPEG metadata a text sweep cannot reach.
5.1 KiB
module torus.ml.torch.hybrid_backprop_linear
Linear layer implementations for backprop FHE-compatible models.
class ForwardModuleLinear
Forward module for linear layers.
method __init__
__init__(weight, bias=None, weight_transposed=False)
method forward
forward(input_tensor)
Forward pass for linear layers.
Args:
input_tensor: The input tensor.
Returns: The output tensor after applying the linear transformation.
class BackwardModuleLinear
Backward module for linear layers.
method __init__
__init__(weight, weight_transposed=False)
method forward
forward(grad_output)
Backward pass for linear layers.
Args:
grad_output: The gradient output tensor.
Returns: The gradient input tensor after applying the backward pass.
class CustomLinear
Custom linear module.
method __init__
__init__(weight, bias=None, weight_transposed=False)
method forward
forward(input_tensor)
Forward pass of the custom linear module.
Args:
input_tensor: The input tensor.
Returns: The output tensor after applying the custom linear module.
class ForwardBackwardModule
Custom autograd function for forward and backward passes.
method backward
backward(ctx, grad_output)
Backward pass of the custom autograd function.
Args:
ctx: The context object.grad_output: The gradient output tensor.
Returns: The gradient input tensor after applying the backward pass.
method forward
forward(ctx, input_tensor, forward_module, backward_module)
Forward pass of the custom autograd function.
Args:
ctx: The context object.input_tensor: The input tensor.forward_module: The forward module.backward_module: The backward module.
Returns: The output tensor after applying the forward pass.