torch_concepts.nn.CAVEmbeddingToConcept¶

class CAVEmbeddingToConcept(in_embeddings: int | Annotations, out_concepts: int | Annotations, **fit_kwargs)[source]¶

Concept encoder based on Concept Activation Vectors (Kim et al., 2018).

The layer is constructed unfitted and trained post hoc with fit(), which fits one logistic-regression probe per concept on frozen activations and stores the unit-normalized probe weights as CAVs. The CAVs are buffers, not parameters: they are invisible to optimizers and are never updated by the main loss, but they move with .to(device) and survive state_dict round-trips.

The forward pass returns the signed distance of each embedding to each concept’s decision boundary, x @ cav_j + bias_j: its sign equals the probe’s prediction (positive means concept present) and its gradient w.r.t. the input is exactly the unit CAV, matching the directional derivative used by TCAV.

cavs¶

Buffer of shape (out_concepts, in_embeddings) holding the unit-norm CAVs (zeros before fit()).

Type:: torch.Tensor

bias¶

Buffer of shape (out_concepts,) holding the probe intercepts rescaled by the same normalization.

Type:: torch.Tensor

Parameters:

in_embeddings – Number of input embedding features.
out_concepts – Number of output concept representations.
**fit_kwargs – Additional keyword arguments for sklearn.linear_model.LogisticRegression (max_iter defaults to 1000).

Example

>>> import torch
>>> from torch_concepts.nn import CAVEmbeddingToConcept
>>>
>>> _ = torch.manual_seed(0)
>>> encoder = CAVEmbeddingToConcept(in_embeddings=16, out_concepts=2)
>>> embeddings = torch.randn(64, 16)
>>> labels = (embeddings[:, :2] > 0).float()
>>> accuracy = encoder.fit(embeddings, labels)
>>> concepts = encoder(embeddings)
>>> print(concepts.shape)
torch.Size([64, 2])

References

Kim et al. “Interpretability Beyond Feature Attribution: Quantitative Testing with Concept Activation Vectors (TCAV)”, ICML 2018. https://proceedings.mlr.press/v80/kim18d

__init__(in_embeddings: int | Annotations, out_concepts: int | Annotations, **fit_kwargs)[source]¶

Initialize the encoder.

Parameters:

in_embeddings – Number of input embedding features.
out_concepts – Number of output concept representations.
**fit_kwargs – Additional keyword arguments for sklearn.linear_model.LogisticRegression (max_iter defaults to 1000).

Methods

`__init__`(in_embeddings, out_concepts, ...)	Initialize the encoder.
`add_module`(name, module)	Add a child module to the current module.
`annotate`(x[, out_concepts])
`apply`(fn)	Apply `fn` recursively to every submodule (as returned by `.children()`) as well as self.
`bfloat16`()	Casts all floating point parameters and buffers to `bfloat16` datatype.
`buffers`([recurse])	Return an iterator over module buffers.
`children`()	Return an iterator over immediate children modules.
`compile`(args, *kwargs)	Compile this Module's forward using `torch.compile()`.
`cpu`()	Move all model parameters and buffers to the CPU.
`cuda`([device])	Move all model parameters and buffers to the GPU.
`double`()	Casts all floating point parameters and buffers to `double` datatype.
`eval`()	Set the module in evaluation mode.
`extra_repr`()	Return the extra representation of the module.
`fit`(embeddings, concept_labels)	Fit one CAV per concept on frozen activations.
`float`()	Casts all floating point parameters and buffers to `float` datatype.
`forward`(embeddings)	Encode embeddings into signed distances to the concept boundaries.
`get_buffer`(target)	Return the buffer given by `target` if it exists, otherwise throw an error.
`get_extra_state`()	Return any extra state to include in the module's state_dict.
`get_parameter`(target)	Return the parameter given by `target` if it exists, otherwise throw an error.
`get_submodule`(target)	Return the submodule given by `target` if it exists, otherwise throw an error.
`half`()	Casts all floating point parameters and buffers to `half` datatype.
`ipu`([device])	Move all model parameters and buffers to the IPU.
`load_state_dict`(state_dict[, strict, assign])	Copy parameters and buffers from `state_dict` into this module and its descendants.
`modules`([remove_duplicate])	Return an iterator over all modules in the network.
`mtia`([device])	Move all model parameters and buffers to the MTIA.
`named_buffers`([prefix, recurse, ...])	Return an iterator over module buffers, yielding both the name of the buffer as well as the buffer itself.
`named_children`()	Return an iterator over immediate children modules, yielding both the name of the module as well as the module itself.
`named_modules`([memo, prefix, remove_duplicate])	Return an iterator over all modules in the network, yielding both the name of the module as well as the module itself.
`named_parameters`([prefix, recurse, ...])	Return an iterator over module parameters, yielding both the name of the parameter as well as the parameter itself.
`parameters`([recurse])	Return an iterator over module parameters.
`prune`(mask)	Prune the predictor by removing connections based on the given mask.
`register_backward_hook`(hook)	Register a backward hook on the module.
`register_buffer`(name, tensor[, persistent])	Add a buffer to the module.
`register_forward_hook`(hook, *[, prepend, ...])	Register a forward hook on the module.
`register_forward_pre_hook`(hook, *[, ...])	Register a forward pre-hook on the module.
`register_full_backward_hook`(hook[, prepend])	Register a backward hook on the module.
`register_full_backward_pre_hook`(hook[, prepend])	Register a backward pre-hook on the module.
`register_load_state_dict_post_hook`(hook)	Register a post-hook to be run after module's `load_state_dict()` is called.
`register_load_state_dict_pre_hook`(hook)	Register a pre-hook to be run before module's `load_state_dict()` is called.
`register_module`(name, module)	Alias for `add_module()`.
`register_parameter`(name, param)	Add a parameter to the module.
`register_state_dict_post_hook`(hook)	Register a post-hook for the `state_dict()` method.
`register_state_dict_pre_hook`(hook)	Register a pre-hook for the `state_dict()` method.
`requires_grad_`([requires_grad])	Change if autograd should record operations on parameters in this module.
`set_extra_state`(state)	Set extra state contained in the loaded state_dict.
`set_submodule`(target, module[, strict])	Set the submodule given by `target` if it exists, otherwise throw an error.
`share_memory`()	See `torch.Tensor.share_memory_()`.
`state_dict`(*args[, destination, prefix, ...])	Return a dictionary containing references to the whole state of the module.
`to`(args, *kwargs)	Move and/or cast the parameters and buffers.
`to_empty`(*, device[, recurse])	Move the parameters and buffers to the specified device without copying storage.
`train`([mode])	Set the module in training mode.
`type`(dst_type)	Casts all parameters and buffers to `dst_type`.
`xpu`([device])	Move all model parameters and buffers to the XPU.
`zero_grad`([set_to_none])	Reset gradients of all model parameters.

Attributes

`T_destination`
`call_super_init`
`dump_patches`
`training`