torch_numopt.utils.param_operations

Functions

param_add(params_a, params_b)	Element-wise addition of two parameter groups.
param_argnums(params)	Return a tuple of indices (0, 1, ..., len(params)-1) for the parameter groups.
param_copy(params)	Creates a deep copy of a parameter group (each tensor cloned).
param_detach(params)	Detach (and clone) all tensors from the computation graph.
param_diff(params_a, params_b)	Element-wise subtraction of two parameter groups.
param_dot(params_a, params_b)	Computes the dot product between two parameter groups treated as a single flat vector.
param_flatten(params)	Flattens an entire parameter group into a single 1-dimensional tensor.
param_is_finite(params)	Checks that all elements in a parameter group are finite.
param_mult(params_a, params_b)	Element-wise (Hadamard) product of two parameter groups.
param_neg(params)	Negates every tensor in a parameter group.
param_norm(params)	Euclidean (L2) norm of a parameter group treated as a flat vector.
param_numel(params)	Returns the total amount of parameters.
param_reshape_like(params_flat, params)	Reshapes a vector into a sequence of matrices with the same shapes as the params parameter.
param_scalar_prod(scalar, params)	Multiplies every tensor in a parameter group by a scalar.
param_scaled_add(params_a, params_b, scale)	Computes params_a + scale * params_b element-wise.
param_sizes(params)	Obtains the shape of every matrix in the parameters provided.
param_transpose(params)	Transposes every 2-dimensional tensor in the parameter group.
param_zero_like(params)	Create a tuple of zero tensors with the same shapes as the input parameters.