Module flowcon.CNF.neural_odes.wrappers.cnf_regularization

Functions

def directional_l2_regularization_fn(x, logp, dx, dlogp, unused_context)
def jacobian_diag_frobenius_regularization_fn(x, logp, dx, dlogp, context)
def jacobian_frobenius_regularization_fn(x, logp, dx, dlogp, context)
def jacobian_offdiag_frobenius_regularization_fn(x, logp, dx, dlogp, context)
def l1_regularzation_fn(x, logp, dx, dlogp, unused_context)
def l2_regularzation_fn(x, logp, dx, dlogp, unused_context)

Classes

class RegularizedODEfunc (odefunc, regularization_fns)

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing to nest them in a tree structure. You can assign the submodules as regular attributes::

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will have their parameters converted too when you call :meth:to, etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

:ivar training: Boolean represents whether this module is in training or evaluation mode. :vartype training: bool

Initialize internal Module state, shared by both nn.Module and ScriptModule.

Expand source code 
class RegularizedODEfunc(nn.Module):
    def __init__(self, odefunc, regularization_fns):
        super(RegularizedODEfunc, self).__init__()
        self.odefunc = odefunc
        self.regularization_fns = regularization_fns

    def before_odeint(self, *args, **kwargs):
        self.odefunc.before_odeint(*args, **kwargs)

    def forward(self, t, state):
        class SharedContext(object):
            pass

        with torch.enable_grad():
            x, logp = state[:2]
            x.requires_grad_(True)
            logp.requires_grad_(True)
            dstate = self.odefunc(t, (x, logp))
            if len(state) > 2:
                dx, dlogp = dstate[:2]
                reg_states = tuple(reg_fn(x, logp, dx, dlogp, SharedContext) for reg_fn in self.regularization_fns)
                return dstate + reg_states
            else:
                return dstate

    @property
    def _num_evals(self):
        return self.odefunc._num_evals

Ancestors

  • torch.nn.modules.module.Module

Class variables

var call_super_init : bool
var dump_patches : bool
var training : bool

Methods

def before_odeint(self, *args, **kwargs)
def forward(self, t, state) ‑> Callable[..., Any]

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the :class:Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.