Skip to content

Prevent numerical issues with poisson_nll_loss when log_input=False #3336

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
soumith merged 3 commits into from
Nov 1, 2017
Merged

Prevent numerical issues with poisson_nll_loss when log_input=False #3336

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
ae0eba2
Prevent numerical issues with poisson_nll_loss when log_input=False
gokceneraslan Oct 28, 2017
5fccdac
PEP8 fix
gokceneraslan Oct 28, 2017
3bd453e
Add epsilon support to PoissonNLLLoss in nn.modules.loss
gokceneraslan Oct 28, 2017
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 5 additions & 3 deletions torch/nn/functional.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -924,7 +924,7 @@ def nll_loss(input, target, weight=None, size_average=True, ignore_index=-100, r
raise ValueError('Expected 2 or 4 dimensions (got {})'.format(dim))


def poisson_nll_loss(input, target, log_input=True, full=False, size_average=True):
def poisson_nll_loss(input, target, log_input=True, full=False, size_average=True, eps=1e-8):
r"""Poisson negative log likelihood loss.

See :class:`~torch.nn.PoissonNLLLoss` for details.
Expand All @@ -934,18 +934,20 @@ def poisson_nll_loss(input, target, log_input=True, full=False, size_average=Tru
target: random sample :math:`target \sim Pois(input)`.
log_input: if True the loss is computed as
`exp(input) - target * input`, if False then loss is
`input - target * log(input)`. Default: True
`input - target * log(input+eps)`. Default: True
full: whether to compute full loss, i. e. to add the Stirling
approximation term. Default: False
`target * log(target) - target + 0.5 * log(2 * pi * target)`.
size_average: By default, the losses are averaged over observations for
each minibatch. However, if the field sizeAverage is set to False,
the losses are instead summed for each minibatch. Default: True
eps (float, optional): Small value to avoid evaluation of log(0) when
log_input=False. Default: 1e-8
"""
if log_input:
loss = torch.exp(input) - target * input
else:
loss = input - target * torch.log(input)
loss = input - target * torch.log(input + eps)
if full:
mask = target > 1
loss[mask] += (target * torch.log(target) - target + 0.5 * torch.log(2 * math.pi * target))[mask]
Expand Down
9 changes: 6 additions & 3 deletions torch/nn/modules/loss.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -190,13 +190,15 @@ class PoissonNLLLoss(_Loss):
Args:
log_input (bool, optional): if True the loss is computed as
`exp(input) - target * input`, if False the loss is
`input - target * log(input)`.
`input - target * log(input+eps)`.
full (bool, optional): whether to compute full loss, i. e. to add the
Stirling approximation term
`target * log(target) - target + 0.5 * log(2 * pi * target)`.
size_average (bool, optional): By default, the losses are averaged over
observations for each minibatch. However, if the field size_average
is set to False, the losses are instead summed for each minibatch.
eps (float, optional): Small value to avoid evaluation of log(0) when
log_input=False. Default: 1e-8

Examples::

Expand All @@ -206,15 +208,16 @@ class PoissonNLLLoss(_Loss):
>>> output = loss(log_input, target)
>>> output.backward()
"""
def __init__(self, log_input=True, full=False, size_average=True):
def __init__(self, log_input=True, full=False, size_average=True, eps=1e-8):
super(PoissonNLLLoss, self).__init__()
self.log_input = log_input
self.full = full
self.size_average = size_average
self.eps = eps

def forward(self, log_input, target):
_assert_no_grad(target)
return F.poisson_nll_loss(log_input, target, self.log_input, self.full, self.size_average)
return F.poisson_nll_loss(log_input, target, self.log_input, self.full, self.size_average, self.eps)


class KLDivLoss(_Loss):
Expand Down