def add_histogram(self,
tag,
values,
global_step=None,
bins='tensorflow',
walltime=None):
"""Add histogram to summary.
Args:
tag (string): Data identifier
values (torch.Tensor, numpy.array, or string/blobname): Values to build histogram
global_step (int): Global step value to record
bins (string): One of {'tensorflow','auto', 'fd', ...}. This determines how the bins are made. You can find
other options in: https://docs.scipy.org/doc/numpy/reference/generated/numpy.histogram.html
walltime (float): Optional override default walltime (time.time())
seconds after epoch of event
Examples::
from torch.utils.tensorboard import SummaryWriter
import numpy as np
writer = SummaryWriter()
for i in range(10):
x = np.random.random(1000)
writer.add_histogram('distribution centers', x + i, i)
writer.close()
Expected result:
.. image:: _static/img/tensorboard/add_histogram.png
:scale: 50 %
"""
self._get_file_writer().add_summary(summary.histogram(
tag, values, step=global_step, buckets=self.default_bins),
global_step=global_step,
walltime=walltime)
def param_summary(model, writer, step):
state = model.state_dict()
for _p in state.keys():
param = state[_p].cpu().numpy()
s = summary.histogram(_p, param.flatten())
writer.add_summary(s, global_step=step)
def test_log_histogram_summary():
logdir = './experiment/histogram'
writer = FileWriter(logdir)
for i in range(10):
mu, sigma = i * 0.1, 1.0
values = np.random.normal(mu, sigma,
10000) # larger for better looking.
hist = summary.histogram('discrete_normal', values)
writer.add_summary(hist, i + 1)
writer.flush()
writer.close()
def log_gradient(self, network_name, gradient):
assert self.summary_writer
summary_value = summary.histogram('{0}'.format(network_name), gradient)
self.summary_writer.add_summary(summary_value)
def test_histogram_summary():
mu, sigma = 0.1, 1.0
values = np.random.normal(mu, sigma, 10)
hist = summary.histogram('discrete_normal', values)
assert len(hist.value) == 1
assert hist.value[0].tag == 'discrete_normal'
def get_grad(g):
# logging using tensorboard
grad = g.asnumpy().flatten()
s = summary.histogram(args.name, grad)
summary_writer.add_summary(s)
return mx.nd.norm(g)/np.sqrt(g.size)
def add_histo_summary(summary_writer, name, value, step):
value = value.view(-1).data.cpu().numpy()
summ = summary.histogram(name=name, values=value)
summary_writer.add_summary(summary=summ, global_step=step)
def log_histogram(name, values, step=None):
if _tf_logger is None:
return ValueError
_tf_logger.add_summary(histogram(name, values), global_step=step)