Spaces:
Runtime error
Runtime error
File size: 7,969 Bytes
6bc94ac |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 |
from datetime import datetime
from time import perf_counter as timer
import numpy as np
import umap
import visdom
from encoder.data_objects.speaker_verification_dataset import Train_Dataset
colormap = np.array([
[76, 255, 0],
[0, 127, 70],
[255, 0, 0],
[255, 217, 38],
[0, 135, 255],
[165, 0, 165],
[255, 167, 255],
[0, 255, 255],
[255, 96, 38],
[142, 76, 0],
[33, 0, 127],
[0, 0, 0],
[183, 183, 183],
], dtype=np.float) / 255
class Visualizations:
def __init__(self, env_name=None, update_every=10, server="http://localhost", disabled=False):
# Tracking data
self.last_update_timestamp = timer()
self.update_every = update_every
self.step_times = []
self.train_losses = []
self.train_eers = []
print("Updating the visualizations every %d steps." % update_every)
# If visdom is disabled TODO: use a better paradigm for that
self.disabled = disabled
if self.disabled:
return
# Set the environment name
now = str(datetime.now().strftime("%d-%m %Hh%M"))
if env_name is None:
self.env_name = now
else:
self.env_name = "%s (%s)" % (env_name, now)
# Connect to visdom and open the corresponding window in the browser
try:
self.vis = visdom.Visdom(server, env=self.env_name, raise_exceptions=True)
except ConnectionError:
raise Exception("No visdom server detected. Run the command \"visdom\" in your CLI to "
"start it.")
# webbrowser.open("http://localhost:8097/env/" + self.env_name)
# Create the windows
self.loss_win = None
self.eer_win = None
# self.lr_win = None
self.implementation_win = None
self.projection_win = None
self.dev_projection_win = None
self.implementation_string = ""
def log_params(self):
if self.disabled:
return
from encoder import params_data
from encoder import params_model
param_string = "<b>Model parameters</b>:<br>"
for param_name in (p for p in dir(params_model) if not p.startswith("__")):
value = getattr(params_model, param_name)
param_string += "\t%s: %s<br>" % (param_name, value)
param_string += "<b>Data parameters</b>:<br>"
for param_name in (p for p in dir(params_data) if not p.startswith("__")):
value = getattr(params_data, param_name)
param_string += "\t%s: %s<br>" % (param_name, value)
self.vis.text(param_string, opts={"title": "Parameters"})
def log_dataset(self, dataset: Train_Dataset):
if self.disabled:
return
dataset_string = ""
dataset_string += "<b>Speakers</b>: %s\n" % len(dataset.speakers)
dataset_string += "\n" + dataset.get_logs()
dataset_string = dataset_string.replace("\n", "<br>")
self.vis.text(dataset_string, opts={"title": "Dataset"})
def log_implementation(self, params):
if self.disabled:
return
implementation_string = ""
for param, value in params.items():
implementation_string += "<b>%s</b>: %s\n" % (param, value)
implementation_string = implementation_string.replace("\n", "<br>")
self.implementation_string = implementation_string
self.implementation_win = self.vis.text(
implementation_string,
opts={"title": "Training implementation"}
)
def update(self, loss, eer, step, dev_loss=None, dev_eer=None):
# Update the tracking data
now = timer()
self.step_times.append(1000 * (now - self.last_update_timestamp))
self.last_update_timestamp = now
self.train_losses.append(loss)
self.train_eers.append(eer)
print(".", end="")
# Update the plots every <update_every> steps
if step % self.update_every != 0:
return
time_string = "Step time: mean: %5dms std: %5dms" % \
(int(np.mean(self.step_times)), int(np.std(self.step_times)))
print("\nStep %6d Train Loss: %.4f Train EER: %.4f Dev Loss: %.4f Dev EER: %.4f %s" %
(step, np.mean(self.train_losses), np.mean(self.train_eers), dev_loss, dev_eer, time_string))
if not self.disabled:
loss_win_id = 'win1'
self.loss_win = self.vis.line(
[np.mean(self.train_losses)],
[step],
win=loss_win_id,
name="Avg. train Loss",
update="append" if loss_win_id else "None",
opts=dict(
xlabel="Step",
ylabel="Loss",
title="Loss",
)
)
self.vis.line(
[dev_loss],
[step],
win=loss_win_id,
name="Avg. dev Loss",
update="append"
)
err_win_id = 'win2'
self.eer_win = self.vis.line(
[np.mean(self.train_eers)],
[step],
win=err_win_id,
name="Avg. train EER",
update="append" if err_win_id else "None",
opts=dict(
xlabel="Step",
ylabel="EER",
title="Equal error rate"
)
)
self.vis.line(
[dev_eer],
[step],
win=err_win_id,
name="Avg. dev EER",
update="append"
)
if self.implementation_win is not None:
self.vis.text(
self.implementation_string + ("<b>%s</b>" % time_string),
win=self.implementation_win,
opts={"title": "Training implementation"},
)
# Reset the tracking
self.train_losses.clear()
self.train_eers.clear()
self.step_times.clear()
def draw_projections(self, embeds, dev_embeds, utterances_per_speaker, step, out_fpath=None, dev_out_fpath=None, max_speakers=10):
import matplotlib.pyplot as plt
max_speakers = min(max_speakers, len(colormap))
# draw train umap projections
embeds = embeds[:max_speakers * utterances_per_speaker]
n_speakers = len(embeds) // utterances_per_speaker
ground_truth = np.repeat(np.arange(n_speakers), utterances_per_speaker)
colors = [colormap[i] for i in ground_truth]
reducer = umap.UMAP()
projected = reducer.fit_transform(embeds)
plt.scatter(projected[:, 0], projected[:, 1], c=colors)
plt.gca().set_aspect("equal", "datalim")
plt.title("UMAP projection (step %d)" % step)
if not self.disabled:
self.projection_win = self.vis.matplot(plt, win=self.projection_win)
if out_fpath is not None:
plt.savefig(out_fpath)
plt.clf()
# draw dev umap projections
dev_embeds = dev_embeds[:max_speakers * utterances_per_speaker]
n_speakers = len(dev_embeds) // utterances_per_speaker
ground_truth = np.repeat(np.arange(n_speakers), utterances_per_speaker)
colors = [colormap[i] for i in ground_truth]
reducer = umap.UMAP()
projected = reducer.fit_transform(dev_embeds)
plt.scatter(projected[:, 0], projected[:, 1], c=colors)
plt.gca().set_aspect("equal", "datalim")
plt.title("dev UMAP projection (step %d)" % step)
if not self.disabled:
self.dev_projection_win = self.vis.matplot(plt, win=self.dev_projection_win)
if dev_out_fpath is not None:
plt.savefig(dev_out_fpath)
plt.clf()
def save(self):
if not self.disabled:
self.vis.save([self.env_name])
|