def cdCmd(path, DEFAULTS, layout):
"""
Set working directory and load default parameters (main).
"""
import sys
from platform import uname
IN_COLAB = "google.colab" in sys.modules
IN_WSL2 = "microsoft-standard" in uname().release
IN_LINUX = not IN_COLAB and not IN_WSL2
out = layout["out"]
path = get_path(path)
with out:
main(cd=path)
logger.info("Loading configuration data ...")
# Tabs
tab = widgets.Tab()
tab.children = [glimpseUI(out, DEFAULTS), fitUI(out, DEFAULTS)]
if IN_LINUX:
tensorboard = widgets.Output()
tab.children = tab.children + (showUI(out, DEFAULTS), tensorboard)
elif IN_WSL2:
tensorboard = None
tensorboard_info = widgets.Label(
value=(
'Run "tensorboard --logdir=<working directory>" in the terminal'
' and then open "localhost:6006" in the browser'
)
)
tab.children = tab.children + (showUI(out, DEFAULTS), tensorboard_info)
elif IN_COLAB:
tensorboard = None
tab.children = tab.children + (
widgets.Label(value="Disabled in Colab"),
widgets.Label(value="Disabled in Colab"),
)
tab.children = tab.children + (logUI(out),)
tab.set_title(0, "Extract AOIs")
tab.set_title(1, "Fit the data")
tab.set_title(2, "View results")
tab.set_title(3, "Tensorboard")
tab.set_title(4, "View logs")
if tensorboard is not None:
with tensorboard:
notebook.start(f"--logdir '{path}'")
notebook.display(height=1000)
# insert tabs into GUI
wd = widgets.Label(value=f"Working directory: {path}")
layout.remove_child("cd")
layout.add_child("tab", tab, beginning=True)
layout.add_child("wd", wd, beginning=True)
with out:
logger.info("Loading configuration data: Done")
out.clear_output(wait=True)
def setup_tensorboard(args):
from tensorboard import notebook, manager
import signal
import shutil
# Kill tensorboard
for info in manager.get_all():
data_source = manager.data_source_from_info(info)
print(f"port {info.port}: {data_source} (pid {info.pid})")
if data_source == "logdir {args.output_dir}":
pid = info.pid
logger.info(f"Killing tensorboard at pid: {pid}")
os.kill(pid, signal.SIGKILL)
break
# Delete output directory
if os.path.exists(args.output_dir):
logger.info(f"Deleting {args.output_dir}")
shutil.rmtree(args.output_dir)
logger.info(f"Creating {args.output_dir}")
os.makedirs(args.output_dir)
# Start notebook
notebook.start(f"--logdir {args.output_dir}")
# Kill tensorboard
for info in manager.get_all():
data_source = manager.data_source_from_info(info)
print(f"port {info.port}: {data_source} (pid {info.pid})")
if data_source == "logdir {args.output_dir}":
port = info.port
print()
notebook.display(port=port, height=1000)
break
def forward(self, x):
x = self.conv1(x)
x = self.pool(x)
x = self.conv2(x)
x = self.pool(x)
x = self.dropout(x)
x = self.adaptive_pool(x)
x = self.flatten(x)
x = self.linear1(x)
x = self.relu(x)
x = self.linear2(x)
y = self.sigmoid(x)
return y
net = Net()
print(net)
writer = SummaryWriter('./data/tensorboard')
writer.add_graph(net,input_to_model = torch.rand(1,3,32,32))
writer.close()
from tensorboard import notebook
#查看启动的tensorboard程序
notebook.list()
#启动tensorboard程序
notebook.start("--logdir ./data/tensorboard")
#等价于在命令行中执行 tensorboard --logdir ./data/tensorboard
#可以在浏览器中打开 http://localhost:6006/ 查看
model.summary()
import datetime
import os
stamp = datetime.datetime.now().strftime("%Y%m%d-%H%M%S")
logdir = os.path.join('data', 'autograph', stamp)
## 在 Python3 下建议使用 pathlib 修正各操作系统的路径
# from pathlib import Path
# stamp = datetime.datetime.now().strftime("%Y%m%d-%H%M%S")
# logdir = str(Path('./data/autograph/' + stamp))
tensorboard_callback = tf.keras.callbacks.TensorBoard(logdir, histogram_freq=1)
model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=0.001),
loss=tf.keras.losses.binary_crossentropy,
metrics=["accuracy"])
history = model.fit(ds_train,
epochs=10,
validation_data=ds_test,
callbacks=[tensorboard_callback],
workers=4)
from tensorboard import notebook
notebook.list()
#在tensorboard中查看模型
notebook.start("--logdir /home/kesci/input/data3483/data/keras_model")
default=18,
type=int,
help="dataloader num_worker")
parser.add_argument("--NETWORK",
default='senet',
type=str,
help="network structure",
choices=['resnet', 'senet', 'repvgg'])
opt = parser.parse_args()
print(json.dumps(vars(opt)))
DEVICE = torch.device('cuda:1' if torch.cuda.is_available() else 'cpu')
torch.backends.cudnn.benchmark = True
log_dir = './log'
timestamp = time.strftime('%Y-%m-%d-%H_%M_%S', time.localtime(time.time()))
notebook.start("--logdir {}".format(log_dir))
writer = SummaryWriter(os.path.join(log_dir, timestamp))
df = pd.read_csv('./dataset/label.csv')
weight_list = df['train_cls'].values
weight_list = 1 / weight_list
weight_list = torch.tensor(weight_list).float()
train_transform = transforms.Compose([
transforms.ToPILImage(),
# transforms.Resize(256),
transforms.RandomAffine(10, shear=0.1, fillcolor=(255, 255, 255)),
transforms.ColorJitter(brightness=0.1,
contrast=0.1,
saturation=0.1,
hue=0.1),