def save(self):
Path( self.get_summary_path() ).write_text( self.get_summary_text() )
self.onSave()
model_data = {
'iter': self.iter,
'options': self.options,
'loss_history': self.loss_history,
'sample_for_preview' : self.sample_for_preview,
'choosed_gpu_indexes' : self.choosed_gpu_indexes,
}
pathex.write_bytes_safe (self.model_data_path, pickle.dumps(model_data) )
if self.autobackup:
current_hour = time.localtime().tm_hour
if self.autobackup_current_hour != current_hour:
self.autobackup_current_hour = current_hour
self.create_backup()
def save_weights(self, filename, force_dtype=None):
d = {}
weights = self.get_weights()
if self.name is None:
raise Exception("name must be defined.")
name = self.name
for w, w_val in zip(weights, nn.tf_sess.run (weights)):
w_name_split = w.name.split('/', 1)
if name != w_name_split[0]:
raise Exception("weight first name != Saveable.name")
if force_dtype is not None:
w_val = w_val.astype(force_dtype)
d[ w_name_split[1] ] = w_val
d_dumped = pickle.dumps (d, 4)
pathex.write_bytes_safe ( Path(filename), d_dumped )
def save(self):
Path( self.get_summary_path() ).write_text( self.get_summary_text() )
self.onSave()
model_data = {
'iter': self.iter,
'options': self.options,
'loss_history': self.loss_history,
'sample_for_preview' : self.sample_for_preview,
'choosed_gpu_indexes' : self.choosed_gpu_indexes,
}
pathex.write_bytes_safe (self.model_data_path, pickle.dumps(model_data) )
if self.autobackup_hour != 0:
diff_hour = int ( (time.time() - self.autobackup_start_time) // 3600 )
if diff_hour > 0 and diff_hour % self.autobackup_hour == 0:
self.autobackup_start_time += self.autobackup_hour*3600
self.create_backup()
def save(self):
summary_path = self.get_strpath_storage_for_file('summary.txt')
Path(summary_path).write_text(self.get_summary_text())
self.onSave()
model_data = {
'iter': self.iter,
'options': self.options,
'loss_history': self.loss_history,
'sample_for_preview': self.sample_for_preview,
'choosed_gpu_indexes': self.choosed_gpu_indexes,
}
pathex.write_bytes_safe(self.model_data_path, pickle.dumps(model_data))
if self.autobackup:
bckp_filename_list = [
self.get_strpath_storage_for_file(filename)
for _, filename in self.get_model_filename_list()
]
bckp_filename_list += [
str(summary_path),
str(self.model_data_path)
]
current_hour = time.localtime().tm_hour
if self.autobackup_current_hour != current_hour:
self.autobackup_current_hour = current_hour
for i in range(15, 0, -1):
idx_str = '%.2d' % i
next_idx_str = '%.2d' % (i + 1)
idx_backup_path = self.autobackups_path / idx_str
next_idx_packup_path = self.autobackups_path / next_idx_str
if idx_backup_path.exists():
if i == 15:
pathex.delete_all_files(idx_backup_path)
else:
next_idx_packup_path.mkdir(exist_ok=True)
pathex.move_all_files(idx_backup_path,
next_idx_packup_path)
if i == 1:
idx_backup_path.mkdir(exist_ok=True)
for filename in bckp_filename_list:
shutil.copy(
str(filename),
str(idx_backup_path / Path(filename).name))
previews = self.get_previews()
plist = []
for i in range(len(previews)):
name, bgr = previews[i]
plist += [(bgr, idx_backup_path /
(('preview_%s.jpg') % (name)))]
for preview, filepath in plist:
preview_lh = ModelBase.get_loss_history_preview(
self.loss_history, self.iter, preview.shape[1],
preview.shape[2])
img = (
np.concatenate([preview_lh, preview], axis=0) *
255).astype(np.uint8)
cv2_imwrite(filepath, img)
