def __log_str(self, model_file=None):
'''
ClinerModel::__log_str()
Build a string of information about training for the model's log file.
@param model_file. A path to optionally identify where the model was saved.
@return A string of the model's training information
'''
assert self._is_trained, 'ClinerModel not trained'
with io.StringIO() as f:
write(f, u'\n')
write(f, '-' * 40)
write(f, u'\n\n')
if model_file:
write(f, 'model : %s\n' % os.path.abspath(model_file))
write(f, u'\n')
if self._use_lstm:
write(f, u'modeltype: LSTM\n')
else:
write(f, u'modeltype: CRF\n')
if 'hyperparams' in self._score:
for name, value in self._score['hyperparams'].items():
write(f, u'\t%-10s: %s\n' % (name, value))
write(f, u'\n')
print_str(f, 'features', self._features)
write(f, u'\n')
write(f, u'\n')
write(f, 'training began: %s\n' % self._time_train_begin)
write(f, 'training ended: %s\n' % self._time_train_end)
write(f, u'\n')
write(f, u'scores\n')
print_vec(f, 'train precision', self._score['train']['precision'])
print_vec(f, 'train recall ', self._score['train']['recall'])
print_vec(f, 'train f1 ', self._score['train']['f1'])
write(f, self._score['train']['conf'])
if 'dev' in self._score:
print_vec(f, u'dev precision ',
self._score['dev']['precision'])
print_vec(f, u'dev recall ', self._score['dev']['recall'])
print_vec(f, u'dev f1 ', self._score['dev']['f1'])
write(f, self._score['dev']['conf'])
if 'test' in self._score:
print_vec(f, u'test precision ',
self._score['test']['precision'])
print_vec(f, u'test recall ',
self._score['test']['recall'])
print_vec(f, u'test f1 ', self._score['test']['f1'])
write(f, self._score['test']['conf'])
if 'history' in self._score:
for label, vec in self._score['history'].items():
print_vec(f, '%-16s' % label, vec)
write(f, u'\n')
if self._training_files:
write(f, u'\n')
write(f, u'Training Files\n')
if len(self._training_files) < 200:
print_files(f, self._training_files)
else:
write(f, '\t%d files\n' % len(self._training_files))
write(f, u'\n')
write(f, u'-' * 40)
write(f, u'\n\n')
# get output as full string
contents = f.getvalue()
return contents
# Use the `.split('>')` list method to get the infiles and outfiles
def check_input(input_string):
"Returns the position of a delimiter in a string"
position = 0
for string in '>':
if string == '>':
return position # what? call combine files and set to a new file?
else:
position += 1
return None
# TODO: Add a flag for this behavior; fix it to take mult. files.
def add_line_num(infile):
"Print the contents of a file with line numbers added."
with open(infile) as f:
iter = 1
for line in f.readlines():
print(f"{iter}\t{line}", end="")
iter += 1
if __name__ == "__main__":
output = tools.print_files(argv)
print(output)
add_line_num(
argv[1:]) # not working - needs to take a str, bytes or file obj