JarveeLee
commented
7 years ago And this is my usually used code for discard missing var from a checkpoint....However it doesn't work as well here.... def optimistic_restore(save_file): reader = tf.train.NewCheckpointReader(save_file) saved_shapes = reader.get_variable_to_shape_map() var_names = sorted([(var.name, var.name.split(':')[0]) for var in tf.global_variables() if var.name.split(':')[0] in saved_shapes]) restore_vars = []
del var_names[0]
#del var_names[0]
#print ('var name ',var_names)
#print ('saved shapes name ',saved_shapes)
reject_list=['Adam','global_step','Adadelta']
#print (tf.global_variables())
#var_names.pop('beta1_power:0')
#var_names.pop('beta2_power:0')
with tf.variable_scope('', reuse=True):
for var_name, saved_var_name in var_names:
print (saved_var_name,saved_shapes[saved_var_name],var_name in saved_shapes,var_name,saved_var_name in saved_shapes)
'''
ok=1
for ele in reject_list:
print (ele in var_name,ele,var_name)
if ele in var_name:
ok=0
break
if ok==0:
continue
'''
tf.get_variable_scope().reuse_variables()
curr_var = tf.get_variable(saved_var_name)
var_shape = curr_var.get_shape().as_list()
if var_shape == saved_shapes[saved_var_name]:
print ('restore var is ',curr_var,' \n')
restore_vars.append(curr_var)
return restore_vars
MaybeShewill-CV
xrf116
I just download your entire project and run under tensorflow =1.1.0 , and want to train on other dataset . I've created own dataset with the same format however load you pretrained model I got this key miss error..... Key shadow/LSTMLayers/stack_bidirectional_rnn/cell_0/bidirectional_rnn/bw/basic_lstm_cell/biases/Adadelta not found in checkpoint