Created
March 28, 2019 03:22
-
-
Save CasiaFan/de2772aeaa100ecb01145b37d4517b22 to your computer and use it in GitHub Desktop.
Measure model flops in tensorflow
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
import tensorflow as tf | |
def load_pb(pb_model): | |
with tf.gfile.GFile(pb_model, "rb") as f: | |
graph_def = tf.GraphDef() | |
graph_def.ParseFromString(f.read()) | |
with tf.Graph().as_default() as graph: | |
tf.import_graph_def(graph_def, name='') | |
return graph | |
def estimate_flops(pb_model): | |
graph = load_pb(pb_model) | |
with graph.as_default(): | |
# placeholder input would result in incomplete shape. So replace it with constant during model frozen. | |
flops = tf.profiler.profile(graph, options=tf.profiler.ProfileOptionBuilder.float_operation()) | |
print('Model {} needs {} FLOPS after freezing'.format(pb_model, flops.total_float_ops)) | |
model = "frozen_inference_graph.pb" | |
estimate_flops(model) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment