We wrap everything in a tf.function() like
` def f():
print('Tracing!')
tf.print('Executing')
tf.function(f)() `
It is the graph mode from TF1 and we use the mode still (in a session).
I will come up with a min working example (probably next week)
Hi CourageousKoala93 , not 100% sure I understand what graphmode is, I see it's a legacy option maybe from TF1? If you can put a small snippet so I can try it on my side that'll be helpful!
I have to look deeper into our codebase to see what exactly happens.
Besides that, the procedure is the same, yes…
Does it make a difference for the autologging if this is done in graphmode ?
Am I doing something differently from you?
Hi Alek,
If I understand correctly, you're basically reporting images to be shown in tensorboard (with tf.summary), am I correct?
When I use this code, I get images logged:
` import tensorflow as tf
from clearml import Task
Task.init('test','test tb image')
w = tf.summary.create_file_writer('test/logs')
with w.as_default():
image1 = tf.random.uniform(shape=[8, 8, 1])
image2 = tf.random.uniform(shape=[8, 8, 1])
tf.summary.image("grayscale_noise", [image1, image2], step=0)
Convert the original dtype=int32 Tensor into dtype=float64.
rgb_image_float = tf.constant([
[[1000, 0, 0], [0, 500, 1000]],
]) / 1000
tf.summary.image("picture", [rgb_image_float], step=0)
Convert original dtype=uint8 Tensor into proper range.
rgb_image_uint8 = tf.constant([
[[1, 1, 0], [0, 0, 1]],
], dtype=tf.uint8) * 255
tf.summary.image("picture", [rgb_image_uint8], step=1) `
Hi Alek
It should be auto logged. Could you please give me some details about your environment ?
