Skip to content

Conversation

@mohomran
Copy link
Contributor

./tools/extra/parse_log.sh was broken by the recent changes to the training output, in particular by the way test accuracy and test loss are now displayed. caffe.log.test had empty columns where accuracy and loss per iteration were supposed to show up.

previously (training output):
I0828 00:56:34.795725 31294 solver.cpp:177] Test score #0: 0.9319
I0828 00:56:34.795765 31294 solver.cpp:177] Test score #1: 0.173513

now (training output):
I0831 01:21:36.325670 8457 solver.cpp:302] Test net output #0: accuracy = 0.9426
I0831 01:21:36.325687 8457 solver.cpp:302] Test net output #1: loss = 0.152957 (* 1 = 0.152957 loss)

culprit in parse_log.sh:
grep 'Test score #0' aux.txt | awk '{print $8}' > aux1.txt
grep 'Test score #1' aux.txt | awk '{print $8}' > aux2.txt

Also, fixed a number of typos and updated the script description.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

aux.txtq is not deleted.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for the heads-up, @kloudkl, just noticed it.

@mohomran mohomran force-pushed the fixing_parse_log_script branch from 44abaae to aae8545 Compare August 31, 2014 14:16
shelhamer added a commit that referenced this pull request Aug 31, 2014
@shelhamer shelhamer merged commit 92f6c26 into BVLC:dev Aug 31, 2014
@shelhamer
Copy link
Member

Thanks.

@mohomran mohomran deleted the fixing_parse_log_script branch September 2, 2014 11:10
mitmul pushed a commit to mitmul/caffe that referenced this pull request Sep 30, 2014
RazvanRanca pushed a commit to RazvanRanca/caffe that referenced this pull request Nov 4, 2014
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants