OpenBioLink / SAFRAN

Scalable and fast non-redundant rule application for link prediction
MIT License
44 stars 6 forks source link

Segmentation fault (core dumped) error #5

Closed fosterreproducibleresearch closed 2 years ago

fosterreproducibleresearch commented 2 years ago

We are trying to run SAFRAN on Ubuntu 20.04 for the WN18RR dataset but we are facing the following error:


ACTION = applynoisy
PATH_TRAINING = ../train.txt
PATH_TEST = ../test.txt
PATH_VALID = ../valid.txt
PATH_RULES = rules_wn/alpha-100
PATH_OUTPUT = rules/safran_wn18rr_rules.txt
WORKER_THREADS = -1
DISCRIMINATION_BOUND = 1000
UNSEEN_NEGATIVE_EXAMPLES = 5
TOP_K_OUTPUT = 10
REFLEXIV_TOKEN = me_myself_i
ONLY_UNCONNECTED = 0
TRIAL = 0

Reading trainingset...
Trainingset read and indexed in 519 ms
Reading rules...
Rules read in 3725 ms
Reading testset...
Testset read in 114 ms
Reading validationset...
Validationset read in 111 ms
Applying rules...
Relation 0/11 _hypernym
Segmentation fault (core dumped)```
nomisto commented 2 years ago

Hello, sorry to hear that, I'll look into it. Did you use the prebuilt application or did you build it yourself?

fosterreproducibleresearch commented 2 years ago

Hello,

Thank you for the quick reply.

I used the prebuilt application.

On Tue, Dec 14, 2021, 11:07 Simon Ott @.***> wrote:

Hello, sorry to hear that, I'll look into it. Did you use the prebuilt application or did you build it yourself?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/OpenBioLink/SAFRAN/issues/5#issuecomment-993376346, or unsubscribe https://github.com/notifications/unsubscribe-auth/ATRMLE6CEZ3NKNRZGACCHZ3UQ4JO5ANCNFSM5KANECAA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

nomisto commented 2 years ago

Also, to retrace the error, could you if possible share your ruleset (alpha-100)? (You can just drag and drop the file in a comment)

fosterreproducibleresearch commented 2 years ago

Here are the first lines of the rules's file

5 3 0.6 _derivationally_related_form(X,00772640) <= _derivationally_related_form(00772640,X) 1852 2 0.0010799136069114472 _derivationally_related_form(X,00772640) <= _derivationally_related_form(X,A) 1977 3 0.0015174506828528073 _has_part(08562243,Y) <= _has_part(Y,A) 7402 2 2.701972439881113E-4 _member_meronym(12097180,Y) <= 5 2 0.4 _derivationally_related_form(X,05869857) <= _derivationally_related_form(00003553,X) 2 2 1.0 _derivationally_related_form(02676496,Y) <= _derivationally_related_form(Y,02676496) 4 2 0.5 _derivationally_related_form(02676496,Y) <= _derivationally_related_form(Y,00722232) 7402 2 2.701972439881113E-4 _member_meronym(02614788,Y) <= 2 2 1.0 _derivationally_related_form(14299070,Y) <= _derivationally_related_form(Y,14299070) 29722 3 1.0093533409595585E-4 _derivationally_related_form(00152018,Y) <= 3 3 1.0 _verb_group(01637982,Y) <= _verb_group(Y,01637982) 309 2 0.006472491909385114 _synset_domain_topic_of(10434160,Y) <= _synset_domain_topic_of(A,Y) 2 2 1.0 _synset_domain_topic_of(10434160,Y) <= _synset_domain_topic_of(10386984,Y) 29722 4 1.3458044546127448E-4 _derivationally_related_form(X,00226071) <= 1433 2 0.0013956734124214933 _hypernym(X,00020827) <= _hypernym(A,X)

nomisto commented 2 years ago

Ah I see! You learned these rules using AnyBURL-JUNO right? This is a newer version of AnyBURL that also learnes rules with no body, like

29722 4 1.3458044546127448E-4 _derivationally_related_form(X,00226071) <=

which basically includes a baseline probability to the predictions. However SAFRAN currently does not support these as it was developed with the AnyBURL-RE version. So in order to run SAFRAN now I suggest you either learn rules with

or you wait until we implemented these zero body rules, however I cannot say right now when this is.

fosterreproducibleresearch commented 2 years ago

Unfortunately, the ZERO_RULES_ACTIVE argument is not recognized

fosterreproducibleresearch commented 2 years ago

But it is fine, I have used AnyBURL-RE and it is working. Thanks :)

nomisto commented 2 years ago

Sorry, I meant to set it in the config file when learning the rules with AnyBURL, not the config for SAFRAN (f.e. in config-learn.properties file that you use when running java -Xmx3G -cp AnyBURL-JUNO.jar de.unima.ki.anyburl.LearnReinforced config-learn.properties).

Oh perfect!

nomisto commented 2 years ago

Closing this now, as I feel this particular issue is resolved for now. Feel free to open it again or create a new issue if you run into further troubles :)