I'm using tf.summary.histogram(var_name, var, family='my_family') to log a histogram. In the tensorboard interface it appears as
my_family/my_family/var_name
Does anybody know what the logic is behind duplicating the family name?
It does seem intentional, as I find the following in tensorflow/tensorflow/python/ops/summary_op_util.py :
# Use family name in the scope to ensure uniqueness of scope/tag.
scope_base_name = name if family is None else '{}/{}'.format(family, name)
with ops.name_scope(scope_base_name, default_name, values) as scope:
if family is None:
tag = scope.rstrip('/')
else:
# Prefix our scope with family again so it displays in the right tab.
tag = '{}/{}'.format(family, scope.rstrip('/'))
The first time family is inserted in scope_base_name = name if family is None else '{}/{}'.format(family, name), and the second time in tag = '{}/{}'.format(family, scope.rstrip('/')), which according to the comments in the code was deliberate.
I too was frustrated by this, but in the context of using tf.summary.scalar. I've resorted to using:
tf.summary.scalar('myfamily/myname', var)
Now the variables show up in Tensorboard without the duplication of the family name.
P.S. I would have made this a "comment" instead of an answer, but my reputation is too low.
Related
Imagine I have a report, a letter actually, which I need to translate to several languages. I have created a greeting field in the form which is filled programatically by an onchange event method.
if self.partner_id.gender == 'female':
self.letter_greeting = _('Dear %s %s,') % ( # the translation should be "Estimada"
self.repr_recipient_id.title.shorcut, surname
)
elif self.partner_id.gender == 'male':
self.letter_greeting = _('Dear %s %s,') % ( # translation "Estimado"
self.repr_recipient_id.title.shorcut, surname
)
else:
self.letter_greeting = _('Dear %s %s,') % ( # translation: "Estimado/a"
self.partner_id.title.shorcut, surname
)
In that case the word Dear should be translated to different Spanish translations depending on which option is used, this is because we use different termination depending on the gender. Exporting the po file I found that all the options are altogether, that make sense because almost all the cases the translations will be the same, but not in this case:
#. module: custom_module
#: code:addons/custom_module/models/sale_order.py:334
#: code:addons/custom_module/models/sale_order.py:338
#: code:addons/custom_module/models/sale_order.py:342
#, python-format
msgid "Dear %s %s,"
msgstr "Dear %s %s,"
Solutions I can apply directly
Put all the terms in different entries to avoid the same translation manually every time I need to update the po file. This can be cumbersome if you have many different words with that problem. If I do it and I open the file with poedit, this error appears: duplicate message definition
Put all the possible combinations with slashes, this is done y some other parts of Odoo. For both gender would be:
#. module: stock
#: model:res.company,msg:stock.res_company
msgid "Dear"
msgstr "Estimado/a"
This is just an example. I can think of many words that look the same in English, but they use different spelling or meanings in other languages depending on the context.
Possible best solutions
I don't know if Odoo know anything aboutu the context of a word to know if it was already translated or not. Adding a context manually could solve the problem, at least for words with different meanings.
The nicest solution would be to have a parameter to the translation module to make sure that the word is exported as an isolated entry for that especific translation.
Do you think that I am giving to it too much importance haha? Do you know if there is any better solution? Why is poedit not taking into account that problem at all?
I propose an extension of models res.partner.title and res.partner.
res.partner.title should get a translateable field for saving salutation prefixes like 'Dear' or 'Sehr geehrter' (German). Maybe it's worth to get something about genders, too, but i won't get into detail here about that.
You probably want to show the configuring user an example like "Dear Mr. Name" or something like that. A computed field should work.
On res.partner you should just implement either a computed field or just a method to get a full salutation for a partner record.
To some degree this is a linguistics problem. I believe the best solution would be to use a different "Source Language", one made up of keys, and then have English as another Translation. The word "Dear" in English does not have a gender context (and typically, much of English doesn't), while the word "Estimado" in Spanish does. The translation from that Spanish word to English is more appropriately "Masculine Dear." Therefore, using keys as your source language, you would have this:
SourceText (EnglishDescription) -> Translation (English) -> Translation (Spanish)
DearMasculine -> Dear -> Estimado
DearFeminine -> Dear -> Estimada
DearNuetral -> Dear -> Estimado/a
I want to know where does SimObject names like mem_ctrls, membus, replacement_policy are set in gem5. After looking at the code, I understood that, these name are used in stats.txt.
I have looked into SimObject code files(py,cc,hh files). I printed all Simobject names by stepping through root descendants in Simulation.py and then searched some of the names like mem_ctrls using vscode, but could not find a place where these names are set.
for obj in root.descendants():
print("object name:%s\n"% obj.get_name())
These names are the Python variable names from the configuration/run script.
For instance, from the Learning gem5 simple.py script...
from m5.objects import *
# create the system we are going to simulate
system = System()
# Set the clock fequency of the system (and all of its children)
system.clk_domain = SrcClockDomain()
system.clk_domain.clock = '1GHz'
system.clk_domain.voltage_domain = VoltageDomain()
# Set up the system
system.mem_mode = 'timing' # Use timing accesses
system.mem_ranges = [AddrRange('512MB')] # Create an address range
The names will be system, clk_domain, mem_ranges.
Note that only the SimObjects will have a name. The other parameters (e.g., integers, etc.) will not have a name.
You can see where this is set here: https://gem5.googlesource.com/public/gem5/+/master/src/python/m5/SimObject.py#1352
I am writing a script with Jira python and I have encountered a big obstacle here.
I need to access to one of the issuelinks under "is duplicated by" but I don't have any idea about the attributes I can use.
I can get to the issuelinks field but I can't go further from here.
This is I've got so far:
issue = jira.issue(ISSUE_NUM) #this is the issue I am handling
link = issue.fields.issuelinks # I 've accessed to the issuelinks field
if hasattr(link, "inwardIssue"):
inwardIssue = link.inwardIssue
and I want to do this from here :
if(str(inwardIssue.type(?)) == "is duplicated by"):
inward Issues can be
is cloned by
is duplicated by
and so on.
how can I get the type of inward Issues??
There seem to be a few types of issue links. So far I've seen: Blocker, Cause, Duplicate and Reference.
In order to identify the type that the IssueLink is you can do the following:
issue = jira.issue(ISSUE_NUM)
all_issue_links = issue.fields.issuelinks
for link in all_issue_links:
if link.type.name == 'Duplicate':
inward_issue = link.inwardIssue
# Do something with link
I'm trying to train my Wit.ai bot in order to recognize the first name of someone. I'm not very sure if I well understand how the NLP works so I'll give you an example.
I defined a lot of expressions like "My name is XXXX", "Everybody calls me XXXX"
In the "Understanding" table I added an entity named "contact_name" and I add almost 50 keywords like "Michel, John, Mary...".
I put the trait as "free-text" and "keywords".
I'm not sure if this process is correctly. So, I ask you:
does it matter the context like "My name is..." for the NLP? I mean...will it help the bot to predict that after this expression probably a fist name will come on?
is that right to add like 50 values to an entity or it's completly wrong?
what do you suggest as a training process in order to get the first name of someone?
You have done it right by keeping the entity's search strategy as "free-text" and "Keywords". But Adding keywords examples to the entity doesn't make any sense because a person's name is not a keyword.
So, I would recommend a training strategy which is as follows:
Create various templates of the message like, "My name is XYZ", "I am XYZ", "This is XYZ" etc. (all possible introduction messages you could think of)
Remove all keywords and expressions for the entity you created and add these two keywords:
"a b c d e f g h i j k l m n o p q r s t u v w x y z"
"XYZ" (can give any name but maintain this name same for validating the templates)
In the 'Understanding' tab enter the messages and extract the name into the entity ("contact_name" in your case) and validate them
Similarly, validate all message templates keeping the name as "XYZ"
After you have done this for all templates your bot will be able to recognise any name in a given template of the message
The logic behind this is your entity is a free-text and keyword which means it first tries to match the keyword if not matched it tries to find the word in the same position of the templates. Keeping the name same for validations helps to train the bot with the templates and learn the position where the name will be usually found.
Hope this works. I have tried this and worked for me. I am not sure how bot trains in background. I recommend you to start a new app and do this exercise.
Comment if there is any problem.
wit.ai has a pre-trained entity extraction method called wit/contact, which
Captures free text that's either the name or a clear reference to a
person, like "Paul", "Paul Smith", "my husband", "the dentist".
It works good even without any training data.
To read about the method refer to duckling.
I think I have a general problem with understanding the structure of matches and the scope in which variables of the match live.
The specific piece of code where I have the problem with is this:
// S sentiment toward A goodFor/badFor T
// => S sentiment toward the idea of A goodFor/badFor T
MATCH (S:A)-[:SOURCE]->(sent1:PS {type:"sentiment"})-[:TARGET]->(gfbf:E {type:"gfbf"}) , (A)-[:SOURCE]->(gfbf)-[:TARGET]->(T) , (Writer:A {type:"writer"})
// if there is some negative belief in any of the writers private state spaces that involve gfbf then inference is blocked
WHERE NOT (Writer)-[*1..]->({type:"believesTrue" , spec:FALSE})-[*1..]->(gfbf)
// if sent1 is in some private state spaces of the writer return all of these
OPTIONAL MATCH p=(Writer)-[*]->(sent1)
WITH NODES(p)[1..-1] AS ps_nodes
WHERE ALL(x IN ps_nodes[1..] WHERE LABELS(x) = "PS")
MERGE (S)-[:SOURCE]->(sent2:PS {type:"sentiment" , spec:(sent1.spec)})-[:TARGET]->(ideaOf:I {name:"ideaOf" , type:"ideaOf"})-[:TARGET]->(gfbf)
ON CREATE SET sent2.name =
CASE sent2.spec
WHEN FALSE THEN "-S"
ELSE "+S"
END
RETURN p
I think it's not relevant to understand what this is for. It suffices to see the structure I assume, but basically what it does is: It looks for a subgraph where there is path S-->sent1-->gfbf and also a path A-->gfbf-->T. If it finds that is makes a new path A-->sent2-->ideaOf-->gfbf, all he while setting the properties of the new nodes depending on the properties of the nodes from the match. Furthermore it looks whether it also has a path writer-->...-->sent where all nodes in the ... part have label PS. If it finds that path then it returns this for further operations in a different part of the program.
The error I am getting is this:
py2neo.cypher.error.statement.InvalidSyntax: sent1 not defined (line 6, column 58 (offset: 421))
"MERGE (S)-[:SOURCE]->(sent2:PS {type:"sentiment" , spec:(sent1.spec)})-[:TARGET]->(ideaOf:I {name:"ideaOf" , type:"ideaOf"})-[:TARGET]->(g"bf)
Why is sent1 no longer defined where I use it and how would I need to restructure the code to make it valid?
sent1 in isn't in the prior WITH - change it so:
WITH NODES(p)[1..-1] AS ps_nodes, sent1