FRom the documentation I found that after contruction of a model, the weights are initialized by calling TFNetwork.initialize_params. I am wondering if there is an API to extract these weights?
There is TFNetwork.save_params_to_file. Is that what you want?
Otherwise, you can simply access all the params individually. E.g. via TFNetwork.get_trainable_params, or TFNetwork.get_param_values_dict, or TFNetwork.get_layer(layer_name).params, etc.
Related
Let's say I have a dynamic list of items which I upload from a webhook at a certain step of a scenario. This list is not a list of some type of pre-defined entities, this is just a list of something.
So,
$session.params.items = ["item_1", "item_2"]
Now, at the next step user can either choose one of the options provided in this list or make something else. It seems that the best way to check whether the next user query has something to do with our list of items, I need to declare a non-required parameter. But if I do this way, they necessarily require an entity type.
So, the question is:
Is is possible to use any list or dict from session parameters as an entity type?
What is the best way to implement checking whether user query falls into one of the session variables like list taking into account that user query may as well go to other intent and so on. I supposed the best way is to have routes for filling in unnecessary parameters and for intents, but still I can't do the first part.
I want to pass the value from one analysis to another analysis. Can I use presentation variable for this purpose? If not what else can I use
A presentation variable can be set by prompts, not an analysis. To pass a value it depends on the kind of action you have in place to go from analysis A to B. If it's a custom built link you can add the value as filter in the URL, if it's a navigation you can use the columns of the analysis A which are sent automatically by the link etc.
Keeping it short: no, your analysis A can't set a presentation variable for analysis B, you must use a prompt to set it or use an alternative way of sending values via the navigation.
When we design Rest apis, it is said that use Path Variable when you need to identify a resource and Request Parameters when you need to do operations like sorting, filtering, searching, pagination. Let us take a scenario of Employee:
Employee has three fields like name, companyName, socialSecurityNo.
Now i want an Employee with a socialSecurityNo = ABC.
It seems fine to have endpoint with path variable like /employees/{socialSecurtityNo}, since we are identifying a resource.
Also it seems intuitive that we are filtering on the basis of socialSecurityNo and have an endpoint like /employees?socialSecurityNo=ABC
What will be the right way as i am confused and think that both apply.
It's a good question.
/employees?socialSecurityNo=ABC
is filtering all employees on socialSecurityNo. If socialSecurityNo is unique to an employee, there's no point in this endpoint existing and a client should use /employees/{socialSecurtityNo}.
There's nothing wrong with filtering on a unique field value (socialSecurtityNo) and if a client finds it easier to use this version (for whatever technical reason) then that's fine. There is no 'right' way. The ultimate reason APIs exist is to allow valuable work to be done by a client. Work with the client to allow that to happen but keep best practice in mind and know when the solution isn't the best but is the most practical in the situation.
I would expect to see:
/employees?surname=Smith
as this is filtering on a non unique field value and should return a collection of Employee objects.
The 'right' thing to do is keep the results consistent. If you have both ways of finding an employee, make sure the returned result is the same in each case.
When I call communityService.getMyCommunities(params) or communityService.getPublicCommunities(params) or communityService.getSubCommunities(parentCommunity, params) I would expect that filling params with e. g. tags=[mytag,yourtag] the call would only lookup communities having at least one of these tags (or both, however).
But to me it looks like this param ("tags") is simply ignored, and I always receive all communities of the given category (my / public / sub).
In case of having lots of communities of the requested category this massively slows down performance when I only want to retrieve communities with e. g. one certain tag: I receive all data over the net and must filter / lookup the received object list locally.
What am I doing wrong?
Is there something missing in the SDK implementation?
As part of the communities/my api, you cannot do any filtering... you need to use Search APIs.
In order to get a filtered list of communities based on the tags, you need to make a request to the following URL.
https://apps.na.collabserv.com/search/atom/mysearch?scope=personalOnly&scope=communities&query=&constraint={%22type%22%3A%22category%22%2C%22values%22%3A[%22Tag%2Fprb%22]}&page=1&pageSize=10
Yes, it is URL encoded, you can then change prb to match your tag, and you can repeat the constraints for each tag
You can also reference Link to Search API Constraints
I want to employ CakePHP's basic caching functionality on my site's home page. However, there is one element on the page that should display different data depending on the visitor's location, as determined by their IP address.
You can't wrap <cake:nocache> around variables that are set in the controller, which is where I was previously determining the location and getting the data. My question therefore is: where can I optimally set a (session?) variable to contain visitor location information before the controller? How can I use this information to populate an array of data for the nocached portion of the view, while completely sidestepping the controller action, which is no longer being called?
Any advice greatly appreciated!
Hmm, well, apparently CakePHP questions aren't of much interest to the world at large: only 8 views in 2 days :(
In any case I investigated a bit further and discovered that, while the <nocache> tags don't let you surround variables to make them dynamic, they DO allow you to make non-caching calls to elements or helpers. I therefore extracted the relevant part of my page into an element, and populated the data array by calling a helper function.
This did mean that I had to access the model from inside the helper with
$this->Modelname =& ClassRegistry::init("Modelname");
and I'm not sure this is necessarily the respectful CakePHP and/or MVC way of doing things, but at least it's some way towards a solution!