How does the BERT model (in Tensorflow or Paddle-paddle frameworks) relate to nodes of the underlying neural-net that’s being trained?

The BERT model in frameworks like TensorFlow/Paddle-paddle shows various kinds of computation nodes (like subtract, accumulate, add, mult etc) in a graph like form in 12 layers.

But this graph doesn’t look anything like a neural-network, one that’s typically shown in textbooks (e.g. like this where each edge has a weight that’s being trained and there is an input layer and output layer.

Instead, when I print out the BERT graph, I can’t figure out how a node in the BERT graph relates to a node in the neural-network that’s being trained.

I have been using the BERT framework models to compile them to a form where we can run the model on a PC/CPU. But I still lack this basic aspect of how BERT relates to neural-net as I don’t see which neural-network topology is being trained (as i’d expect topology/connections between/among various layers/nodes of the neural-net dictate how training of the neural net occurs).

Could someone explain what underlying neural-net is being trained by BERT? How do nodes in the BERT graph relate to neural-net nodes and weights on neural-net edges?

Pros and Cons of Different Action/Turn Frameworks

I would like to have a clearer understanding of the pros and cons of various frameworks of turn composition and action economies. That is, I would like to see the relative benefits and drawbacks of different ways game systems handle answering the question about what actions a character can take in a turn (and how many thereof, or how their efficiency is influenced by their number etc.).

Now, of course the lines between framework will not always be sharp, but I think I can post a (non-exhaustive) list of what seems like some of the common paradigms or families thereof, in the hopes that this will better scope the question. Also please note that for the sake of maintaining focus and preventing tangents, I’m not asking about systems of initiative (who goes first), nor about comparing systems with vs. without a way of increasing the number of allowed actions above the default.

Freeform Composition / Build Your Own Turn

Games which give the players the basic ground rules for deciding which actions to fit into a turn, each turn. Often defence and movement are treated just as any other action you can take. Examples would be older books in the Storyteller / Classic World of Darkness (e.g. VtM 2e), where you can Split your dice pool into as many actions as you need and can afford to (e.g. into three actions for performing two swings of your tyre iron [attack] and one dodge [defence]), or Fallout PnP (where you get an allotment of Action Points, and actions cost some number thereof, and one AP is also the cost of moving 1m or boosting defence by a certain value, though ‘default’ defences are free).

This framework seems to be equally capable of supporting the assumption that everyone gets defences for free and the assumption that defences are actions like any other.

Type-Specific Action Slots

Games where a turn usually consists of a certain number (either fixed or upgradeable) of ‘slots’ of specific types, e.g. one move action, one ‘do your thing’ action that can be used for a variety of things (attack, move again, do a noncombat action etc.), and one or two reflex actions (often used for defences if the system puts a limit on the number of defences). Sometimes a systems allows choosing between several different sets of slots. Examples would be Dark Heresy and Eclipse Phase.

This framework also seems to be equally capable of supporting the assumption that everyone gets defences for free and the assumption that defences are actions like any other.

Turn Structure Templates

Games where a player chooses one of a limited, often rather specific sets of actions for a turn, often with specific modifiers baked in. My go-to example would be GURPS, where you can choose e.g. Move and Attack (full movement, poor attack, slightly compromised defence), Move (full move, no attack), Attack (single step move, normal defence, normal attack), All-Out Attack (half of maximum move, attack with a bonus of choice, no defence at all), Committed Attack (attack with a modest bonus, seriously compromised defence, one- or two-step movement), Ready (get a special non-attack action, no attack, one step of movement, and normal defence), and some others.

This framework seems imply a strict regulation of what defences a given turn action structure template includes, and I think it predisposes one to lean towards having a ‘with defences’ structure as the more default one.

Other Frameworks

I’m sure there are other framework that are distinct from the above three, but I’m currently not able to think of an example I’m well-acquainted with nor a generalised description. A comparison of any such additional frameworks is also welcome if someone is open to including them in an analysis.

Legal/Performace ramifications of using outdated frameworks & unsupported languages?

I hope this is the best place for this – I would like to know how people deal with companies that use outdated frameworks or languages that are no longer supported and have reached end-of-life.

I know someone who is working for a firm that are using a very outdated PHP framework. Zend 1.x to be precise which was end-of-life in 2016 and their custom Zend framework hasn’t been updated for approx 7-8 years.

As these frameworks no longer receive any more support the chances of getting hacked increase not to mention the general performance decreases with outdated versions)

How would I approach this colleague as I feel using all the outdated stuff will just burn them in the ass – how do I best approach this?

Is it advisable for software developers to focus on frameworks like React or Vue?

As a front end developer just 2 years into my career, I am starting to be confused by the job industry nowadays where the job recruiters are looking for people who had experience with React or Angular for X amount of years. I have read what people wrote online saying that do not focus too much on frameworks as they come and go, rather focus on picking up OOP languages like Java or Python.

For front end developers, this is a very difficult situation because if frameworks come and go, that means our ROI learning and working on those frameworks will be disappointing. So would that mean back end developers will have an edge in the industry because what they learn (database, backend OOP languages)would still be practical 10 years down the line?

Should I switch my job to become a backend developer then? Or rather instead of focusing on learning frameworks, I should just focus on learning a backend programming language then (either Python, Java, Golang)? Really need advise as a newbie programmer here.

How to validate usability of software frameworks

My thesis will produce a software framework that can be used by programmers to solve problems in a paticular domain. One of the case studies I want to do to verify the contribution I’m making is to have some measure of usability of the framework. I’m looking for techniques to test a software framework that will be used by other programmers, not so much end users. When including ‘usability’ in my searches the results lead to end-user UX/UI usability.

I ultimately want to run a test where one or a few programmer try to use the framework on a given problem. To help make sure the test has usable results I want to plan the strategy before hand and first consider existing techniques. But finding existing research or tests about usability of code libraries/frameworks is difficult to find.

Please let me know if you can think of a way to validate the usability of my software framework.

Thanks for reading my question

Subtrair DateTime já persistida com DateTime Atual juntamente com condição de igualdade entre tags com Entity Frameworks


Acabo de cadastrar os seguintes dados.

Tag: 19 e Data: 04/06/2019 13:00

Agora se por ventura tiver que cadastrar esta Tag: 19 novamente só poderei depois de 60 minutos, sendo assim

Tag: 19 e Data: 04/06/2019 14:01

A ideia é esta logo abaixo:

Se (((04/06/2019 13:00 - 04/06/2019 13:55) < 60) && (19 == 19)) {     //Não posso recadastrar a passagem da Tag 19 pois à subtração das datas é menor que 60 minutos } 

Fiz algumas tentativa com o Entity mas não obtive sucesso.

if (_context.OrdensChegadas.Where((e => e.Data.Subtract(DateTime.Now)) < 60       & (e.Tag.Equals(ordemChegada.Tag))) {    //Não realizo cadastro  } 

I have Visual Studio 2019 and have installed all the developer kits/targeting frameworks looking to use .Net 4.6.1 but have issues

I have Visual Studio 2019 and have installed all the developer kits/targeting frameworks looking to use .Net 4.6.1 for entity framework 6.2.0 but when I go to the project properties I do not see anything besides .Net Core 2.2 in the targeting frameworks? I could use some help figuring out what I am doing wrong.

Visual Studio 2019