I'm trying to load Encog created network in C# code:
private void buttonImportNetwork_Click(object sender, EventArgs e)
{
NeuralNetworkEffect.Network =
(BasicNetwork) EncogDirectoryPersistence.LoadObject(new FileInfo(_networkFilePath));
}
I'm getting System.NullReferenceException. Here is the call stack:
"at Encog.Neural.Networks.PersistBasicNetwork.Read(Stream mask0)
at Encog.Persist.EncogDirectoryPersistence.LoadObject(Stream mask0)
at Encog.Persist.EncogDirectoryPersistence.LoadObject(FileInfo file)"
I tried multiple times with various files and paths, result the same.
What am I doing wrong, did anybody encounter this and solved the issue?
OK, I found the reason. The EG file containing network definition produced by Encog Workbench contained invalid type for activation function, probably compatible with Java version of Encog package, but not with .Net version.
In the EG file it was:
"org.encog.engine.network.activation.ActivationTANH"
but it should be:
"ActivationTANH".
After removing Java namespaces the network was loaded without issues.
Related
I tried to use distilbert classifier. but I am getting the following error.
This is my code
(X_train,y_train),(X_test,y_test),prepro
=text.texts_from_df(train_df=data_train,text_column="Cleaned",label_columns=col
,val_df=data_test,maxlen=500,preprocess_mode="distilbert")
and here is the error
OSError: Model name 'distilbert-base-uncased' was not found in tokenizers model name list (distilbert-base-uncased, distilbert-base-uncased-distilled-squad, distilbert-base-cased, distilbert-base-cased-distilled-squad, distilbert-base-german-cased, distilbert-base-multilingual-cased). We assumed 'distilbert-base-uncased' was a path, a model identifier, or url to a directory containing vocabulary files named ['vocab.txt'] but couldn't find such vocabulary files at this path or url._
Due to my office current environmental issue, I can only work on tf 2.2 and python 3.8. Right now I am using 0.19.
Do you think it will affect my current environment if I downgrade it to 0.16?
This error may happen if there is a network or firewall issue preventing download of the tokenizer files. See this FAQ entry for remedies.
Also, when you use preprocess_mode='distilbert', texts_from* functions return TransformerDataset instances, not arrays. You'll need to replace (X_train, y_train) with train_data, for example. See this example notebook.
I have written a program in c++11. My intention is to use MATLAB environment for plotting.
So, I followed these two links to have some idea:MATLAB_Build C++ Engine programs and c++ start MATLAB session. (There must be other ways to interface MATLAB and c++. But I want to know where is my fault before trying other ones)
As those above links have suggested I built my project accordingly using visual studio '17. The initial errors that I found was, my application could not locate the following .dll files libMatlabEngine.dll, libMatlabdataArray.lib.
So, I moved my .dlls to the executable location as suggested here. But after then my code is throwing the following error: 0xC0000005: Access violation executing location 0x0000000000000000.
Below is the piece of code that I'm trying. I guess it's the startMATLAB() which is returning a NULL pointer and the code is failing for that. But I can't figure out what's the reason.
Appreciate your help.
//Start the MATLAB engine
using namespace matlab::engine;
using namespace matlab::data;
unique_ptr<MATLABEngine> matlabptr = startMATLAB();
//Cretae a matlab array factory
ArrayFactory factory;
auto inputArray = factory.createArray({ 1, 6 }, cppData.cbegin(), cppData.cend());
auto time_array = factory.createArray({ 1, 6 }, time.cbegin(), time.cend());
matlabptr->setVariable(convertUTF8StringToUTF16String("data"), move(inputArray));
matlabptr->setVariable(convertUTF8StringToUTF16String("time"), move(time_array));
matlabptr->eval(convertUTF8StringToUTF16String("plot(time, data)"));
I am using embedded Neo4j, version 3.0.3. Following this guide, I have created Neo4j/Java code. It creates a database, adds two nodes (one for java, one for scala) and adds a relationship.
package examples;
import java.io.File;
import org.neo4j.graphdb.*;
import org.neo4j.graphdb.factory.GraphDatabaseFactory;
public class HelloWorld {
public static void main(String[] args) {
GraphDatabaseFactory dbFactory = new GraphDatabaseFactory();
GraphDatabaseService db = dbFactory.newEmbeddedDatabase(new File("Test_DB"));
try (Transaction tx = db.beginTx()) {
Node javaNode = db.createNode(Tutorials.JAVA);
javaNode.setProperty("TutorialID", "JAVA001");
Node scalaNode = db.createNode(Tutorials.SCALA);
scalaNode.setProperty("TutorialID", "SCALA001");
Relationship relationship = javaNode.createRelationshipTo(scalaNode, TutorialRelationships.JVM_LANGUAGES);
relationship.setProperty("Id", "1234");
tx.success();
}
}
}
enum Tutorials implements Label {
JAVA, SCALA, SQL, NEO4J;
}
enum TutorialRelationships implements RelationshipType {
JVM_LANGUAGES, NON_JVM_LANGUAGES;
}
I program using Eclipse, so all the libraries are imported and I can just click the 'run' button on Eclipse to get the code running, and it seems to work without any issues. Upon running the code, I now have a folder Test_DB in the ~/workspace/project_name/Test_DB directory, where project_name is the name of the overall Eclipse folder. My goal is now to visualize this database in a web browser. The guide I linked to earlier shows an example of this; the user was able to look at the nodes in the web browser (see the bottom of the webpage). Unfortunately, I am using a Linux computer with Firefox, and that tutorial was in Windows, and I can't figure out how to get the visualization.
There have been a few other questions related to this. Unfortunately, some of them (such as this one) propose using software other than the default visualization. I don't own the computer and I have to go through a roundabout process to get external code installed. To be clear what I mean, this link discusses the default Neo4j browser. This is what I would like to see.
This question here directly tackles the same issue, and in fact, it uses the exact same tutorial I used! The answer proposes changing the path in the neo4j-server.properties file. Unfortunately, that file doesn't exist, and upon further analysis, it seems like Neo4j 3.0 changed the configuration naming, which I found out by reading the answer to this similar question. There is now a file conf/neo4j.conf with this information. I entered the following information in the first few lines, keeping the other settings the default:
# The name of the database to mount
dbms.active_database=Test_DB
# Paths of directories in the installation.
dbms.directories.data=/home/username/workspace/project_name/
This does not appear to work. Am I using these settings correctly? When I open the neo4j web browser after running ./bin/neo4j start and click on the database symbol in the left hand side, I see "Name: Test_DB", but it also says there are no nodes and no relationships in the database, and returning a match all query provides nothing. Is it possible for the browser to connect to my database so it can see the nodes (e.g., the two nodes in my Java code above)?
Or is it that I'm not using this code correctly; does the code somehow have to avoid quitting (i.e., replace tx.success() with something else?) to keep the data there?
Sorry about answering my own question, but I finally figured out how to do this! Here's what happens: according to the github change log for 3.0.0.RC-1:
Databases are now stored in a directory called databases under the directory specified in dbms.directories.data
So what we actually have to do is make sure our data base is in the following location:
/home/username/workspace/project_name/databases/
The issue is that when we run it in Eclipse, we get the database in the following folder:
/home/username/workspace/project_name/
Thus, the solution is to make sure the new database folder is preceded by a databases name, i.e., I would change one line to:
GraphDatabaseService db = dbFactory.newEmbeddedDatabase(new File("databases/Test_DB"));
Using pentaho 5.3, Modified Java Script Value step to inject those data.
I want to dynamically set path and names in the MongoDB Output step. Here is my code
var meta = new org.pentaho.di.trans.TransMeta( source_path );
var mongoStep = meta.findStep("MongoDB Output");
mongoStep.setDescription('This is MongoDB Output by Ray');
Alert(mongoStep.getName()); // code is ok until here.
var mongoStepMeta = mongoStep.getStepMetaInterface() // error occurs here
When I want to get the getStepMetaInterface() to use step functions, the error occurs.
java.lang.LinkageError: loader constraint violation: loader (instance of org/pentaho/di/core/plugins/KettleURLClassLoader) previously initiated loading for a different type with name "org/pentaho/metastore/api/IMetaStore"
This error seems to be generated by the violation of .jar.
But when I use those original steps, like "Microsoft Access Input", I can successfully get getStepMetaInterface(). In this way, I can use all the functions defined in AccessInputMeta.java.
From my point of view, this problem may have a relation to the MongoDB Output step, because this is a plugin for kettle, but I am not sure.
Any response is appreciated!!
Yep, it's because of the plugin system. The Modified Java Script step and the Access Input step (among many others) are all in the Kettle engine and share a classloader. External plugins (like MongoDB Input/Output) have their own isolated classloaders. You have to do some voodoo with thread context classloaders and reflection in order to get at the Meta classes for those steps.
Here's an example of the hoops you have to jump through to get at external plugins (the Big Data plugin in this example) from the Modified Java Script step:
https://github.com/brosander/pentaho-hadoop-shims/blob/verification-2/test/verification/jobs/dependencies/ForceHiveToConnectRemotely.ktr
First, thanks to Mass who solves the problem.
Here is his advice:
This problem is caused by jar file conflict.
In pentaho installation there are two jar files:
-lib/metastore-5.3.0.0-213.jar
-plugins/pentaho-mongodb-plugin/lib/metastore-5.3.0.0-213.jar
Just remove the last jar file, this error will disappear.
This question already has answers here:
MetadataException: Unable to load the specified metadata resource
(47 answers)
Closed 6 years ago.
This code:
using (EntityConnection conn = new EntityConnection("name=ELSCommonEntities"))
{
conn.Open();
}
Gives me the following error:
Test method ELS.Service.Business.IntegrationTest.Base.ServiceBaseIntegrationTest.StartLoggingTestMethod threw exception: System.Data.MetadataException: Unable to load the specified metadata resource..
With the following stack trace:
System.Data.Metadata.Edm.MetadataArtifactLoaderCompositeResource.LoadResources(String assemblyName, String resourceName, ICollection`1 uriRegistry, MetadataArtifactAssemblyResolver resolver)
System.Data.Metadata.Edm.MetadataArtifactLoaderCompositeResource.CreateResourceLoader(String path, ExtensionCheck extensionCheck, String validExtension, ICollection`1 uriRegistry, MetadataArtifactAssemblyResolver resolver)
System.Data.Metadata.Edm.MetadataArtifactLoader.Create(String path, ExtensionCheck extensionCheck, String validExtension, ICollection`1 uriRegistry, MetadataArtifactAssemblyResolver resolver)
System.Data.EntityClient.EntityConnection.SplitPaths(String paths)
System.Data.EntityClient.EntityConnection.GetMetadataWorkspace(Boolean initializeAllCollections)
System.Data.EntityClient.EntityConnection.InitializeMetadata(DbConnection newConnection, DbConnection originalConnection, Boolean closeOriginalConnectionOnFailure)
System.Data.EntityClient.EntityConnection.Open()
ELS.Service.Business.Base.ServiceBase.StartLogging(String userWindowsLogon) in C:\C-TOM\ELS-RELEASE1\ELS.Service.Business\Base\ServiceBase.cs: line 98
ELS.Service.Business.IntegrationTest.Base.ServiceBaseIntegrationTest.StartLoggingTestMethod() in C:\C-TOM\ELS-RELEASE1\ELS.Service.Business.IntegrationTest\Base\ServiceBaseIntegrationTest.cs: line 65
However, this code which uses the same connection string:
using (ELSCommonEntities db = new ELSCommonEntities())
{
var res = from c in db.Logging
select c;
int i = res.Count();
}
Does not give an error.
The connection string is:
<add name="ELSCommonEntities" connectionString="metadata=res://*/Common.CommonModel.csdl|res://*/Common.CommonModel.ssdl|res://*/Common.CommonModel.msl;provider=System.Data.SqlClient;provider connection string="Data Source=localhost;Initial Catalog=els5_demo;Integrated Security=True;MultipleActiveResultSets=True"" providerName="System.Data.EntityClient" />
I have also opened up the dll in reflector and the metadata looks ok.
Found the problem.
The standard metadata string looks like this:
metadata=res://*/Model.csdl|res://*/Model.ssdl|res://*/Model.msl
And this works fine in most cases. However, in some (including mine) Entity Framework get confused and does not know which dll to look in. Therefore, change the metadata string to:
metadata=res://nameOfDll/Model.csdl|res://nameOfDll/Model.ssdl|res://nameOfDll/Model.msl
And it will work. It was this link that got me on the right track:
http://itstu.blogspot.com/2008/07/to-load-specified-metadata-resource.html
Although I had the oposite problem, did not work in unit test, but worked in service.
I had the same error message, and the problem was also the metadata part of the connection string, but I had to dig a little deeper to solve it and wanted to share this little nugget:
The metadata string is made up of three sections that each look like this:
res://
(assembly)/
(model name).(ext)
Where ext is "csdl", "ssdl", and "msl".
For most people, assembly can probably be "*", which seems to indicate that all loaded assemblies will be searched (I haven't done a huge amount of testing of this).
This part wasn't an issue for me, so I can't comment on whether you need the assembly name or file name (i.e., with or without ".dll"), though I have seen both suggested.
The model name part should be the name and namespace of your .edmx file, relative to your assembly.
So if you have a My.DataAccess assembly and you create DataModels.edmx in a Models folder, its full name is My.DataAccess.Models.DataModels. In this case, you would have "Models.DataModels.(ext)" in your metadata.
If you ever move or rename your .edmx file, you will need to update your metadata string manually (in my experience), and remembering to change the relative namespace will save a few headaches.
There are several possible catches.
I think that the most common error is in this part of the connection string:
res://xxx/yyy.csdl|res://xxx/yyy.ssdl|res://xxx/yyy.msl;
This is no magic. Once you understand what is stands for you'll get the connection string right.
First the xxx part. That's nothing else than an assembly name where you defined you EF context clas. Usually it would be something like MyProject.Data.
Default value is * which stands for all loaded assemblies. It's always better to specify a particular assembly name.
Now the yyy part. That's a resource name in the xxx assembly.
It will usually be something like a relative path to your .edmx file with dots instead of slashes. E.g. Models/Catalog - Models.Catalog
The easiest way to get the correct string for your application is to build the xxx assembly. Then open the assembly dll file in a text editor (I prefer the Total Commander's default viewer) and search for ".csdl". Usually there won't be more than 1 occurence of that string.
Your final EF connection string may look like this:
res://MyProject.Data/Models.Catalog.DataContext.csdl|res://MyProject.Data/Models.Catalog.DataContext.ssdl|res://MyProject.Data/Models.Catalog.DataContext.msl;
As Shiraz Bhaiji answered, the metadata=res:///Model.csdl|res:///Model.ssdl|res://*/Model.msl was the case.
However I still had problems with constructing the proper string based on my Model localization, namespaces and assemby name. The very simple solution was to rename the .edmx file in Visual Studio(after than rename and get back to the original name), which triggered the automatic refreshing of the string in my Web.config
I had the same problem with three projects in one solution and all of the suggestions didn't work until I made a reference in the reference file of the web site project to the project where the edmx file sits.
I moved my Database First DataModel to a different project midway through development. Poor planning (or lack there of) on my part.
Initially I had a solution with one project. Then I added another project to the solution and recreated my Database First DataModel from the Sql Server Dataase.
To fix the problem - MetadataException when using Entity Framework Entity Connection. I copied my the ConnectionString from the new Project Web.Config to the original project Web.Config. However, this occurred after I updated my all the references in the original project to new DataModel project.
It might just be a connection string error, which is solved by the above process, but if you are using the dll's in multiple projects then making sure the connection string is named properly will fix the error for sure.
I had this problem when moving my .edmx database first model from one project to another.
I simply did the following:
Deleted the connection strings in the app.config or web.config
Deleted the 'Model.edmx'
Re-added the model to the project.