Azure Machine Learning Studio - Stucked in Queue
Hi, I tried to run my experiment in Azure Learning Studio but it just get stuck at "Queue". I cancelled and ensured I just ran 1 experiment but I still get the same issue. Please help. Thanks.
Random forests on Azure GPU VM using the SDK
Can you please share any code examples for training random forests with GPU on Azure using libraries. I want to run on the multiple nodes.
Azure Stream Analytics: ML Service function call in cloud job results in no output events
Hey, I've got a problem with an Azure Stream Analytics (ASA) job that should call an Azure ML Service function to score the provided input data. The query was developed und tested in Visual Studio (VS) 2019 with the "Azure Data Lake and…
Azure ML Enpoint deployment failed EAST US region
I have an Azure ML Real-time inference endpoint deployed ran for a month till yesterday. Today it is in the state of "Failed". I did create a new compute and did a new deployment in the same region EAST US and it failed again. What's going?…
Feature request: module for selecting rows in a dataset
I'm looking for a module similar to Select Columns in Dataset but for rows. For example, one of the columns in my dataset is a string containing the quality control level. If the QC level equals some value, I would like to keep that row and discard the…
Question: how to define custom Model in Designer
In Azure ML Designer, I can't figure out how to define a custom Model. The builtin NN Regression module has many bugs (will open a separate issue for those), so I need to make my own custom model. The closest thing I've found so far is Create Python…
Question: what framework/language does Designer use behind the scenes?
I'm using Azure ML Designer, but it isn't clear what framework/language is being used behind the scenes. For example, deep learning is incredibly computation-heavy, so I would like to run any matrix multiplications on a GPU instead of on a CPU. For the…
Real-time endpoint response is empty
Hi, I've created and deployed NLP pipeline as a real-time endpoint, but the response that I get is an empty result ( {"Results":{}} ) with Status 200 OK. I have the Owner role, so i think that the permission is not a problem. I've also…
User program failed with ValueError: ZIP does not support timestamps before 1980
Hi i'm running into an issue in Azure ML- I'm getting the error "User program failed with ValueError: ZIP does not support timestamps before 1980" when it calls the when i run the following part of the code as a part of a pipeline from…
Time out after Succeded
Hi, I was trying to deploy my model as a web service after I run this command (Webservice.deploy_from_model), it said it was succeded. After that, I run this command (service.wait_for_deployment). this process told me the "Time out" condition.…
Azure Automated ML(interface) Does k-fold cross validation in autoML use just random sampling?
Is k-fold cross validation in automated ML(interface) stratified sampling or random sampling by default? I have ran several automated ML experiments using a training set with five data points for the least common class(say class A), and started to…
Question: distinction between Modules and Models in Designer
In Azure ML, under designer, there are 3 categories: Datasets Modules Models Datasets are pretty straightforward, but I don't understand the distinction between Modules and Models. As an ML researcher, when I think of a "model", I…
Bug report: can't parametrize # hidden node or momentum in NN regression model
I'm using the NN regression model in Azure ML Designer, but it appears to have a bug. If I switch trainer mode to ParameterRange, it allows me to use a semicolon-separated list of hyperparameters for learning rate and epochs. However, I'm unable to use a…
Feature request: more customizability in builtin models
In Azure ML Designer, I would like to customize and parametrize other aspects of the models. For example, in the NN Regression model, I would like to try various optimizers (Adam, Adagrad, SGD) and activation functions (ReLU, Sigmoid, Tanh).
Feature request: module for handling timestamps
In Azure ML Designer, there are dozens of builtin Modules for data transformation, but none of them can handle timestamps. What I'm looking for is a module in which the user: Selects which column of the input dataset contains a timestamp Selects…
Feature request: models only support single prediction column
Many machine learning tasks involve input data in the form of scalars, vectors, matrices, and higher-order tensors. Similarly, the output predictions are often in the same format. The builtin NN regression model in Azure ML Designer only supports scalar…
Bug report: Apply Math Operation doesn't cache and reuse results from previous run
In Azure ML Designer, I have a complex data preprocessing pipeline. One of the main features of Designer pipelines is that they are supposed to cache the results of each run, so that they don't need to be re-executed every time if nothing that they…
Azure Automated ML(interface) how do models created from an Automated ML experiment handle Imbalanced Data?
I have run automated ML experiments with imbalanced data (10:1, 20:1, sometimes 30:1) and deployed the best models which all showed fantastic results. When I looked up the link …
Why does Azure Machine Learning Studio (classic) take additonal time to execute python scripts?
Hi, I have been working with ML Studio (classic) and facing a problem with "Execute Python" scripts. I have noticed that it takes time to perform some internal tasks after which it starts executing the actual Python code in ML Studio. This…
Machine Learning Model Deployment
I am new to ML model and am researching using Azure Databricks and MLFlow to train a model. My question is once the model is created, is there a way to host the model that can be downloaded and inferenced remotely ? I am looking for options other than…