Fine Tuning a LLM
I am trying to fine tune a Language Model available on Azure Machine Learning Workspace. It involves two steps for data ingestion: data Uploading data mapping But after I am able to successfully upload data, I am not getting any functionality to map…
How to fix "ERROR: pip's dependency resolver does not currently take into account all the packages that are installed." while using Promt Flow.
I have made a custom image to use python's version 3.10.1 on Azure Ai Foundry 's Prompt Flow ( Earlier Ai Studio). The Flow is running successfully by ignoring the error. ERROR: pip's dependency resolver does not currently take into account all the…
Online Endpoint Deployment in Azure ML Studio
I was using the Azure Machine learning studio yesterday and my endpoints were deployed perfectly. I came back today and I am unable to test my endpoints because "V1 deployment testing is not supported." Does anyone know how to dix this issue
"ManagedIdentityCredential.get_token_info failed" in train.py (serverless)
In my Azure ML pipeline, I have a training step which uses a file called train.py to train my model. The issue is, whether I use DefaultAzureCredential() or ManagedIdentityCredential(), I get errors similar to the one…
Cannot get ML data set created by following the tutorial
Hi I am trying to create data set using the tutorial here https://zcusa.951200.xyz/en-us/azure/machine-learning/tutorial-explore-data?view=azureml-api-2 I cannot get the following code to run - It has an issue with a missing parameter I think in the …
Settings used from Microsoft Learn challenge but gpt-4to still shows exceeded token limit after one question
Hi guys, I follow the Microsoft Learn Challenge for "Trustworthy AI" Unfortunately every time I set up a gpt-4to playground I get the following error after just asking one question: "Requests to the ChatCompletions_Create Operation under…
AzureMLException: Error in Streaming from Input Data Sources
An AzureMLException is encountered with the message "Error Code: ScriptExecution.StreamAccess.Unexpected Native Error: error in streaming from input data sources" when using the 'Apply SQL Transformation' component in the Azure ML designer…
Mlflow and Spark
Hi, I encountered the following problem when I tried to use a model for spark inference (via mlflow.pyfunc.spark_udf) that I had previously trained in pandas and saved in mlflow. I saved a model via from mlflow.tracking import MlflowClient from…
"Error 'Resource provider not registered' when deploying model in Azure ML Tutorial (Free Trial)"
Hi, I'm following the Microsoft Learn tutorial "Create a classification model with Azure Machine Learning designer" (https://microsoftlearning.github.io/mslearn-ai-fundamentals/Instructions/Labs/01-machine-learning.html) with an Azure Free…
Set up as Workspace Owner and Data Scientist role, but cannot create notebooks/write files
I just set up a Workspace under my Subscription, and am the Owner on both. I'm trying to work in Azure ML in this Workspace, where I have a Azure ML Data Scientist role for myself as well, and keep getting this message when I try to create a new…
MLflow Model Deployment Error - Can't Find MLmodel File in Registered Model
Hi, I'm trying to deploy a registered MLflow model in Azure ML. I've already fixed the environment issues by creating a custom environment with the correct dependencies for GPU support and tensor parallelism. However, now I'm getting an error in the…
Error training model in ML Studio
Hello, I get the following error while running an "Automated ML" job in ML Studio. Failed to download snapshot from storage using SAS url. Error message: This request is not authorized to perform this operation. Authentication error when…
How to correct Error Code: ScriptExecution.StreamAccess.Unexpected
I am using AMLS for a class. First week of usage...no issues. Second week of usage...no loaded csv file is recognized by any node. I continue to receive the above error, even with a csv file containing 3 columns and 15 rows of data. I can read the data…
Azure ML endpoint - getting predict_proba output
Hi, I've created AutoML job to do binary classification and registered some models to create endpoints, the endpoints are not producing predic_probabilities outputs event when I'm passing to json "GlobalParameters": { "method":…
AzureMLCompute job failed 500: [REDACTED]: Some(true) Error while creating custom environment in azure ml
Hello everyone, I am trying to create a custom environment to train and deploy a catboost regression model in azure ml SDK. However when I submit the job, it's running for a while and then throws "AzureMLCompute job failed 500: [REDACTED]:…
I get the following error message on datasets that worked previously, how do I fix this problem?
AzureMLException: Message: Error Code: ScriptExecution.StreamAccess.Unexpected Native Error: error in streaming from input data sources StreamError(Unknown("Dataflow at inmemory://dataflow/dea9ef925fd2e0f84b13b7cb3c8aa074 is not valid.",…
Logging in Python SDK
Hi, I ask whether these differences exist in the Python SDK between logging in training and logging after deployment: https://zcusa.951200.xyz/en-us/azure/machine-learning/how-to-track-experiments-mlflow?view=azureml-api-2 "The Azure Machine…
How I can update the version of python using flow.dag.yaml file?
While using Azure Ai Foundary's Prompt Flow when the compute session needs to be set up it asks for an image it can be provided by azure ( tick mark using Latest image), or a custom image can be provided here. In the image provided by azure that version…
Unable to deploy fine-tuned Phi 3.5 MoE model on NC24ads_A100_v4 despite successful training on NC96ads_A100_v4
Hello, I'm trying to deploy a fine-tuned Phi 3.5 MoE model as a real-time endpoint in Azure ML. I've encountered an interesting situation: Training Environment: Successfully fine-tuned on NC96ads_A100_v4 (96-core machine) Using DeepSpeed with tensor…
HttpResponseError: Operation returned an invalid status 'OK' - Deploying Pipeline Component to BatchEndpoint
Hi everyone, I try to deploy a pipeline component from custom azure machine learning registry to a batch endpoint. The batch endpoint was created successfully. with the following code: # connection to prod workspace ml_client_dev = MLClient( …