Use uv tool for isolate cicd env #2597

bopeng1234 · 2025-03-03T07:27:28Z

Existing Sample Changes

Description

The UV tool is designed to streamline the management of Python environments for multiple test cases. One of its standout features is its ability to operate without altering the existing Python environment, ensuring that the each unit test sample (sample.json) won't affect primary setup (AI Tool did with conda environment). This makes it an ideal solution without the hassle of environment conflicts or dependencies issues, and makes each sample isolation.

add uv tool management for samples:

1.AI-and-Analytics/End-to-end-Workloads/JobRecommendationSystem 
2.AI-and-Analytics/Features-and Functionality/INC_QuantizationAwareTraining_TextClassification 
3.AI-and-Analytics/Features-and-Functionality/Intel_Extension_For_SKLearn_Performance_SVC_Adult 
4.AI-and-Analytics/Features-and-Functionality/IntelPython_daal4py_DistributedKMeans 
5.AI-and-Analytics/Features-and-Functionality/IntelPython_daal4py_DistributedLinearRegression 
6.AI-and-Analytics/Features-and-Functionality/IntelPython_GPU_dpnp_Genetic_Algorithm 
7.AI-and-Analytics/Features-and-Functionality/IntelPython_Numpy_Numba_dpnp_kNN 
8.AI-and-Analytics/Features-and-Functionality/IntelPython_XGBoost_Performance 
9.AI-and-Analytics/Features-and-Functionality/IntelPyTorch_GPU_InferenceOptimization_with_AMP 
10.AI-and-Analytics/Features-and-Functionality/IntelPyTorch_TrainingOptimizations_AMX_BF16 
11.AI-and-Analytics/Features-and-Functionality/IntelTensorFlow_AMX_BF16_Inference 
12.AI-and-Analytics/Features-and-Functionality/IntelTensorFlow_AMX_BF16_Training 
13.AI-and-Analytics/Features-and-Functionality/IntelTensorFlow_Enabling_Auto_Mixed_Precision_for_TransferLearning 
14.AI-and-Analytics/Features-and-Functionality/IntelTensorFlow_for_LLMs 
15.AI-and-Analytics/Features-and-Functionality/IntelTensorFlow_TextGeneration_with_LSTM 
16.AI-and-Analytics/Features-and-Functionality/IntelTransformers_Quantization 
17.AI-and-Analytics/Getting-Started-Samples/INC-Quantization-Sample-for-PyTorch 
18.AI-and-Analytics/Getting-Started-Samples/INC-Sample-for-Tensorflow 
19.AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_SKLearn_GettingStarted 
20.AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_TensorFlow_GettingStarted 
21.AI-and-Analytics/Getting-Started-Samples/IntelPython_daal4py_GettingStarted 
22.AI-and-Analytics/Getting-Started-Samples/IntelPython_XGBoost_GettingStarted 
23.AI-and-Analytics/Getting-Started-Samples/Modin_GettingStarted 
24.AI-and-Analytics/Getting-Started-Samples/Modin_Vs_Pandas

Type of change

Please delete options that are not relevant. Add a 'X' to the one that is applicable.

Bug fix (non-breaking change which fixes an issue)
New feature (non-breaking change which adds functionality)
Implement fixes for ONSAM Jiras

ONSAM 1917

1.AI-and-Analytics/End-to-end-Workloads/JobRecommendationSystem 2.AI-and-Analytics/Features-and Functionality/INC_QuantizationAwareTraining_TextClassification 3.AI-and-Analytics/Features-and-Functionality/Intel_Extension_For_SKLearn_Performance_SVC_Adult 4.AI-and-Analytics/Features-and-Functionality/IntelPython_daal4py_DistributedKMeans 5.AI-and-Analytics/Features-and-Functionality/IntelPython_daal4py_DistributedLinearRegression 6.AI-and-Analytics/Features-and-Functionality/IntelPython_GPU_dpnp_Genetic_Algorithm 7.AI-and-Analytics/Features-and-Functionality/IntelPython_Numpy_Numba_dpnp_kNN 8.AI-and-Analytics/Features-and-Functionality/IntelPython_XGBoost_Performance 9.AI-and-Analytics/Features-and-Functionality/IntelPyTorch_GPU_InferenceOptimization_with_AMP 10.AI-and-Analytics/Features-and-Functionality/IntelPyTorch_TrainingOptimizations_AMX_BF16 11.AI-and-Analytics/Features-and-Functionality/IntelTensorFlow_AMX_BF16_Inference 12.AI-and-Analytics/Features-and-Functionality/IntelTensorFlow_AMX_BF16_Training 13.AI-and-Analytics/Features-and-Functionality/IntelTensorFlow_Enabling_Auto_Mixed_Precision_for_TransferLearning 14.AI-and-Analytics/Features-and-Functionality/IntelTensorFlow_for_LLMs 15.AI-and-Analytics/Features-and-Functionality/IntelTensorFlow_TextGeneration_with_LSTM 16.AI-and-Analytics/Features-and-Functionality/IntelTransformers_Quantization 17.AI-and-Analytics/Getting-Started-Samples/INC-Quantization-Sample-for-PyTorch 18.AI-and-Analytics/Getting-Started-Samples/INC-Sample-for-Tensorflow 19.AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_SKLearn_GettingStarted 20.AI-and-Analytics/Getting-Started-Samples/Intel_Extension_For_TensorFlow_GettingStarted 21.AI-and-Analytics/Getting-Started-Samples/IntelPython_daal4py_GettingStarted 22.AI-and-Analytics/Getting-Started-Samples/IntelPython_XGBoost_GettingStarted 23.AI-and-Analytics/Getting-Started-Samples/Modin_GettingStarted 24.AI-and-Analytics/Getting-Started-Samples/Modin_Vs_Pandas Signed-off-by: Xu, He <[email protected]>

Ankur-singh

Left a few comments mostly around:

Activating conda environment
ipykernel as dev dependency
Why is numpy installed separately?

Furthermore, I don't see pyproject.toml and uv.lock files. The reproducibility aspect of UV is totally missed, this PR only uses UV as a pip replacement. Ideally, we want to use UV as a complete package manager.

Ankur-singh · 2025-03-06T20:17:37Z

AI-and-Analytics/End-to-end-Workloads/JobRecommendationSystem/sample.json

-          "pip install -r requirements.txt",
-          "python -m ipykernel install --user --name=user_tensorflow-gpu",
-          "python JobRecommendationSystem.py"
+          "conda activate tensorflow-gpu",


Any particular reason to activate conda environment? As far as I know, UV will create an isolated environment.

reply at below

Ankur-singh · 2025-03-06T20:22:12Z

AI-and-Analytics/End-to-end-Workloads/JobRecommendationSystem/sample.json

+          "uv add seaborn",
+          "uv add numpy==1.26.4",
+          "uv run python -m spacy download en_core_web_sm",


super nit but can have them in a single uv add command?

Ankur-singh · 2025-03-06T20:25:39Z

...ytics/Features-and-Functionality/IntelPyTorch_GPU_InferenceOptimization_with_AMP/sample.json

+            "uv add --dev ipykernel",
+            "uv run ipython kernel install --user --name=pytorch-gpu",


We don't have to install ipykernel if we not using it. I know its there in current JSON file, but its good opportunity to clean it up.

Ankur-singh · 2025-03-06T20:28:35Z

...Analytics/Features-and-Functionality/IntelPyTorch_TrainingOptimizations_AMX_BF16/sample.json

+              "uv python pin $(which python)",
+              "uv venv --system-site-packages",
+              "uv add -r requirements.txt",
+              "uv add --dev ipykernel notebook",


Can you please install ipykernel in standard dependency? No need for --dev

removed --dev

Ankur-singh · 2025-03-06T20:34:45Z

AI-and-Analytics/Features-and-Functionality/IntelTensorFlow_AMX_BF16_Training/sample.json

+        "export execution_mode=eager",
+        "uv run python Intel_TensorFlow_AMX_BF16_Training.py -m eager",
+        "uv run ipython kernel install --user --name tensorflow",
+        "uv run jupyter nbconvert --ExecutePreprocessor.enabled=True --ExecutePreprocessor.kernel_name=tensorflow --to notebook IntelTensorFlow_AMX_BF16_Training.ipynb"


Don't think we need kernel_name argument any more

Not sure, if the .ipynb file not specify the kernel name and command line also not include it, we might meet issue like jupyter_client.kernelspec.NoSuchKernel: No such kernel named tf_amx_bf16_2, maybe keep the arg for now.

Ankur-singh · 2025-03-06T20:37:04Z

...nd-Analytics/Features-and-Functionality/IntelTensorFlow_TextGeneration_with_LSTM/sample.json

+                "conda activate tensorflow-gpu",
+                "pip install uv notebook",
+                "uv init",
+                "uv python pin $(which python)",
+                "uv venv --system-site-packages",
+                "uv add -r requirements.txt",


Here, notebook is install separately. Both notebook and ipykernel are not used.

yeah, removed

Ankur-singh · 2025-03-06T20:41:14Z

AI-and-Analytics/Getting-Started-Samples/INC-Sample-for-Tensorflow/sample.json

+				"pip install notebook",
+				"uv add --dev ipykernel",
+				"uv run ipython kernel install --user --name tensorflow"
+
        ],
 		"id": "neural-compressor tensorflow",
 		"steps": [


I think you missed uv run in the next line

yeah, thanks !

Ankur-singh · 2025-03-06T20:43:50Z

...-Analytics/Getting-Started-Samples/Intel_Extension_For_TensorFlow_GettingStarted/sample.json

+				"conda activate base",
+				"pip install uv notebook",
+				"uv init",
+				"uv python pin $(which python)",
+				"uv venv --system-site-packages",
+				"uv add -r requirements.txt",
+				"uv add numpy==1.26.4",
+				"uv run ipython kernel install --user --name=tensorflow",
+				"uv run ipython kernel install --user --name=tensorflow-gpu",
+


The original sample.json file tried to test the sample in both CPU and GPU. This doesn't seems to be doing that.

ok, already remove cpu one and keep the gpu one

Ankur-singh · 2025-03-06T20:44:20Z

AI-and-Analytics/Getting-Started-Samples/Modin_GettingStarted/sample.json

  		"id": "Intel_Modin_GS_py",
  		"steps": [
-			"source /intel/oneapi/intelpython/bin/activate",
-			"pip install -r requirements.txt",
-			"pip install jupyter ipykernel",
 			"jupyter nbconvert --ExecutePreprocessor.enabled=True --to notebook Modin_GettingStarted.ipynb"


missed uv run

thanks, added

…tarted as test

Getting-Started-Samples/INC-Quantization-Sample-for-PyTorch Getting-Started-Samples/INC-Sample-for-Tensorflow Getting-Started-Samples/IntelPython_daal4py_GettingStarted Getting-Started-Samples/Intel_Extension_For_SKLearn_GettingStarted Getting-Started-Samples/Intel_Extension_For_TensorFlow_GettingStarted Getting-Started-Samples/Modin_GettingStarted Getting-Started-Samples/Modin_Vs_Pandas

Features-and-Functionality/IntelPyTorch_GPU_InferenceOptimization_with_AMP Features-and-Functionality/IntelPyTorch_TrainingOptimizations_AMX_BF16 Features-and-Functionality/IntelTensorFlow_AMX_BF16_Inference Features-and-Functionality/IntelTensorFlow_AMX_BF16_Training Features-and-Functionality/IntelTensorFlow_Enabling_Auto_Mixed_Precision_for_TransferLearning Features-and-Functionality/IntelTensorFlow_TextGeneration_with_LSTM Features-and-Functionality/IntelTensorFlow_for_LLMs Features-and-Functionality/IntelTransformers_Quantization

bopeng1234 · 2025-03-07T09:10:16Z

Left a few comments mostly around:

Activating conda environment

ipykernel as dev dependency

Why is numpy installed separately?

Furthermore, I don't see pyproject.toml and uv.lock files. The reproducibility aspect of UV is totally missed, this PR only uses UV as a pip replacement. Ideally, we want to use UV as a complete package manager.

The reasons why we use AI Tool's Conda environments alongside uv:

Isolation and Reusability
uv allows us to maintain a local venv environment while also leveraging packages from the Conda environment installed by AI Tool. These two environments remain independent, ensuring that uv run can utilize both local dependencies and Conda-installed packages such as PyTorch and TensorFlow. Additionally, uv sync or uv add does not modify the system environment, achieving both isolation (sample's requirements) and package reuse (AI Tool's ipex/itex).
CI Validation of AI Tool + Sample
The purpose of CI (Continuous Integration) is to verify that AI Tool and its associated samples work correctly. we need to use the environment provided by AI Tool, to validate the samples works well with it.
Avoiding Redundant Installations
Without using the AI Tool environment, every sample would need to install pytorch, tensorflow (which is from AI Tool Conda env) separately. This results in duplicate installations, excessive resource consumption, and significantly increases CI run time.

bopeng1234 · 2025-03-07T09:12:22Z

ipykernel issue is resolved by removing --dev
numpy separately installation also removed.

bopeng1234 · 2025-03-07T09:16:51Z

This PR isn’t just about using uv as a simple replacement for pip. Instead, it takes advantage of AI Tool’s base image while allowing each sample to create its own minimal local venv. This ensures that every sample remains independent while still reusing the system environment, ultimately enabling CI to validate that AI Tool + sample work correctly.

bopeng1234 · 2025-03-07T09:28:28Z

Thanks for the review @Ankur-singh , we refactor the code and added pyproject.toml and uv sync

End-to-end-Workloads/JobRecommendationSystem Features-and Functionality/INC_QuantizationAwareTraining_TextClassification Features-and-Functionality/Intel_Extension_For_SKLearn_Performance_SVC_Adult Features-and-Functionality/IntelPython_daal4py_DistributedKMeans Features-and-Functionality/IntelPython_daal4py_DistributedLinearRegression Features-and-Functionality/IntelPython_GPU_dpnp_Genetic_Algorithm Features-and-Functionality/IntelPython_Numpy_Numba_dpnp_kNN Features-and-Functionality/IntelPython_XGBoost_Performance Signed-off-by: troy818 <[email protected]>

Ankur-singh · 2025-03-07T21:07:34Z

This PR isn’t just about using uv as a simple replacement for pip. Instead, it takes advantage of AI Tool’s base image while allowing each sample to create its own minimal local venv. This ensures that every sample remains independent while still reusing the system environment, ultimately enabling CI to validate that AI Tool + sample work correctly.

This makes sense. The PR looks much better now. Thank you so much for all the work.

There is one major question that needs to be answered. Next oneAPI release for AI tools will not have offline installer. All the packages will either be distributed via apt-get or pip. This change is quite significant for two reasons:

We won't have conda environments to start with.
UV is good for python packages but not OS level packages.

We will have to give some thought. I will be happy to hear your ideas.

cc @jimmytwei

bopeng1234 · 2025-03-10T02:23:25Z

Oh, I see! I wasn’t aware that the Conda environment would be removed in the next AI Tool release. If that’s the case, the changes in this PR will be suitable for the current release but could also serve as a reference for the next version.

For each sample, use uv to install the required packages individually, such as PyTorch, TensorFlow, and other dependencies. Similar to this PR, the automatically generated pyproject.toml should be pushed to the repository. In sample.json, sync the packages from pyproject.toml using uv sync to create an isolated environment, and use uv run to execute each sample independently.

bopeng1234 · 2025-03-10T02:27:40Z

I'm not sure if all the required packages for each sample can already be installed via apt or pip. If they can, we can refactor the uv workflow accordingly and remove the Conda-related components from the current AI Tool.

Ankur-singh · 2025-03-11T15:25:11Z

If that’s the case, the changes in this PR will be suitable for the current release but could also serve as a reference for the next version.

Yes, I think we should be doing it incrementally.

I'm not sure if all the required packages for each sample can already be installed via apt or pip.

I'm not sure about it either, we will have to wait for the next release. @jimmytwei do you have any insights about this?

This was referenced Mar 3, 2025

Hackthon ai tool 1 #2592

Closed

Use uv tool for isolate cicd env, including samples: #2593

Closed

Add uv support for sample ci validation #2594

Closed

use uv tool for isolate cicd env, samples include #2595

Closed

jimmytwei added the Hackathon label Mar 3, 2025

troy818 force-pushed the 2025Hack_AI_Tool branch from b4c85f0 to 3b2414b Compare March 5, 2025 02:54

Ankur-singh reviewed Mar 6, 2025

View reviewed changes

bopeng1234 added 5 commits March 7, 2025 05:36

refactor uv, use Getting-Started-Samples/IntelPython_XGBoost_GettingS…

283801d

…tarted as test

Getting-Started-Samples/IntelPython_XGBoost_GettingStarted ready

8cc42c2

add kernel name for getstart sample

99d6abd

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Use uv tool for isolate cicd env #2597

Use uv tool for isolate cicd env #2597

bopeng1234 commented Mar 3, 2025

Ankur-singh left a comment

Ankur-singh Mar 6, 2025

bopeng1234 Mar 7, 2025 •

edited

Loading

Ankur-singh Mar 6, 2025

bopeng1234 Mar 7, 2025

Ankur-singh Mar 6, 2025

bopeng1234 Mar 7, 2025

Ankur-singh Mar 6, 2025

bopeng1234 Mar 7, 2025 •

edited

Loading

Ankur-singh Mar 6, 2025

bopeng1234 Mar 7, 2025

Ankur-singh Mar 6, 2025

bopeng1234 Mar 7, 2025

Ankur-singh Mar 6, 2025

bopeng1234 Mar 7, 2025

Ankur-singh Mar 6, 2025

bopeng1234 Mar 7, 2025

Ankur-singh Mar 6, 2025

bopeng1234 Mar 7, 2025

bopeng1234 commented Mar 7, 2025

bopeng1234 commented Mar 7, 2025

bopeng1234 commented Mar 7, 2025

bopeng1234 commented Mar 7, 2025

Ankur-singh commented Mar 7, 2025

bopeng1234 commented Mar 10, 2025

bopeng1234 commented Mar 10, 2025

Ankur-singh commented Mar 11, 2025

		"uv add --dev ipykernel",
		"uv run ipython kernel install --user --name=pytorch-gpu",

Use uv tool for isolate cicd env #2597

Are you sure you want to change the base?

Use uv tool for isolate cicd env #2597

Conversation

bopeng1234 commented Mar 3, 2025

Existing Sample Changes

Description

Type of change

Ankur-singh left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

bopeng1234 Mar 7, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

bopeng1234 Mar 7, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

bopeng1234 commented Mar 7, 2025

bopeng1234 commented Mar 7, 2025

bopeng1234 commented Mar 7, 2025

bopeng1234 commented Mar 7, 2025

Ankur-singh commented Mar 7, 2025

bopeng1234 commented Mar 10, 2025

bopeng1234 commented Mar 10, 2025

Ankur-singh commented Mar 11, 2025

bopeng1234 Mar 7, 2025 •

edited

Loading

bopeng1234 Mar 7, 2025 •

edited

Loading