René's URL Explorer Experiment


Title: NumPy Random Seeds — Global Overwrite Ruins Reproducibility

Open Graph Title: NumPy Random Seeds — Global Overwrite Ruins Reproducibility

Description: ±3% validation accuracy swing on same seed — two libraries call np.random.seed() independently. Fix with explicit Generator seeds for reproducible ML runs.

Open Graph Description: ±3% validation accuracy swing on same seed — two libraries call np.random.seed() independently. Fix with explicit Generator seeds for reproducible ML runs.

Opengraph URL: https://thecodeforge.io/python/numpy-random-module/

direct link

Domain: thecodeforge.io


Hey, it has json ld scripts:
{
  "@context": "https://schema.org",
  "@type": "TechArticle",
  "mainEntityOfPage": {
    "@type": "WebPage",
    "@id": "https://thecodeforge.io/python/numpy-random-module/"
  },
  "headline": "NumPy Random Seeds — Global Overwrite Ruins Reproducibility",
  "description": "±3% validation accuracy swing on same seed — two libraries call np.random.seed() independently. Fix with explicit Generator seeds for reproducible ML runs.",
  "keywords": "NumPy Random Seeds — Global Overwrite Ruins Reproducibility, Python, Intermediate, Python Libraries, ±3% validation accuracy swing on same seed — two libraries call np.random.seed() independently. Fix with explicit Generator seeds for reproducible ML runs.",
  "proficiencyLevel": "Intermediate",
  "author": {
    "@type": "Person",
    "@id": "https://thecodeforge.io/about/#naren",
    "name": "Naren",
    "url": "https://thecodeforge.io/about/"
  },
  "publisher": {
    "@type": "Organization",
    "@id": "https://thecodeforge.io/#org",
    "name": "TheCodeForge",
    "url": "https://thecodeforge.io",
    "logo": {"@type": "ImageObject", "url": "https://thecodeforge.io/assets/og-image.png"}
  },
  "datePublished": "2026-03-16T00:00:00+00:00",
  "dateModified": "2026-08-30T00:00:00+00:00",
  "url": "https://thecodeforge.io/python/numpy-random-module/",
  "inLanguage": "en",
  "educationalLevel": "Intermediate",
  "image": "https://thecodeforge.io/assets/og-image.png",
  "isPartOf": {"@id": "https://thecodeforge.io/#website"}
}
{
  "@context": "https://schema.org",
  "@type": "BreadcrumbList",
  "itemListElement": [
    {"@type": "ListItem", "position": 1, "name": "Home", "item": "https://thecodeforge.io/"},
    {"@type": "ListItem", "position": 2, "name": "Python", "item": "https://thecodeforge.io/python/"},
    {"@type": "ListItem", "position": 3, "name": "NumPy Random Seeds — Global Overwrite Ruins Reproducibility", "item": "https://thecodeforge.io/python/numpy-random-module/"}
  ]
}
{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "What is the difference between np.random.seed() and np.random.default_rng()?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "np.random.seed() sets a global seed that affects all legacy numpy.random functions. np.random.default_rng() creates an independent Generator object. The Generator approach is better because it avoids shared global state — multiple generators with different seeds can run independently in the same process."
      }
    },
    {
      "@type": "Question",
      "name": "Why does my random data change every time I run my script?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "You are not seeding the generator. Add seed=42 (or any integer) to np.random.default_rng(). The exact number does not matter — what matters is that you use the same number consistently."
      }
    },
    {
      "@type": "Question",
      "name": "Can I mix legacy and Generator API in the same script?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Yes, but avoid it. The legacy API uses a global RandomState that can interact unpredictably with Generator objects. Migrate all code to Generator for consistency."
      }
    },
    {
      "@type": "Question",
      "name": "How do I generate the same random numbers in Python 2 and Python 3?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Use the bit-level Generator API and a fixed seed. In Python 2, use np.random.RandomState(seed).randn() (legacy) but note that Python 2 is unsupported. Better: use the same NumPy version and the Generator API which is stable within major versions."
      }
    }
  ]
}

og:typewebsite
og:imagehttps://thecodeforge.io/assets/og-image.png
og:image:width1200
og:image:height630

Links:

TheCodeForge https://thecodeforge.io/
Pythonhttps://thecodeforge.io/python/
Buy me a coffee https://ko-fi.com/thecodeforge
1NumPy Basicshttps://thecodeforge.io/python/numpy-basics/
2NumPy Arrays and Operationshttps://thecodeforge.io/python/numpy-arrays-operations/
3Pandas Basicshttps://thecodeforge.io/python/pandas-basics/
4Pandas DataFrameshttps://thecodeforge.io/python/pandas-dataframes/
5Matplotlib Basicshttps://thecodeforge.io/python/matplotlib-basics/
6Seaborn for Data Visualisationhttps://thecodeforge.io/python/seaborn-data-visualisation/
7Requests Library in Pythonhttps://thecodeforge.io/python/requests-library-python/
8Flask Web Framework Basicshttps://thecodeforge.io/python/flask-web-framework-basics/
9Django Web Framework Basicshttps://thecodeforge.io/python/django-web-framework-basics/
10SQLAlchemy Basicshttps://thecodeforge.io/python/sqlalchemy-basics/
11itertools Module in Pythonhttps://thecodeforge.io/python/itertools-python/
12collections Module in Pythonhttps://thecodeforge.io/python/collections-module-python/
13datetime Module in Pythonhttps://thecodeforge.io/python/datetime-module-python/
14regex Module in Pythonhttps://thecodeforge.io/python/regex-module-python/
15threading and multiprocessing in Pythonhttps://thecodeforge.io/python/threading-multiprocessing-python/
16FastAPI Basicshttps://thecodeforge.io/python/fastapi-basics/
17Celery for Task Queues in Pythonhttps://thecodeforge.io/python/celery-task-queues-python/
18Pytest Fixtureshttps://thecodeforge.io/python/pytest-fixtures/
19Beautiful Soup Web Scrapinghttps://thecodeforge.io/python/beautiful-soup-web-scraping/
20Selenium with Pythonhttps://thecodeforge.io/python/selenium-python/
21Pydantic for Data Validationhttps://thecodeforge.io/python/pydantic-data-validation/
22Streamlit for Data Appshttps://thecodeforge.io/python/streamlit-data-apps/
23Playwright Python — Browser Automation and Testinghttps://thecodeforge.io/python/playwright-python/
24Advanced Network Interception and Mocking in Playwright Pythonhttps://thecodeforge.io/python/playwright-python-network-mocking/
25NumPy Broadcasting — How It Actually Workshttps://thecodeforge.io/python/numpy-broadcasting/
26NumPy Indexing and Slicing — Beyond the Basicshttps://thecodeforge.io/python/numpy-indexing-slicing/
27NumPy Shape Manipulation — reshape, flatten, ravel, transposehttps://thecodeforge.io/python/numpy-shape-manipulation/
28NumPy Mathematical Functions — ufuncs, aggregations and statisticshttps://thecodeforge.io/python/numpy-mathematical-functions/
29NumPy Random Module — Generating and Controlling Random Datahttps://thecodeforge.io/python/numpy-random-module/
30NumPy Linear Algebra — dot, matmul, linalg explainedhttps://thecodeforge.io/python/numpy-linear-algebra/
31NumPy Boolean Indexing and Fancy Indexinghttps://thecodeforge.io/python/numpy-boolean-fancy-indexing/
32NumPy Performance Tips — Vectorisation vs Loopshttps://thecodeforge.io/python/numpy-performance-vectorisation/
33NumPy with Pandas — How They Work Togetherhttps://thecodeforge.io/python/numpy-with-pandas/
34NumPy dtype and Memory Layout — float32, int64 and C vs F orderhttps://thecodeforge.io/python/numpy-dtype-memory/
35NumPy loadtxt and savetxt — Reading and Writing Array Datahttps://thecodeforge.io/python/numpy-loadtxt-savetxt/
36NumPy where, select and piecewise — Conditional Array Operationshttps://thecodeforge.io/python/numpy-where-select-piecewise/
37FastAPI Path Parameters and Query Parametershttps://thecodeforge.io/python/fastapi-path-query-parameters/
38FastAPI Request Body and Pydantic Modelshttps://thecodeforge.io/python/fastapi-request-body-pydantic/
39FastAPI Response Models and Status Codeshttps://thecodeforge.io/python/fastapi-response-models-status-codes/
40FastAPI Dependency Injection — How and Why to Use Ithttps://thecodeforge.io/python/fastapi-dependency-injection/
41FastAPI Authentication — JWT and OAuth2 with Password Flowhttps://thecodeforge.io/python/fastapi-authentication-jwt/
42FastAPI Background Tasks and Async Endpointshttps://thecodeforge.io/python/fastapi-background-tasks-async/
43FastAPI Database Integration with SQLAlchemyhttps://thecodeforge.io/python/fastapi-sqlalchemy-database/
44FastAPI File Uploads and Form Datahttps://thecodeforge.io/python/fastapi-file-uploads-forms/
45FastAPI Middleware — Logging, CORS and Custom Middlewarehttps://thecodeforge.io/python/fastapi-middleware-cors/
46FastAPI Testing with pytest and TestClienthttps://thecodeforge.io/python/fastapi-testing-pytest/
47FastAPI WebSockets — Real-time Communicationhttps://thecodeforge.io/python/fastapi-websockets/
48FastAPI Deployment — Docker, Uvicorn and Gunicornhttps://thecodeforge.io/python/fastapi-deployment-docker/
49FastAPI vs Flask vs Django — When to Use Whichhttps://thecodeforge.io/python/fastapi-vs-flask-vs-django/
50FastAPI Error Handling and Custom Exception Handlershttps://thecodeforge.io/python/fastapi-error-handling/
51PySpark Tutorial: Big Data Processing with Pythonhttps://thecodeforge.io/python/pyspark-tutorial/
52FastAPI CLI — dev, run and deployhttps://thecodeforge.io/python/fastapi-cli-dev-run-deploy/
53FastAPI OpenAPI Customization — Tags, Examples and Schemahttps://thecodeforge.io/python/fastapi-openapi-customization/
54FastAPI Lifespan Events — Startup, Shutdown and Context Managershttps://thecodeforge.io/python/fastapi-lifespan-events/
55FastAPI Streaming Responses and File Responseshttps://thecodeforge.io/python/fastapi-streaming-responses/
56FastAPI Async SQLAlchemy — Alembic Migrations and Production Patternshttps://thecodeforge.io/python/fastapi-async-sqlalchemy/
57FastAPI Cloud — Official Deployment Platformhttps://thecodeforge.io/python/fastapi-cloud-deployment/
58Django OperationalError No Such Table Fixhttps://thecodeforge.io/python/django-operationalerror-no-such-table/
59Django ImproperlyConfigured Fixhttps://thecodeforge.io/python/django-improperlyconfigured/
← Back to Python track https://thecodeforge.io/python/
Homehttps://thecodeforge.io/
Pythonhttps://thecodeforge.io/python/
Follow https://www.linkedin.com/in/thecodeforge-io-a0676b3b5/
 ● Production Incidenthttps://thecodeforge.io/python/numpy-random-module/#v2-production-incident
 🔎 Debug Guidehttps://thecodeforge.io/python/numpy-random-module/#v2-debug-guide
 ⚙ Triage Commandshttps://thecodeforge.io/python/numpy-random-module/#v2-triage-commands
NumPyhttps://thecodeforge.io/python/numpy-basics/
functionshttps://thecodeforge.io/python/functions-python/
designhttps://thecodeforge.io/java/design-patterns-java/
machinehttps://thecodeforge.io/ml-ai/introduction-machine-learning/
generatorshttps://thecodeforge.io/python/generators-python/
arrayshttps://thecodeforge.io/c-cpp/arrays-c/
→NumPy Broadcasting — Silent OOM That Killed 5M Profiles5M profiles OOM-killed a container because broadcasting silently inflated a 2D operation into 3D. Avoid GFG's toy examplhttps://thecodeforge.io/python/numpy-arrays-operations/
→NumPy Loop vs Vectorisation: 45-Minute Training BottleneckPython for-loop over 50M rows caused 48-minute single-core bottleneck. Vectorised cumsum cut runtime to 3 minutes — applhttps://thecodeforge.io/python/numpy-performance-vectorisation/
→NumPy Indexing — Unintended View Dropped Accuracy 92%→37%A NumPy slice created a view that mutated training array, dropping accuracy 92%→37%. Use np.shares_memory() to prevent shttps://thecodeforge.io/python/numpy-indexing-slicing/
→NumPy Mathematical Functions — The NaN Trap in np.meannp.mean returns NaN if any element is NaN, propagating through pipelines and breaking predictions. Swap to np.nanmean anhttps://thecodeforge.io/python/numpy-mathematical-functions/
→Universal Hashing — 2011 Attack That Forced Random SeedsCPU spikes on POST requests? The 2011 hash DoS attack exploited deterministic hash tables. Our production debug guide hehttps://thecodeforge.io/dsa/universal-hashing/
Model Monitoring and Drift DetectionAdvancedhttps://thecodeforge.io/ml-ai/model-monitoring-drift-detection/
Follow https://www.linkedin.com/in/thecodeforge-io-a0676b3b5/
←PreviousNumPy Mathematical Functions — ufuncs, aggregations and statisticshttps://thecodeforge.io/python/numpy-mathematical-functions/
NextNumPy Linear Algebra — dot, matmul, linalg explained→https://thecodeforge.io/python/numpy-linear-algebra/
LinkedIn https://www.linkedin.com/in/thecodeforge-io-a0676b3b5/
Facebook https://www.facebook.com/profile.php?id=61583020033815
YouTube https://www.youtube.com/channel/UCMar7CF2EmCF1I2tw1tNVuQ
Javahttps://thecodeforge.io/java/
Pythonhttps://thecodeforge.io/python/
C / C++https://thecodeforge.io/c-cpp/
JavaScripthttps://thecodeforge.io/javascript/
DSAhttps://thecodeforge.io/dsa/
Databasehttps://thecodeforge.io/database/
System Designhttps://thecodeforge.io/system-design/
DevOpshttps://thecodeforge.io/devops/
C# / .NEThttps://thecodeforge.io/csharp/
PHPhttps://thecodeforge.io/php/
ML / AIhttps://thecodeforge.io/ml-ai/
Interviewhttps://thecodeforge.io/interview/
CS Fundamentalshttps://thecodeforge.io/cs-fundamentals/
Rusthttps://thecodeforge.io/rust/
Toolshttps://thecodeforge.io/tools/
Cheatsheetshttps://thecodeforge.io/cheatsheets/
Mobilehttps://thecodeforge.io/mobile/
Data Analyticshttps://thecodeforge.io/data-analytics/
Securityhttps://thecodeforge.io/security/
Frontendhttps://thecodeforge.io/frontend/
Data Engineeringhttps://thecodeforge.io/data-engineering/
Testinghttps://thecodeforge.io/testing/
Cloudhttps://thecodeforge.io/cloud/
Observabilityhttps://thecodeforge.io/observability/
Web Platformhttps://thecodeforge.io/web-platform/
🧯 Errors Explainedhttps://thecodeforge.io/errors/
⚒ Blueprintshttps://thecodeforge.io/blueprints/
⌨ Practicehttps://thecodeforge.io/practice/
Interview Prephttps://thecodeforge.io/interview/
DSA Problemshttps://thecodeforge.io/dsa/
Developer Toolshttps://thecodeforge.io/tools/
Cheat Sheetshttps://thecodeforge.io/cheatsheets/
Abouthttps://thecodeforge.io/about/
Sitemaphttps://thecodeforge.io/sitemap.xml
Privacyhttps://thecodeforge.io/privacy.html

Viewport: width=device-width, initial-scale=1.0

Robots: max-image-preview:large


URLs of crawlers that visited me.