From 1c7415e5ba987890156b2312f5859f0868e34be8 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?M=C3=A9d=C3=A9ric=20Hurier=20=28Fmind=29?= Date: Mon, 10 Aug 2026 20:15:55 +0200 Subject: [PATCH] fix(docs): repoint dead neptune.ai links and drop the wound-down product OpenAI acquired Neptune in December 2025 and every neptune.ai/blog URL now 308-redirects to the acquisition announcement, so four referenced articles were unreachable. Each one is repointed at a pinned Wayback capture of the original article, verified through the CDX API as a real text/html 200 of 43-62 KB rather than a redirect stub. Neptune also wound down its external service, so the course was recommending a product readers can no longer sign up for. Those three recommendations now name Comet and ClearML alongside Weights & Biases, and chapter 5.5 says plainly what happened: a tracker disappearing with its documentation is exactly the vendor risk that chapter should teach. Also unblocks docs-only commits. dprint.jsonc excludes docs/**, so the pre-commit hook handed dprint a staged file list it had nothing to do with and dprint exited 14. Every chapter-only contribution failed its own hook. --- docs/0. Overview/0.3. Platforms.md | 2 +- docs/1. Initializing/1.0. System.md | 2 +- docs/5. Refining/5.5. AI-ML Experiments.md | 6 ++++-- docs/5. Refining/5.6. Model Registries.md | 4 ++-- docs/7. Observability/7.0. Reproducibility.md | 2 +- mise.toml | 5 ++++- 6 files changed, 13 insertions(+), 8 deletions(-) diff --git a/docs/0. Overview/0.3. Platforms.md b/docs/0. Overview/0.3. Platforms.md index 2b763d0..1a0ae7a 100644 --- a/docs/0. Overview/0.3. Platforms.md +++ b/docs/0. Overview/0.3. Platforms.md @@ -18,7 +18,7 @@ Core capabilities of an MLOps platform typically include: - **Data and Artifact Storage**: Secure and scalable systems like [Amazon S3](https://aws.amazon.com/s3/) or [Google Cloud Storage](https://cloud.google.com/storage) for managing datasets, models, and other versioned artifacts. - **Compute Resources**: On-demand access to computational power, such as [Kubernetes](https://kubernetes.io/) clusters or managed services like [Databricks](https://databricks.com), for model training and inference. - **Workflow Orchestration**: Tools like [Apache Airflow](https://airflow.apache.org/), [Metaflow](https://metaflow.org/), or [Prefect](https://www.prefect.io/) that automate and manage the complex workflows and data pipelines involved in ML projects. -- **Model Registries and Experiment Tracking**: Centralized platforms like [MLflow](https://mlflow.org/), [Neptune.ai](https://neptune.ai/), or [Weights & Biases](https://wandb.ai/site) for versioning models, tracking experiment parameters, and comparing results. +- **Model Registries and Experiment Tracking**: Centralized platforms like [MLflow](https://mlflow.org/), [Weights & Biases](https://wandb.ai/site), or [Comet](https://www.comet.com/site/) for versioning models, tracking experiment parameters, and comparing results. The choice between open-source tools and managed enterprise solutions depends on trade-offs between flexibility, cost, and operational overhead. Smaller teams might combine tools like [MLflow](https://mlflow.org/) and [Airflow](https://airflow.apache.org/) for a custom, low-cost stack, while large organizations often prefer comprehensive platforms like [Databricks](https://databricks.com) or [AWS SageMaker](https://aws.amazon.com/sagemaker/) for their scalability and support. diff --git a/docs/1. Initializing/1.0. System.md b/docs/1. Initializing/1.0. System.md index 11587a0..9384a32 100644 --- a/docs/1. Initializing/1.0. System.md +++ b/docs/1. Initializing/1.0. System.md @@ -49,4 +49,4 @@ When using cloud services, be mindful of resource management and usage quotas, p - [GitHub Codespaces](https://github.com/features/codespaces) - [Google Cloud Workstations](https://cloud.google.com/workstations) -- [MLOps Landscape in 2024: Top Tools and Platforms](https://neptune.ai/blog/mlops-tools-platforms-landscape) +- [MLOps Landscape in 2024: Top Tools and Platforms](https://web.archive.org/web/20251207223323/https://neptune.ai/blog/mlops-tools-platforms-landscape) diff --git a/docs/5. Refining/5.5. AI-ML Experiments.md b/docs/5. Refining/5.5. AI-ML Experiments.md index 50558d6..a8dbcf2 100644 --- a/docs/5. Refining/5.5. AI-ML Experiments.md +++ b/docs/5. Refining/5.5. AI-ML Experiments.md @@ -8,7 +8,7 @@ description: Master AI/ML experiment tracking with MLflow. Learn how to effectiv ## What is an AI/ML experiment? -[An AI/ML experiment](https://neptune.ai/blog/ml-experiment-tracking) is a systematic and iterative process for building robust machine learning models. It involves testing different algorithms, tuning hyperparameters, and using various datasets to discover the optimal configuration for a specific predictive task. Each experiment is a structured trial designed to measure the impact of changes on model performance, such as accuracy, efficiency, and reliability. +[An AI/ML experiment](https://web.archive.org/web/20251127141526/https://neptune.ai/blog/ml-experiment-tracking) is a systematic and iterative process for building robust machine learning models. It involves testing different algorithms, tuning hyperparameters, and using various datasets to discover the optimal configuration for a specific predictive task. Each experiment is a structured trial designed to measure the impact of changes on model performance, such as accuracy, efficiency, and reliability. ## Why is experiment tracking essential in AI/ML? @@ -22,7 +22,9 @@ In MLOps, the complexity and often non-deterministic nature of model development ## Which experiment tracking solution should you use? -Numerous solutions are available for tracking AI/ML experiments. Major cloud platforms like [Google Cloud (Vertex AI)](https://cloud.google.com/vertex-ai/docs/experiments/intro-vertex-ai-experiments), [Azure (Azure ML)](https://learn.microsoft.com/en-us/python/api/azureml-core/azureml.core.experiment.experiment?view=azure-ml-py), and [AWS (SageMaker)](https://aws.amazon.com/fr/sagemaker/experiments/) offer powerful, integrated MLOps capabilities. Specialized commercial tools like [Weights & Biases](https://wandb.ai/) and [Neptune AI](https://neptune.ai/) also provide excellent features. +Numerous solutions are available for tracking AI/ML experiments. Major cloud platforms like [Google Cloud (Vertex AI)](https://cloud.google.com/vertex-ai/docs/experiments/intro-vertex-ai-experiments), [Azure (Azure ML)](https://learn.microsoft.com/en-us/python/api/azureml-core/azureml.core.experiment.experiment?view=azure-ml-py), and [AWS (SageMaker)](https://aws.amazon.com/fr/sagemaker/experiments/) offer powerful, integrated MLOps capabilities. Specialized commercial tools like [Weights & Biases](https://wandb.ai/), [Comet](https://www.comet.com/site/), and [ClearML](https://clear.ml/) also provide excellent features. + +> **A note on vendor risk**: this chapter used to recommend Neptune.ai here. OpenAI acquired it in December 2025 and it wound down its external service, taking its documentation with it. That is worth internalising before you build a workflow on a hosted tracker: an experiment store is where your history lives, so prefer one whose format you can export, and treat a vendor's continued existence as an assumption rather than a given. It is one of the reasons this course teaches MLflow, which you can run yourself. For those starting out or preferring an open-source, framework-agnostic solution, [MLflow](https://mlflow.org/) is an outstanding choice. It is versatile, robust, and integrates with a wide array of ML libraries. diff --git a/docs/5. Refining/5.6. Model Registries.md b/docs/5. Refining/5.6. Model Registries.md index f4cb843..ea7e679 100644 --- a/docs/5. Refining/5.6. Model Registries.md +++ b/docs/5. Refining/5.6. Model Registries.md @@ -8,7 +8,7 @@ description: Explore the use of model registries with MLflow for managing model ## What is a model registry? -[A model registry](https://neptune.ai/blog/ml-model-registry) is a centralized repository designed to manage the lifecycle of machine learning models. It acts as a version control system for models, tracking their journey from training and experimentation to staging and production deployment. This makes it an indispensable tool for collaborative and scalable MLOps. +[A model registry](https://web.archive.org/web/20251005125802/https://neptune.ai/blog/ml-model-registry) is a centralized repository designed to manage the lifecycle of machine learning models. It acts as a version control system for models, tracking their journey from training and experimentation to staging and production deployment. This makes it an indispensable tool for collaborative and scalable MLOps. ## Why is a model registry essential? @@ -24,7 +24,7 @@ A model registry provides critical governance and operational capabilities: The right choice depends on your existing ecosystem and requirements. - **Cloud-Based Platforms**: Major cloud providers offer tightly integrated solutions, such as [Google Vertex AI](https://cloud.google.com/vertex-ai/docs/model-registry/introduction), [AWS SageMaker](https://docs.aws.amazon.com/sagemaker/latest/dg/model-registry.html), and [Azure ML](https://learn.microsoft.com/en-us/azure/machine-learning/how-to-manage-models). -- **Third-Party Solutions**: Platforms like [Weights & Biases](https://wandb.ai/site) and [Neptune AI](https://neptune.ai/) provide comprehensive experiment tracking and model management features. +- **Third-Party Solutions**: Platforms like [Weights & Biases](https://wandb.ai/site), [Comet](https://www.comet.com/site/), and [ClearML](https://clear.ml/) provide comprehensive experiment tracking and model management features. - **Open-Source**: [MLflow Model Registry](https://mlflow.org/docs/latest/model-registry.html) is a popular, framework-agnostic option that you can host yourself. To begin with MLflow, install it in your project: diff --git a/docs/7. Observability/7.0. Reproducibility.md b/docs/7. Observability/7.0. Reproducibility.md index 019c8fe..788742c 100644 --- a/docs/7. Observability/7.0. Reproducibility.md +++ b/docs/7. Observability/7.0. Reproducibility.md @@ -8,7 +8,7 @@ description: Explore the crucial concept of reproducibility in MLOps and learn h ## What is reproducibility in MLOps? -[Reproducibility in MLOps](https://neptune.ai/blog/how-to-solve-reproducibility-in-ml) is the ability to re-create the exact same results of a machine learning experiment or model, given the same code, data, and environment. This is a fundamental requirement for validating findings, debugging models, and ensuring consistent behavior over time. Achieving reproducibility builds trust and transparency, enabling independent verification and accelerating development by providing a stable foundation. +[Reproducibility in MLOps](https://web.archive.org/web/20251006163636/https://neptune.ai/blog/how-to-solve-reproducibility-in-ml) is the ability to re-create the exact same results of a machine learning experiment or model, given the same code, data, and environment. This is a fundamental requirement for validating findings, debugging models, and ensuring consistent behavior over time. Achieving reproducibility builds trust and transparency, enabling independent verification and accelerating development by providing a stable foundation. ## What is the difference between reproducibility and replicability? diff --git a/mise.toml b/mise.toml index 03cfd3b..8ba47ff 100644 --- a/mise.toml +++ b/mise.toml @@ -54,7 +54,10 @@ depends = ["format:dprint"] [tasks."format:dprint"] description = "Format JSON, Markdown, TOML, YAML (dprint)" -run = "dprint fmt" +# `--allow-no-files` because the pre-commit hook passes staged paths and dprint.jsonc +# excludes `docs/**`: without it, dprint exits 14 on a commit that touches only chapter +# files, which is every content contribution this repository asks for. +run = "dprint fmt --allow-no-files" # CHECK