AI-900 Practice Question: Describe features of Natural Language Processing workloads on Azure

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Moving a trained model from one Azure region to another for deployment

  Why it's wrong here

  Deploying a model to a different Azure region is a pure infrastructure concern—you are moving a serialized artifact (e.g., a pickle file or MLflow model) and setting up a new endpoint, often for latency, data residency, or disaster recovery. This operation has nothing to do with how the model's weights were learned. Transfer learning instead refers to reusing the knowledge encoded in a model's parameters to accelerate learning on a new task, not relocating where inference runs.

• ✓

  Using knowledge from a large pre-trained model as the starting point for learning a new, related task

  Why this is correct

  Transfer learning is a training methodology where an existing model that was pre-trained on a large, general corpus (e.g., a transformer trained on massive text) is fine-tuned on a smaller, task-specific dataset. The lower layers retain broad feature detectors—like edges in images or syntax in language—so the network only needs to adapt the higher layers to the new objective. This dramatically reduces the amount of labeled data and compute required compared to training from scratch, because the model already encodes general world knowledge.

• ✗

  Transferring labeled training data between different ML projects

  Why it's wrong here

  Sharing labeled data between projects is an act of data management—you are moving or copying examples, which may implicate data governance, privacy, and licensing rules, but the model parameters have not been touched. Transfer learning does not transfer exemplars; it transfers the knowledge abstracted from data into weights, such as word embeddings or convolutional filters. Even if the same dataset is reused, the model still learns from scratch unless you initialize it with a pre-trained model's parameters.

• ✗

  Automatically translating ML models from Python to other programming languages

  Why it's wrong here

  Automatically converting a model from Python to another programming language (e.g., via ONNX, PMML, or TensorFlow Serving) is a model interoperability and deployment concern—you are translating a computational graph or a serialized artifact, not transferring learned knowledge to a new task. Even if the translation succeeds, the model's parameters remain exactly the same; the model has not been trained or adapted. Transfer learning, by contrast, changes the model's weights through learning a new task, often starting from a pre-trained checkpoint.

Go deeper

Related to this question