concept#Artificial Intelligence#Machine Learning#Model Architecture#Model Platform

Pre-Trained Model

Machine learning models pre-trained on large datasets and reused or fine-tuned for downstream tasks to accelerate development and reduce resource needs.

Pre-trained models are machine learning models trained on large generic datasets and reused or fine-tuned for specific downstream tasks.

Maturity

Established

Cognitive loadMedium

360° overview

Pre-Trained Model

Six perspectives place the building block in context. The numbers show where each perspective continues in the reading path.

The building block at a glance

Pre-Trained Model

360°

Classification

2

Complexity: Medium
Impact area: Technical

Technical context

2

Integrations

PyTorch or TensorFlow models in ML pipelines

+2

Principles & goals

1

Assess domain alignment before reuse.

Value stream stage: Build

Use cases & scenarios

4

Use cases: 3
Scenarios: 3

Compromises

3

Risks: 3
Best practices: 3

I/O & resources

4

Inputs: 3
Outputs: 3