Implement CNN, KNN & Naive Bayes

I need three clean, self-contained classification modules that I can drop into an existing Python pipeline.

1. Convolutional Neural Network
• Data: images only
• Goal: image classification
• Framework: TensorFlow (eager execution, TF 2.x)
• Deliverable: a .py file or notebook that builds, trains, and evaluates the model on a sample dataset I will supply, plus saved weights and a short README.

2. k-Nearest Neighbour
• Data: the same image set, pre-processed as needed
• Goal: baseline image classification to benchmark the CNN
• Deliverable: training script, accuracy report, and a way to tweak k from the command line.

3. Naive Bayes
• Data: a small corpus of labelled text documents
• Goal: text classification
• Deliverable: script/notebook that tokenises, vectorises, trains, and outputs precision, recall, and F1.

Acceptance criteria
• Reproducible runs: a requirements.txt or environment.yml.
• Clear separation of the three models; no shared global state.
• Code commented well enough for a mid-level Python user to follow.
• All results must be repeatable on CPU-only hardware.

Let me know if any extra dataset specs are needed so I can provide them before you start.