Experiment Tracking Setup AI Prompt

Set up comprehensive experiment tracking for this ML training pipeline.

Use {{tracking_tool}} (MLflow / Weights & Biases / Neptune).

1. Run initialization:
   - Create a run with a descriptive name including: model architecture, dataset version, timestamp
   - Tag run with: git commit hash, environment (dev/staging/prod), dataset version

2. Hyperparameter logging:
   - Log all hyperparameters at run start: learning rate, batch size, epochs, optimizer, scheduler, architecture config
   - Log data config: train/val split, augmentations, preprocessing steps

3. Metric logging per epoch:
   - Training: loss, primary metric, learning rate, gradient norm
   - Validation: loss, primary metric, all secondary metrics
   - System: GPU memory used, step time, throughput (samples/sec)

4. Artifact logging:
   - Best model checkpoint
   - Final model checkpoint
   - Confusion matrix or prediction plots at end of training
   - Feature importance if applicable

5. Run comparison:
   - Show how to use the tracking UI to compare runs by val metric
   - Show how to retrieve the best run programmatically

6. Reproducibility:
   - Log environment: requirements.txt or conda env YAML
   - Log random seeds

Return: tracking setup code integrated into the training loop, and a run naming convention guide.