Use this for every training experiment.

Experiment

• Experiment ID:
• Date:
• Module:
• Question:
• Dataset/source:
• Intended use:
• Not intended for:

Data

• Number of rows/items:
• Input shape:
• Target or label:
• Split method:
• Known limitations:
• Leakage checks:

Model

• Baseline:
• PyTorch model:
• Loss:
• Optimizer:
• Learning rate:
• Epochs:
• Batch size:
• Random seed:
• CPU/GPU:

Results

Diagnostics

• Training curve pattern:
• Overfitting signs:
• Underfitting signs:
• Instability signs:
• Error examples inspected:

Decision

• Accept / revise / reject / run more checks:
• Reason:
• Next change:
• Limitations:

Mark complete