ml-experiment

What problem does it solve?

This Skill addresses the common issue of losing track of ML experiment progress, hypotheses, and learnings, leading to repeated failures and slow iteration.

Core Features & Use Cases

• Structured Experiment Logging: Enforces logging hypotheses before experiments and results after.
• Persistent Journaling: Maintains experiments/journal.md and experiments/lessons.md for cross-session memory.
• Iterative Improvement: Guides users to review past experiments and distill lessons for future work.
• Use Case: When fine-tuning a large language model, this skill ensures that every hyperparameter change, dataset modification, or architectural tweak is logged with its expected outcome, preventing the agent from trying the same failed approaches repeatedly.

Quick Start

Log a new experiment by writing a hypothesis and planned changes in the journal.