When does this skill trigger?

It triggers automatically when you modify training notebooks, GPU environment configurations, reward functions, or any training-related source code.

Why does it check weight sums?

It ensures that reward function components (direction, pnl, drawdown, etc.) sum to exactly 1.0, preventing mathematical imbalances during the training phase.

How does it handle Google Colab paths?

It explicitly differentiates between local Windows paths and unzipped Colab paths (/content/...) to prevent file-not-found errors during cloud execution.

Can it verify notebook cells?

Yes, it identifies cells by their content rather than their index number, which prevents errors when cells are added or shifted during editing.

What tests does this skill run?

It prompts for the execution of pytest suites for PPO trainers, GPU training modules, and data fetchers to ensure no regressions were introduced.

Training Code Verification

Name: Training Code Verification
Author: smith6jt-cop

bysmith6jt-cop

Data Science & ML

Validates machine learning training scripts and configurations to ensure path accuracy, weight consistency, and environment compatibility.

About

This skill provides a rigorous verification framework for developers modifying reinforcement learning or algorithmic trading training pipelines. It prevents common failures such as path mismatches in Google Colab, inconsistent reward function weights, and outdated API key references by enforcing a 'verify, don't assume' workflow. It is particularly useful when switching between local development and cloud environments like Colab, ensuring that environment-specific configurations and notebook cell sequences remain intact and functional before code is committed or executed.

Key Features

0 GitHub stars
Version string consistency checking across notebook and script files
Configuration value validation for GPU environment settings and weight sums
Cell-by-content identification to prevent notebook execution errors
Automated path and filename verification for local and Colab environments
Pre-commit test execution for PPO trainers and data fetchers

Use Cases

Modifying reward functions and penalty weights in reinforcement learning agents
Porting training workflows from local development machines to Google Colab
Updating Alpaca API keys and environment-specific paper trading configurations

About

Key Features

0 GitHub stars
Version string consistency checking across notebook and script files
Configuration value validation for GPU environment settings and weight sums
Cell-by-content identification to prevent notebook execution errors
Automated path and filename verification for local and Colab environments
Pre-commit test execution for PPO trainers and data fetchers

Use Cases

Modifying reward functions and penalty weights in reinforcement learning agents
Porting training workflows from local development machines to Google Colab
Updating Alpaca API keys and environment-specific paper trading configurations