ml-model-training

Train ML models with scikit-learn, PyTorch, TensorFlow. Use for classification/regression, neural networks, hyperparameter tuning, or encountering overfitting, underfitting, convergence issues.

Install with Tessl CLI

npx tessl i github:secondsky/claude-skills --skill ml-model-training

What are skills?

Review — 93%

Does it follow best practices?

Validation — 10 / 11 Passed

Validation for skill structure

SKILL.md

Review

Evals

Evaluation results

100%

Customer Churn Preprocessing Pipeline

Data splitting and preprocessing pipeline

Criteria

Without context

With context

70/15/15 split

100%

Scaler fit on train only

100%

No all-data fitting

50%

100%

Categorical encoding

100%

Feature scaling applied

100%

Random seed set

100%

Validation set present

100%

Correct pandas loading

100%

Feature/target separation

100%

Without context: $0.3114 · 1m 4s · 14 turns · 15 in / 3,787 out tokens

With context: $0.8644 · 2m 7s · 26 turns · 27 in / 6,936 out tokens

100%

Fraud Detection Model for Payment Transactions

Class imbalance handling and hyperparameter tuning

Criteria

Without context

With context

Class imbalance addressed

100%

SMOTE or class weights

100%

GridSearch/CV on training only

100%

Test set reserved for final eval

100%

MLflow tracking

100%

MLflow run context

100%

Classification report

100%

AUC-ROC reported

100%

Cross-validation used

100%

Hyperparameters documented

100%

Without context: $0.9018 · 3m 59s · 27 turns · 26 in / 12,008 out tokens

With context: $0.7613 · 3m 23s · 29 turns · 77 in / 8,851 out tokens

100%

15%

Patient Readmission Risk Prediction with Neural Networks

PyTorch neural network training loop

Criteria

Without context

With context

BatchNorm in hidden layers

100%

Dropout in hidden layers

100%

Adam optimizer used

100%

ReduceLROnPlateau scheduler

100%

Early stopping implemented

100%

Best model checkpoint saved

70%

100%

DataLoader used

100%

Batch size 32

100%

Random seeds set

100%

Separate val loop

100%

Classification report

100%

Validation set included

100%

Without context: $0.4057 · 4m 54s · 15 turns · 387 in / 6,188 out tokens

With context: $0.7039 · 7m 44s · 23 turns · 272 in / 7,702 out tokens

Evaluated: 3 days ago
Agent: Claude Code

Table of Contents

Customer Churn Preprocessing Pipeline Fraud Detection Model for Payment Transactions Patient Readmission Risk Prediction with Neural Networks

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.