cs-ml

Purpose

This skill handles machine learning tasks, including supervised/unsupervised learning, reinforcement learning, CNN/RNN/Transformer models, training pipelines, evaluation metrics, MLOps workflows, and LLM fine-tuning. It integrates with OpenClaw to automate code generation and execution for ML projects.

When to Use

Use this skill when building ML models from scratch, fine-tuning pre-trained models like BERT, deploying models via MLOps, or evaluating performance. Apply it for tasks involving large datasets, neural networks, or production pipelines, such as image recognition with CNNs or text generation with Transformers.

Key Capabilities

• Train supervised models using algorithms like linear regression or decision trees via scikit-learn integration.
• Implement unsupervised learning with K-means clustering or PCA for dimensionality reduction.
• Build and train deep learning models: CNNs for images (e.g., using Keras), RNNs for sequences, or Transformers for NLP tasks.
• Handle RL environments with libraries like Stable Baselines, including Q-learning loops.
• Evaluate models with metrics like accuracy, F1-score, or ROC curves, and generate confusion matrices.
• Manage MLOps: model deployment to containers, monitoring with MLflow, and CI/CD integration.
• Fine-tune LLMs like GPT variants using Hugging Face Transformers, with techniques like LoRA for efficiency.