train-sft

What problem does it solve?

This Skill helps users fine-tune language models using Supervised Fine-Tuning (SFT) with the ART framework, enabling them to create custom models for specific tasks.

Core Features & Use Cases

• SFT Training: Guides users through setting up and generating SFT training scripts.
• Dataset Handling: Supports training from JSONL files or distillation from a teacher model.
• Backend Options: Offers both ServerlessBackend (managed GPUs) and LocalBackend (local GPU) training.
• Use Case: Train a specialized chatbot for customer support by fine-tuning a base model on a dataset of past customer interactions.

Quick Start

Use the train-sft skill to start a new SFT training process.