Important

Adds a recipe for supervised fine-tuning using Unsloth, including setup, data preparation, model training, and deployment.

• Setup:
  • Adds devcontainer.json for Unsloth development environment setup.
  • Specifies dependencies in pyproject.toml and requirements.txt.
• Script (unsloth_nb.py):
  • Demonstrates supervised fine-tuning using Unsloth with TensorZero data.
  • Configures model parameters, data filtering, and training settings.
  • Implements data preparation, model instantiation, and tokenizer setup.
  • Converts TensorZero inferences to ChatML format for training.
  • Splits data into training and validation sets.
  • Configures and trains model with optional LoRA adaptation.
  • Deploys fine-tuned model using Fireworks for serverless inference.
  • Provides deployment and configuration instructions for the fine-tuned model.

^{This description was created by for 0c66447. You can customize this summary. It will automatically update as commits are pushed.}