mshumer/gpt-llm-trainer ? reverse-engineered prompt

Build me a Colab or Jupyter notebook that lets me describe a task in plain English, then automatically creates a training dataset for that task, writes a good system prompt, splits the data into training and validation sets, and fine tunes a model for me.

I want it to work with either a LLaMA 2 style model or GPT 3.5, and I should be able to set how creative the generated examples are and how many examples to make. After training, I want to test the model right in the notebook, then save it so I can use it later.

Make the whole flow as simple as possible, since I’m trying to go from an idea to a working task specific model without having to build the dataset or training pipeline by hand. If you need to check current notebook or API details online, go ahead and look them up.