Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.veri.studio/llms.txt

Use this file to discover all available pages before exploring further.

​
Overview

This cookbook is a set of practical starting points for common fine-tuning workflows in Veri. It is intentionally lightweight for now: the goal is to help you get unstuck quickly, not bury you in framework-specific details.

​
Common Starting Points

​
Fine-tune for structured outputs

Use this when you want:
  • valid JSON responses
  • predictable schemas
  • strict formatting requirements
Best paired with:

​
Connect a dataset from an existing source

Use this when your training data already lives in:
  • S3
  • GCS
  • Azure Blob
  • Hugging Face
  • Postgres or another database
Best paired with:

​
Fine-tune for task-specific quality

Use this when you want the model to become better at:
  • a domain-specific task
  • a constrained format
  • a known evaluation rubric
Best paired with:

​
Example Workflow

{
  "base_model": "Qwen/Qwen3-4B",
  "dataset_id": "ds_abc123",
  "reward_function_id": "rf_def456",
  "output_name": "veri-cookbook-run",
  "hyperparameters": {
    "learning_rate": 1e-6,
    "rollouts_per_prompt": 8,
    "max_response_length": 2048
  }
}

​
More Resources