Skip to content

Project settings

Settings define what the agent optimizes and how an experiment runs. They're the project-level defaults the agent respects on every experiment.

View

autolab settings

Shows the current objective, constraints, run/setup commands, and the assigned LLM key.

Edit

Pass any flag to edit those fields, then it prints the updated settings:

autolab settings --objective "minimize validation bits-per-byte"
autolab settings --constraints "no extra params; train < 2h on 8xH100"
autolab settings --run "python train.py --config big.yaml" --prep "uv sync"
Flag Meaning
--objective What to optimize — the research goal (your metric, in words).
--constraints Rules the agent must respect.
--run How one experiment runs.
--prep Environment setup before the run (e.g. uv sync).
--name Display name.
--description Short project description.
--source Source repo URL.

Clear a field with an empty string

autolab settings --constraints ""

Objective, metric & constraints

Autolab optimizes a single objective stated in plain language — minimize val_loss, maximize MMLU, or even an LLM-as-judge comparison. The agent reads the metric out of your run's logs and the objective text, so be specific about the number you care about. Constraints are the guardrails: parameter budgets, time limits, "don't touch the tokenizer", and so on.

Stop policy & cost cap

A project's termination condition (--stop-policy) and hard spend cap (--max-cost) are set when you create it:

autolab init -y --name nanochat --objective "min val loss" \
  --stop-policy "stop after 30 experiments with no improvement" \
  --max-cost 100

Adjust them later from the dashboard project settings.

Collaborators

Managing who can access the project has its own subcommand — see Collaborators:

autolab settings collab          # list members

Delete a project

Owners only, and permanent:

autolab settings delete          # confirms first
autolab settings delete --yes    # skip the confirmation

Changing the objective or constraints steers the live agent on its next experiment. To pause it first, see Drive the agent.