Run an SFT experiment¶

Train a supervised fine-tune in an Advanced Training project: attach an SFT dataset and pick algorithm SFT.

Prerequisites¶

Advanced Training project with an SFT dataset (prompt / target columns).
See SFT datasets if you still need to upload data.

1. Open the project¶

On Experiments, open your Advanced Training project.

Advanced Training project experiments table — The Advanced Training project page.¶

2. New experiment¶

Click New experiment, name it, and save to open the wizard.

New experiment dialog in advanced project — Creating a new experiment.¶

3. Environment step¶

On Environment, select your SFT dataset. The preview shows Prompt and Target.

Wizard dataset selection step with SFT dataset — Selecting an SFT dataset in the wizard.¶

4. Agent¶

Choose SFT and a pretrained model. Tune hyperparameters if you need to. Click Next.

Wizard agent step with SFT selected — The **Agent** step with **SFT**.¶

5. Training¶

Set training steps and related options, then continue to HPO or Summary.

Wizard training step — The **Training** step.¶

6. Summary¶

Review the configuration on Summary.

Wizard summary before train — **Summary** before starting the run.¶

7. Running experiment¶

Click Train. The experiment row shows status while the job runs.

Experiments table with running status — An experiment with an active training status.¶

From the terminal¶

Save a default spec from Python as run-spec.json, then submit. Copy a resource id from arena resources list.

Install the CLI and sign in (Authentication):

pip install "agilerl[arena]"
arena login
# or: export ARENA_API_KEY=arena_pat_…

import json
from pathlib import Path

from agilerl.arena import ArenaClient

with ArenaClient() as client:
    spec = client.get_default_run_spec(
        algorithm="SFT",
        dataset_name="support-tickets-sft",
    )
    Path("run-spec.json").write_text(json.dumps(spec, indent=2))

arena resources list

arena experiments submit ./run-spec.json --project llm-work --experiment-name sft-run --resource-id a2-highgpu-1g --num-nodes 2

Python¶

from agilerl.arena import ArenaClient

with ArenaClient() as client:
    spec = client.get_default_run_spec(
        algorithm="SFT",
        dataset_name="support-tickets-sft",
    )
    client.submit_experiment(
        manifest=spec,
        project="llm-work",
        resource_id="a2-highgpu-1g",
        experiment_name="sft-run",
        num_nodes=2,
    )
    print(client.list_experiments(project="llm-work"))