Runs

Track training runs and experiments.

All methods are accessed via client.runs.

create_run()

POST/v1/runs

Create a new training run.

Parameters

team_idstrRequired

Team ID for the run

user_idstrRequired

User ID who owns the run

run_idstrdefault: None

Optional run ID (will be generated if not provided)

model_idstrdefault: None

Optional associated model ID

namestrdefault: None

Optional run name

metadataDictdefault: None

Optional metadata dictionary

Returns

str — The run ID (either provided or newly generated)

Example

1result = client.runs.create_run(

2 team_id="team_abc123",

3 user_id="user_abc123",

4 run_id="run_abc123",

5 model_id="model_abc123",

6 name="My Resource",

7 metadata="..."

8)

---

list_team_runs()

GET/v1/runs/teams/{team_id}

List the team's runs, newest first. A run records one piece of work on the platform: training a model, refitting it, or an agentic session. It is the thread reports and lineage hang off, so this is how you find the run_id a report needs when you were not the one who created the model. Pass model_id to get just that model's runs; the newest is the run that produced its latest version.

Parameters

model_idstrdefault: None

Only runs that produced this model (optional).

team_idstrdefault: None

Team to list for; defaults to the active team.

limitintdefault: 50

Maximum runs to return (newest first).

offsetintdefault: 0

Skip this many, for paging.

Returns

list — List of runs, each with run_id, name, model_id, dataset_id, status and created.

Example

1result = client.runs.list_team_runs(

2 model_id="model_abc123",

3 team_id="team_abc123",

4 limit=1,

5 offset=1

6)

---

get_run()

GET/v1/runs/{run_id}

Get run details by ID.

Parameters

run_idstrRequired

The run ID

Returns

Dict — Run details dictionary

Example

1result = client.runs.get_run(

2 run_id="run_abc123"

3)

---