[Q] How to import runs from MLflow

[Dariusz Kajtoch]

Hi, I am using the following piece of code from documentation to import runs from mlflow. I do not know what should be the argument dst_base_url and dst_api_key. I am not hosting my own server just web version.

import mlflow
from wandb.apis.importers.mlflow import MlflowRun
from wandb.apis.importers.mlflow import MlflowImporter
from tqdm.auto import tqdm
from wandb.apis.importers import Namespace

client = mlflow.tracking.MlflowClient("http://localhost:6100")

project = "my_project"
runs = []
for run in tqdm(get_runs(client, experiment_name=project)):
    runs.append(MlflowRun(run, client))

importer = MlflowImporter(
    dst_base_url="",
    dst_api_key="my_api_key",
    mlflow_tracking_uri="http://localhost:6100"
)

for run in tqdm(runs):
    importer._import_run(run, namespace=Namespace(entity, project), artifacts=True)

[fmamberti-wandb]

Hi @dkajtoch, thank you for reaching out.
If you are using W&B cloud platform the dst_base_url should be https://api.wandb.ai (you can see this running the wandb status command via terminal) while dst_api_key should be your api key for the platform that you can find at https://wandb.ai/authorize.

Please let me know if you have any further questions.

[fmamberti-wandb]

Hi @dkajtoch I wanted to follow up on this request. Please let us know if we can be of further assistance or if your issue has been resolved.

[fmamberti-wandb]

Hi @dkajtoch , since we have not heard back from you we are going to close this request. If you would like to re-open the conversation, please let us know!