Runs

Runs#

This page discusses ways to manipulate your runs.

import mlflow
mlflow.set_tracking_uri("file:///tmp/runs_page")

Search runs#

Using mlflow.search_runs, you can load a set of runs that correspond to specified conditions. The most useful parameters are:

Parameter

Description

experiments_ids

The IDs of the experiments from which the runs will be retrieved

search_all_experiments

Returns the runs from all available experiments

The following cell generates a set of experiments, as well as a set of runs for each experiment.

for i in range(5):
    experiment_name = f"experiment {i + 1}"
    mlflow.create_experiment(experiment_name)
    mlflow.set_experiment(experiment_name=experiment_name)
    for j in range(5):
        mlflow.start_run(run_name = f"run {j + 1}")
        mlflow.end_run()

The following cell shows the runs of the experiment 1.

mlflow.search_runs(experiment_ids=[mlflow.get_experiment_by_name("experiment 1").experiment_id])
run_id experiment_id status artifact_uri start_time end_time tags.mlflow.user tags.mlflow.runName tags.mlflow.source.name tags.mlflow.source.type
0 1d3b84196bf04311b515df92ded6e29b 460233828277669480 FINISHED file:///tmp/runs_page/460233828277669480/1d3b8... 2025-09-26 12:37:43.683000+00:00 2025-09-26 12:37:43.685000+00:00 user run 5 /home/user/.virtualenvironments/python/lib/pyt... LOCAL
1 7d0de82b412d421aac32258101c81136 460233828277669480 FINISHED file:///tmp/runs_page/460233828277669480/7d0de... 2025-09-26 12:37:43.680000+00:00 2025-09-26 12:37:43.682000+00:00 user run 4 /home/user/.virtualenvironments/python/lib/pyt... LOCAL
2 c6d2dde7251a4847941a1abe664832e6 460233828277669480 FINISHED file:///tmp/runs_page/460233828277669480/c6d2d... 2025-09-26 12:37:43.676000+00:00 2025-09-26 12:37:43.679000+00:00 user run 3 /home/user/.virtualenvironments/python/lib/pyt... LOCAL
3 d8ac0efee591417bb46b36614c037ccd 460233828277669480 FINISHED file:///tmp/runs_page/460233828277669480/d8ac0... 2025-09-26 12:37:43.671000+00:00 2025-09-26 12:37:43.675000+00:00 user run 2 /home/user/.virtualenvironments/python/lib/pyt... LOCAL
4 5e93a7b1b5bc418cb7aa59a526b7e2d0 460233828277669480 FINISHED file:///tmp/runs_page/460233828277669480/5e93a... 2025-09-26 12:37:43.664000+00:00 2025-09-26 12:37:43.669000+00:00 user run 1 /home/user/.virtualenvironments/python/lib/pyt... LOCAL

The same, but also runs of the experiment 2 are added.

experiments_ids = [
    mlflow.get_experiment_by_name("experiment 1").experiment_id,
    mlflow.get_experiment_by_name("experiment 2").experiment_id,
]
mlflow.search_runs(experiment_ids=experiments_ids)
run_id experiment_id status artifact_uri start_time end_time tags.mlflow.user tags.mlflow.runName tags.mlflow.source.name tags.mlflow.source.type
0 e6c7219efbe34a8582a4500a11d3ccac 459248949934636608 FINISHED file:///tmp/runs_page/459248949934636608/e6c72... 2025-09-26 12:37:43.700000+00:00 2025-09-26 12:37:43.702000+00:00 user run 5 /home/user/.virtualenvironments/python/lib/pyt... LOCAL
1 ddc7e8c1884c421d885158211f9a5acd 459248949934636608 FINISHED file:///tmp/runs_page/459248949934636608/ddc7e... 2025-09-26 12:37:43.697000+00:00 2025-09-26 12:37:43.699000+00:00 user run 4 /home/user/.virtualenvironments/python/lib/pyt... LOCAL
2 20fd7890661a4ca6bd2a56d61fb88dfc 459248949934636608 FINISHED file:///tmp/runs_page/459248949934636608/20fd7... 2025-09-26 12:37:43.694000+00:00 2025-09-26 12:37:43.696000+00:00 user run 3 /home/user/.virtualenvironments/python/lib/pyt... LOCAL
3 4ff4fdacb5bf408092498a187f8e6a0a 459248949934636608 FINISHED file:///tmp/runs_page/459248949934636608/4ff4f... 2025-09-26 12:37:43.691000+00:00 2025-09-26 12:37:43.693000+00:00 user run 2 /home/user/.virtualenvironments/python/lib/pyt... LOCAL
4 4706fd7e7596464b9d25deee41f9b2db 459248949934636608 FINISHED file:///tmp/runs_page/459248949934636608/4706f... 2025-09-26 12:37:43.688000+00:00 2025-09-26 12:37:43.690000+00:00 user run 1 /home/user/.virtualenvironments/python/lib/pyt... LOCAL
5 1d3b84196bf04311b515df92ded6e29b 460233828277669480 FINISHED file:///tmp/runs_page/460233828277669480/1d3b8... 2025-09-26 12:37:43.683000+00:00 2025-09-26 12:37:43.685000+00:00 user run 5 /home/user/.virtualenvironments/python/lib/pyt... LOCAL
6 7d0de82b412d421aac32258101c81136 460233828277669480 FINISHED file:///tmp/runs_page/460233828277669480/7d0de... 2025-09-26 12:37:43.680000+00:00 2025-09-26 12:37:43.682000+00:00 user run 4 /home/user/.virtualenvironments/python/lib/pyt... LOCAL
7 c6d2dde7251a4847941a1abe664832e6 460233828277669480 FINISHED file:///tmp/runs_page/460233828277669480/c6d2d... 2025-09-26 12:37:43.676000+00:00 2025-09-26 12:37:43.679000+00:00 user run 3 /home/user/.virtualenvironments/python/lib/pyt... LOCAL
8 d8ac0efee591417bb46b36614c037ccd 460233828277669480 FINISHED file:///tmp/runs_page/460233828277669480/d8ac0... 2025-09-26 12:37:43.671000+00:00 2025-09-26 12:37:43.675000+00:00 user run 2 /home/user/.virtualenvironments/python/lib/pyt... LOCAL
9 5e93a7b1b5bc418cb7aa59a526b7e2d0 460233828277669480 FINISHED file:///tmp/runs_page/460233828277669480/5e93a... 2025-09-26 12:37:43.664000+00:00 2025-09-26 12:37:43.669000+00:00 user run 1 /home/user/.virtualenvironments/python/lib/pyt... LOCAL

Child & Parent#

MLFlow allows the creation of nested runs. There is an outer run - the parent run and an inner run - the child run. To create a child run, you need to start another run while one is running and pass nested=True to the mlflow.start_run.

Check out the Understanding Parent and Child Runs in MLflow tutorial for more details.

The following cell creates a “parent” run, and inside it iteratively starts some other runs with the names child_0, child_1 and so on.

mlflow.set_experiment("nested_runs")
with mlflow.start_run(run_name="parent"):
    for i in range(10):
        with mlflow.start_run(run_name=f"child_{i}", nested=True):
            pass

🏃 View run child_0 at: http://localhost:5000/#/experiments/362948453640932348/runs/2562f6f88bd24fd18e590fefc985cc33
🧪 View experiment at: http://localhost:5000/#/experiments/362948453640932348
🏃 View run child_1 at: http://localhost:5000/#/experiments/362948453640932348/runs/9b0233bb13e045e581d79a73e1ef1f73
🧪 View experiment at: http://localhost:5000/#/experiments/362948453640932348
🏃 View run child_2 at: http://localhost:5000/#/experiments/362948453640932348/runs/17799d3fd383422c8ebf2086237854e2
🧪 View experiment at: http://localhost:5000/#/experiments/362948453640932348
🏃 View run child_3 at: http://localhost:5000/#/experiments/362948453640932348/runs/c1283382a89549959c423a12fe09b6fb
🧪 View experiment at: http://localhost:5000/#/experiments/362948453640932348
🏃 View run child_4 at: http://localhost:5000/#/experiments/362948453640932348/runs/4b1163851f0d46eeaebf210102b5d1ab
🧪 View experiment at: http://localhost:5000/#/experiments/362948453640932348
🏃 View run child_5 at: http://localhost:5000/#/experiments/362948453640932348/runs/3fdd20b7cfdb4746b44ce73eb9e19bc9
🧪 View experiment at: http://localhost:5000/#/experiments/362948453640932348
🏃 View run child_6 at: http://localhost:5000/#/experiments/362948453640932348/runs/74eaf04c30c043edbbf6eec9a48b5866
🧪 View experiment at: http://localhost:5000/#/experiments/362948453640932348
🏃 View run child_7 at: http://localhost:5000/#/experiments/362948453640932348/runs/e1d880c924ec4a7cb896a043f29615ed
🧪 View experiment at: http://localhost:5000/#/experiments/362948453640932348
🏃 View run child_8 at: http://localhost:5000/#/experiments/362948453640932348/runs/c292b24e66e24fd7aa8ea632e478388e
🧪 View experiment at: http://localhost:5000/#/experiments/362948453640932348
🏃 View run child_9 at: http://localhost:5000/#/experiments/362948453640932348/runs/2225de115c2e45e0be8c8149dc8fb85c
🧪 View experiment at: http://localhost:5000/#/experiments/362948453640932348
🏃 View run parent at: http://localhost:5000/#/experiments/362948453640932348/runs/a6793e5651944acea30396464bb3bdd7
🧪 View experiment at: http://localhost:5000/#/experiments/362948453640932348

The next code displays the names of the runs and the parent ids of the runs that were just created.

id = mlflow.get_experiment_by_name("nested_runs").experiment_id
runs = mlflow.search_runs(experiment_ids=id)
runs.loc[:, ["run_id", "tags.mlflow.runName", "tags.mlflow.parentRunId"]]
run_id tags.mlflow.runName tags.mlflow.parentRunId
0 2225de115c2e45e0be8c8149dc8fb85c child_9 a6793e5651944acea30396464bb3bdd7
1 c292b24e66e24fd7aa8ea632e478388e child_8 a6793e5651944acea30396464bb3bdd7
2 e1d880c924ec4a7cb896a043f29615ed child_7 a6793e5651944acea30396464bb3bdd7
3 74eaf04c30c043edbbf6eec9a48b5866 child_6 a6793e5651944acea30396464bb3bdd7
4 3fdd20b7cfdb4746b44ce73eb9e19bc9 child_5 a6793e5651944acea30396464bb3bdd7
5 4b1163851f0d46eeaebf210102b5d1ab child_4 a6793e5651944acea30396464bb3bdd7
6 c1283382a89549959c423a12fe09b6fb child_3 a6793e5651944acea30396464bb3bdd7
7 17799d3fd383422c8ebf2086237854e2 child_2 a6793e5651944acea30396464bb3bdd7
8 9b0233bb13e045e581d79a73e1ef1f73 child_1 a6793e5651944acea30396464bb3bdd7
9 2562f6f88bd24fd18e590fefc985cc33 child_0 a6793e5651944acea30396464bb3bdd7
10 a6793e5651944acea30396464bb3bdd7 parent None

As a result, in the tags.mlflow.parentRunId column, for the “parent” run there is None, but all other runs refer to the “parent” through this column.

Note that in the mlflow GUI runs are ordered in a hierachal order - check how to work with it in the Understanding Parent and Child Runs in MLflow tutorial.

Contents