search

google / ml-metadata

For recording and retrieving metadata associated with ML developer and data scientist workflows.
https://www.tensorflow.org/tfx/guide/mlmd
Apache License 2.0
626 stars 148 forks source link

Suitable for computer vision project? #170

Closed ryo-matsuzaka closed 1 year ago

ryo-matsuzaka commented 1 year ago

Hi there, I am a noob of ML Metadata.
I wanna manage metadata of computer vision project.
Is ML Metadata is suitable for its purpose?
When I see get started, it uses RDBMS such as MySQL so I guess it is beneficial to such data type?

BrianSong commented 1 year ago

Hi @ryo-matsuzaka, if the "metadata of computer vision project" means the metadata for ML training related metadata, then yes, MLMD definitely is suitable for it.

I am not sure if I understand your second question though, could you be more specific about the "data type"?

ryo-matsuzaka commented 1 year ago

Hi @BrianSong , thank you very much for your answer.

if the "metadata of computer vision project" means the metadata for ML training related metadata

Yes, that is what I meant.

I am not sure if I understand your second question though, could you be more specific about the "data type"?

Sorry for my vague question. I am not familiar with SQL so I imagined that MLMD does not support string type such as explanations about experiments. But now I found the following description in the get started documentation and understood it supports the feature I want. I think I can write and store any descriptions about ML experiments.

# Group the model and the trainer run to an experiment.
my_experiment = metadata_store_pb2.Context()
my_experiment.type_id = experiment_type_id
# Give the experiment a name
my_experiment.name = "exp1"
my_experiment.properties["note"].string_value = "My first experiment."
[experiment_id] = store.put_contexts([my_experiment])