gokart
gokart copied to clipboard
m3dev
Gokart solves reproducibility, task dependencies, constraints of good code, and ease of use for Machine Learning Pipeline.
Thanks for this useful module. Let me report a strange behavior. One of my tasks dumps a dict with length 1 (like `InputTask` as an example below). When I load...
I would like to add `.lmdb` to the file formats supported by TaskOnKart.make_target(). `.lmdb` is the format used by several popular datasets, and is actually suitable for handling large datasets...
https://github.com/m3dev/gokart/blob/489e2af926074d4c702c1c857b52834f664c1e81/gokart/__init__.py#L1 This line overwrites gokart.build module with gokart.build() function. We can't use `gokart.build.GokartBuildError ` directly. `gokart.build` module might ought to be renamed.
Hello, thank you for developing such a great tool! ## Summary I have one feature request to validate `TaskInstanceParameter()` at runtime by its subclass bound like the following: ```python class...
I'll create a task like `Function` that won't be saved. for example ```python import gokart class Pipeline(gokart.TaskOnKart): def requires(self): data = LoadData() features = [MakeFeatureA(data=data), MakeFeatureB(data=data), MakeFeatureC(data=data)] # `Flatten` is...
I would like to see a function that can be used for tests where you don't want to leave a pkl. now ``` df = gokart.build(Foo()) # dump pkl ./resource/......
I want to be able to create dependent tasks easily for regression test. image: ``` X = gokart.MockTask(dump_value='hoge') class Example(gokart.TaskOnKart): target = gokart.TaskInstanceParameter() ... task = Example(target=X) df = gokart.build(task)...
Gokart is based on a DataFrame. But `pd.testing.assert_frame_equal` is not a simple way to compare df. This is especially true for sort and index problems. now ``` expected = pd.DataFrame(...)...
I want a mock task instance generator to make test more convenient. now ``` from_date = datetime.date(2020, 1, 1) to_date = datetime.date(2020, 1, 5) target = gokart.TaskOnKart() task = ExampleTask(from_date=from_date,...
Task parameters can be set at `gokart.run()` as following example. ``` gokart.run(['SampleTask', '--local-scheduler', '--param=hello']) ``` When using `gokart.build()` this is not possible, which makes changing parameters bothersome.
