diff --git a/doc/python/pipeline_example_dask.py b/doc/python/pipeline_example_dask.py index e1df7d89e0f1221b04f54e158a097cfa8db06146..d6e365a1570b5ab50ed48f304f5b614b5f1a9f5b 100644 --- a/doc/python/pipeline_example_dask.py +++ b/doc/python/pipeline_example_dask.py @@ -25,7 +25,7 @@ class MyFitTranformer(TransformerMixin, BaseEstimator): def __init__(self): self._fit_model = None - def transform(self, X): + def transform(self, X, metadata=None): # Transform `X` return [x @ self._fit_model for x in X] @@ -40,7 +40,7 @@ X = numpy.zeros((2, 2)) X_as_sample = [Sample(X, key=str(i), metadata=1) for i in range(10)] # Building an arbitrary pipeline -model_path = "~/dask_tmp" +model_path = "./dask_tmp" os.makedirs(model_path, exist_ok=True) pipeline = make_pipeline(MyTransformer(), MyFitTranformer()) @@ -48,13 +48,15 @@ pipeline = make_pipeline(MyTransformer(), MyFitTranformer()) pipeline = bob.pipelines.wrap( ["sample", "checkpoint", "dask"], pipeline, - model_path=model_path, + model_path=os.path.join(model_path, "model.pickle"), + features_dir=model_path, transform_extra_arguments=(("metadata", "metadata"),), ) # Create a dask graph from a pipeline # Run the task graph in the local computer in a single tread X_transformed = pipeline.fit_transform(X_as_sample).compute(scheduler="single-threaded") + import shutil shutil.rmtree(model_path) diff --git a/doc/python/pipeline_example_dask_sge.py b/doc/python/pipeline_example_dask_sge.py index eba1709ec9655c1d9fdedba2ae3a7a76bac4a05a..ae071657619c18ee618286cacfd37f155cdd7bc7 100644 --- a/doc/python/pipeline_example_dask_sge.py +++ b/doc/python/pipeline_example_dask_sge.py @@ -28,7 +28,7 @@ class MyFitTranformer(TransformerMixin, BaseEstimator): def __init__(self): self._fit_model = None - def transform(self, X): + def transform(self, X, metadata=None): # Transform `X` return [x @ self._fit_model for x in X] @@ -43,7 +43,7 @@ X = numpy.zeros((2, 2)) X_as_sample = [Sample(X, key=str(i), metadata=1) for i in range(10)] # Building an arbitrary pipeline -model_path = "~/dask_tmp" +model_path = "./dask_tmp" os.makedirs(model_path, exist_ok=True) pipeline = make_pipeline(MyTransformer(), MyFitTranformer()) @@ -51,7 +51,8 @@ pipeline = make_pipeline(MyTransformer(), MyFitTranformer()) pipeline = bob.pipelines.wrap( ["sample", "checkpoint", "dask"], pipeline, - model_path=model_path, + model_path=os.path.join(model_path, "model.pickle"), + features_dir=model_path, transform_extra_arguments=(("metadata", "metadata"),), ) @@ -62,6 +63,7 @@ client = Client(cluster) # Creating the scheduler # Run the task graph in the local computer in a single tread X_transformed = pipeline.fit_transform(X_as_sample).compute(scheduler=client) + import shutil shutil.rmtree(model_path) diff --git a/doc/python/pipeline_example_dask_sge_adaptive.py b/doc/python/pipeline_example_dask_sge_adaptive.py index 5f9e791c582d162804eaf94da57779efbb085570..5fab7e92920992004f8ee2a3fcbff16374ae581f 100644 --- a/doc/python/pipeline_example_dask_sge_adaptive.py +++ b/doc/python/pipeline_example_dask_sge_adaptive.py @@ -28,7 +28,7 @@ class MyFitTranformer(TransformerMixin, BaseEstimator): def __init__(self): self._fit_model = None - def transform(self, X): + def transform(self, X, metadata=None): # Transform `X` return [x @ self._fit_model for x in X] @@ -43,7 +43,7 @@ X = numpy.zeros((2, 2)) X_as_sample = [Sample(X, key=str(i), metadata=1) for i in range(10)] # Building an arbitrary pipeline -model_path = "~/dask_tmp" +model_path = "./dask_tmp" os.makedirs(model_path, exist_ok=True) pipeline = make_pipeline(MyTransformer(), MyFitTranformer())