def split_key_value_store(
input_store: KeyValueStore,
*,
num_parts: int,
random_seed: Optional[int] = None) -> Tuple[KeyValueStore]:
"""
Splits *input_store* into *num_parts* pieces of nearly equal size.
Some of the resulting key-value stores may be empty.
"""
if num_parts <= 0:
raise RuntimeError("Number of parts must be positive")
split_locator = input_store.locator / "split"
split_output_dir = directory_for(split_locator)
param_args = {
"input": input_store.input_parameters(),
"num_slices": num_parts,
"output_dir": split_output_dir,
}
if random_seed:
param_args["random_seed"] = random_seed
split_job = run_python_on_parameters(
split_locator,
split_entry_point,
Parameters.from_mapping(param_args),
depends_on=input_store,
)
return tuple(
ZipKeyValueStore(
path=split_output_dir / f"{slice_index}.zip",
depends_on=split_job,
locator=split_locator / str(slice_index),
) for slice_index in range(num_parts))
def downsample(input_store: KeyValueStore,
*,
limit: int,
output_locator: Optional[Locator] = None) -> KeyValueStore:
"""
Convince function to run `vistautils.scripts.downsample_key_value_store` as a Pegasus Job
"""
if not output_locator:
output_locator = input_store.locator / f"downsampled-{limit}"
output_zip_path = directory_for(output_locator) / "downsampled.zip"
downsample_job = run_python_on_parameters(
output_locator,
downsample_key_value_store,
Parameters.from_mapping({
"input": input_store.input_parameters(),
"output_zip_path": output_zip_path,
"num_to_sample": limit,
"random_seed": 0,
}),
depends_on=input_store,
)
return ZipKeyValueStore(
path=output_zip_path,
locator=output_locator,
depends_on=[input_store.depends_on, downsample_job],
)
def test_not_clearing_ckpts(monkeypatch, tmp_path):
workflow_params = Parameters.from_mapping({
"workflow_name":
"Test",
"workflow_created":
"Testing",
"workflow_log_dir":
str(tmp_path / "log"),
"workflow_directory":
str(tmp_path / "working"),
"site":
"saga",
"namespace":
"test",
"partition":
"scavenge",
"home_dir":
str(tmp_path),
})
initialize_vista_pegasus_wrapper(workflow_params)
multiply_job_name = Locator(_parse_parts("jobs/multiply"))
multiply_output_file = tmp_path / "multiplied_nums.txt"
multiply_input_file = tmp_path / "raw_nums.txt"
multiply_params = Parameters.from_mapping({
"input_file": multiply_input_file,
"output_file": multiply_output_file,
"x": 4
})
multiple_dir = directory_for(multiply_job_name)
checkpointed_multiply_file = multiple_dir / "___ckpt"
checkpointed_multiply_file.touch()
multiply_output_file.touch()
run_python_on_parameters(multiply_job_name,
multiply_by_x_main,
multiply_params,
depends_on=[])
monkeypatch.setattr("builtins.input", lambda _: "n")
write_workflow_description()
assert checkpointed_multiply_file.exists()
def join_to_key_value_zip(key_value_zips_to_join: Iterable[ZipKeyValueStore],
*, output_locator: Locator) -> ZipKeyValueStore:
key_value_zips_to_join = tuple(key_value_zips_to_join)
output_zip_path = directory_for(output_locator) / "joined.zip"
join_job = run_python_on_parameters(
output_locator,
join_key_value_stores,
Parameters.from_mapping({
"input_store_list_file": [p.path for p in key_value_zips_to_join],
"output": {
"type": "zip",
"path": output_zip_path
},
}),
depends_on=key_value_zips_to_join,
)
return ZipKeyValueStore(path=output_zip_path,
locator=output_locator,
depends_on=join_job)
def gaze_ablation_runner_entry_point(params: Parameters) -> None:
"""This function creates all possible gaze ablation param files within a given range"""
initialize_vista_pegasus_wrapper(params)
# Get the baseline experiment parameters for gaze ablation -- these are things common to all of
# the experiments, like:
#
# include_image_links: true
# sort_learner_descriptions_by_length: True
# num_pretty_descriptions: 5
baseline_parameters = params.namespace("gaze_ablation")
# get the minimum and maximum number of objects in a scene
min_num_objects = params.integer("min_num_objects", default=1)
max_num_objects = params.integer("max_num_objects", default=7)
# this gets the number of different accuracies to try; default = increment by 0.1
num_accuracy_increments = params.integer("num_increments", default=11)
values_for_accuracy = np.linspace(0, 1, num_accuracy_increments)
# the number of noise instances to be included
min_num_noise_instances = params.integer("min_num_noise", default=0)
max_num_noise_instances = params.integer("max_num_noise", default=0)
# get the number of instances in the entire curriculum
min_num_instances_in_curriculum = params.integer("min_instances",
default=10)
max_num_instances_in_curriculum = params.integer("max_instances",
default=20)
# all possible numbers of noise instances
for num_noise_instances in range(min_num_noise_instances,
max_num_noise_instances + 1):
# all possible numbers of instances in the curriculum
for num_instances in range(min_num_instances_in_curriculum,
max_num_instances_in_curriculum + 1):
# all possible numbers of instances
for num_objects_in_instance in range(min_num_objects,
max_num_objects + 1):
# all possible accuracies
for prob_given in values_for_accuracy:
for prob_not_given in values_for_accuracy:
# both ignoring and perceiving gaze
for add_gaze in [True, False]:
# Define the experiment name, which is used both as a job name and to
# choose a directory in which to store the experiment results.
experiment_name_string = EXPERIMENT_NAME_FORMAT.format(
num_instances=num_instances,
num_noise_instances=num_noise_instances,
num_objects_in_instance=num_objects_in_instance,
prob_given=prob_given,
prob_not_given=prob_not_given,
add_gaze=add_gaze,
)
experiment_name = Locator(
experiment_name_string.split("-"))
# Note that the input parameters should include the root params and
# anything else we want.
experiment_params = baseline_parameters.unify(
FIXED_PARAMETERS).unify({
"experiment":
experiment_name_string,
"experiment_group_dir":
directory_for(experiment_name),
"hypothesis_log_dir":
directory_for(experiment_name) /
"hypotheses",
"learner_logging_path":
directory_for(experiment_name),
"log_learner_state":
True,
"resume_from_latest_logged_state":
True,
"pursuit-curriculum-params": {
"num_instances": num_instances,
"num_noise_instances":
num_noise_instances,
"num_objects_in_instance":
num_objects_in_instance,
"add_gaze": add_gaze,
"prob_given": float(prob_given),
"prob_not_given":
float(prob_not_given),
},
})
run_python_on_parameters(
experiment_name,
log_experiment_script,
experiment_params,
depends_on=[],
)
write_workflow_description()
def main(params: Parameters):
adam_root = params.existing_directory("adam_root")
m13_experiments_dir = adam_root / "parameters" / "experiments" / "m13"
use_pegasus = params.boolean("use_pegasus", default=False)
if use_pegasus:
initialize_vista_pegasus_wrapper(params)
param_files: List[Path] = []
if params.boolean("include_objects", default=True):
param_files.append(m13_experiments_dir / "objects.params")
if params.boolean("include_imprecise_size", default=True):
param_files.append(m13_experiments_dir / "imprecise_size.params")
if params.boolean("include_imprecise_temporal", default=True):
param_files.append(m13_experiments_dir / "imprecise_temporal.params")
if params.boolean("include_subtle_verb", default=True):
param_files.append(m13_experiments_dir / "subtle_verb.params")
if params.boolean("include_object_restrictions", default=True):
param_files.append(m13_experiments_dir / "object_restrictions.params")
if params.boolean("include_functionally_defined_objects", default=True):
param_files.append(m13_experiments_dir / "functionally_defined_objects.params")
if params.boolean("include_relations", default=True):
param_files.append(m13_experiments_dir / "relations.params")
if params.boolean("include_generics", default=True):
param_files.append(m13_experiments_dir / "generics.params")
if params.boolean("include_verbs_with_dynamic_prepositions", default=True):
param_files.append(
m13_experiments_dir / "events_with_dynamic_prepositions.params"
)
if params.boolean("include_m9_complete", default=False):
param_files.append(m13_experiments_dir / "m9_complete.params")
if params.boolean("include_m13_complete", default=False):
param_files.append(m13_experiments_dir / "m13_complete.params")
if params.boolean("include_m13_shuffled", default=False):
param_files.append(m13_experiments_dir / "m13_shuffled.params")
# This activates a special "debug" curriculum,
# which is meant to be edited in the code by a developer to do fine-grained debugging.
if params.boolean("include_debug", default=False):
param_files.append(m13_experiments_dir / "debug.params")
# If any of the param files don't exist, bail out earlier instead of making the user
# wait for the error.
for param_file in param_files:
if not param_file.exists():
raise RuntimeError(f"Expected param file {param_file} does not exist")
for param_file in param_files:
logging.info("Running %s", param_file)
experiment_params = YAMLParametersLoader().load(param_file)
if not use_pegasus:
log_experiment_entry_point(experiment_params)
else:
experiment_name = Locator(experiment_params.string("experiment"))
experiment_params = experiment_params.unify(
{
"experiment_group_dir": directory_for(experiment_name) / "output",
"hypothesis_log_dir": directory_for(experiment_name) / "hypotheses",
# State pickles will go under experiment_name/learner_state
"learner_logging_path": directory_for(experiment_name),
"log_learner_state": True,
"resume_from_latest_logged_state": True,
"log_hypothesis_every_n_steps": params.integer(
"save_state_every_n_steps"
),
"debug_learner_pickling": params.boolean(
"debug_learner_pickling", default=False
),
}
)
run_python_on_parameters(
experiment_name, log_experiment_script, experiment_params, depends_on=[]
)
if use_pegasus:
write_workflow_description()
def test_dax_with_python_into_container_jobs(tmp_path):
docker_tar = Path(f"{tmp_path}/docker/tar.tar")
docker_build_dir = tmp_path
docker_image_name = "pegasus_wrapper_container_demo"
docker_image_tag = "0.2"
# Generating parameters for initializing a workflow
# We recommend making workflow directory, site, and partition parameters
# in an research workflow
workflow_params = Parameters.from_mapping({
"workflow_name":
"Test",
"workflow_created":
"Testing",
"workflow_log_dir":
str(tmp_path / "log"),
"workflow_directory":
str(tmp_path / "working"),
"site":
"saga",
"namespace":
"test",
"home_dir":
str(tmp_path),
"partition":
"scavenge",
})
saga31_request = SlurmResourceRequest.from_parameters(
Parameters.from_mapping({
"run_on_single_node": "saga31",
"partition": "gaia"
}))
# Our source input for the sample jobs
input_file = tmp_path / "raw_nums.txt"
add_y_output_file_nas = tmp_path / "nums_y.txt"
sorted_output_file_nas = tmp_path / "sorted.txt"
random = Random()
random.seed(0)
nums = [int(random.random() * 100) for _ in range(0, 25)]
# Base Job Locator
job_locator = Locator(("jobs", ))
docker_python_root = Path("/home/app/")
# Write a list of numbers out to be able to run the workflow
with input_file.open("w") as mult_file:
mult_file.writelines(f"{num}\n" for num in nums)
initialize_vista_pegasus_wrapper(workflow_params)
build_container_locator = job_locator / "build_docker"
build_container = run_bash(
build_container_locator,
command=[
"mkdir -p /scratch/dockermount/pegasus_wrapper_tmp",
f"cd {docker_build_dir}",
f"docker build . -t {docker_image_name}:{docker_image_tag}",
f"docker save -o /scratch/dockermount/pegasus_wrapper_tmp/{docker_tar.name} {docker_image_name}:{docker_image_tag}",
f"cp /scratch/dockermount/pegasus_wrapper_tmp/{docker_tar.name} {docker_tar.absolute()}",
f"chmod go+r {docker_tar.absolute()}",
],
depends_on=[],
resource_request=saga31_request,
)
build_container_dir = directory_for(build_container_locator)
assert (build_container_dir / "script.sh").exists()
python36 = add_container(
f"{docker_image_name}:{docker_image_tag}",
"docker",
str(docker_tar.absolute()),
image_site="saga",
bypass_staging=True,
)
job_profile = PegasusProfile(namespace="pegasus",
key="transfer.bypass.input.staging",
value="True")
mongo4_4 = add_container("mongo:4.4",
"docker",
"path/to/tar.tar",
image_site="saga",
bypass_staging=True)
with pytest.raises(RuntimeError):
_ = stop_docker_as_service(mongo4_4,
depends_on=[],
resource_request=saga31_request)
start_mongo = start_docker_as_service(
mongo4_4,
depends_on=[build_container],
docker_args=f"-v /scratch/mongo/data/db:/data/db",
resource_request=saga31_request,
)
mongo4_4_dir = directory_for(Locator(("containers", mongo4_4.name)))
assert (mongo4_4_dir / "start.sh").exists()
assert (mongo4_4_dir / "stop.sh").exists()
add_y_locator = job_locator / "add"
add_y_job = run_python_on_args(
add_y_locator,
docker_python_root / "add_y.py",
set_args=f"{input_file} {add_y_output_file_nas} --y 10",
depends_on=[build_container],
job_profiles=[job_profile],
resource_request=saga31_request,
container=python36,
input_file_paths=[input_file],
output_file_paths=[add_y_output_file_nas],
)
add_y_dir = directory_for(add_y_locator)
assert (add_y_dir / "___run.sh").exists()
with pytest.raises(RuntimeError):
_ = run_python_on_args(
add_y_locator,
docker_python_root / "add_y.py",
set_args=f"{input_file} {add_y_output_file_nas} --y 10",
depends_on=[build_container],
job_profiles=[job_profile],
resource_request=saga31_request,
container=python36,
input_file_paths=[input_file, input_file],
output_file_paths=[add_y_output_file_nas],
)
sort_job_locator = job_locator / "sort"
sort_job = run_python_on_parameters(
sort_job_locator,
sort_nums_main,
{
"input_file": add_y_output_file_nas,
"output_file": sorted_output_file_nas
},
depends_on=[add_y_job],
container=python36,
job_profiles=[job_profile],
resource_request=saga31_request,
input_file_paths=add_y_output_file_nas,
output_file_paths=sorted_output_file_nas,
)
assert sort_job == run_python_on_parameters(
sort_job_locator,
sort_nums_main,
{
"input_file": add_y_output_file_nas,
"output_file": sorted_output_file_nas
},
depends_on=[add_y_job],
container=python36,
job_profiles=[job_profile],
resource_request=saga31_request,
input_file_paths=add_y_output_file_nas,
output_file_paths=sorted_output_file_nas,
)
sort_job_dir = directory_for(sort_job_locator)
assert (sort_job_dir / "___run.sh").exists()
assert (sort_job_dir / "____params.params").exists()
with pytest.raises(RuntimeError):
_ = run_python_on_parameters(
sort_job_locator,
sort_nums_main,
{
"input_file": add_y_output_file_nas,
"output_file": sorted_output_file_nas
},
depends_on=[add_y_job],
container=python36,
job_profiles=[job_profile],
resource_request=saga31_request,
input_file_paths=add_y_output_file_nas,
output_file_paths=[sorted_output_file_nas, sorted_output_file_nas],
)
celebration_bash_locator = job_locator / "celebrate"
celebration_bash = run_bash(
celebration_bash_locator,
'echo "Jobs Runs Successfully"',
depends_on=[sort_job],
job_profiles=[job_profile],
)
assert celebration_bash == run_bash(
celebration_bash_locator,
'echo "Jobs Runs Successfully"',
depends_on=[sort_job],
job_profiles=[job_profile],
)
celebration_bash_dir = directory_for(celebration_bash_locator)
assert (celebration_bash_dir / "script.sh").exists()
_ = stop_docker_as_service(mongo4_4,
depends_on=[start_mongo, sort_job],
resource_request=saga31_request)
# Generate the Pegasus DAX file & a Submit Script
dax_file_one = write_workflow_description(tmp_path)
assert dax_file_one.exists()
submit_script_one = tmp_path / "submit.sh"
assert submit_script_one.exists()
def test_dax_with_job_on_saga_with_dict_as_params(tmp_path):
workflow_params = Parameters.from_mapping({
"workflow_name":
"Test",
"workflow_created":
"Testing",
"workflow_log_dir":
str(tmp_path / "log"),
"workflow_directory":
str(tmp_path / "working"),
"site":
"saga",
"namespace":
"test",
"partition":
"gaia",
"experiment_name":
"fred",
"home_dir":
str(tmp_path),
})
slurm_params = Parameters.from_mapping({
"partition": "gaia",
"num_cpus": 1,
"num_gpus": 0,
"memory": "4G"
})
multiply_input_file = tmp_path / "raw_nums.txt"
random = Random()
random.seed(0)
nums = immutableset(int(random.random() * 100) for _ in range(25))
multiply_output_file = tmp_path / "multiplied_nums.txt"
sorted_output_file = tmp_path / "sorted_nums.txt"
add_output_file = tmp_path / "add_nums.txt"
with multiply_input_file.open("w") as mult_file:
mult_file.writelines(f"{num}\n" for num in nums)
multiply_params = {
"input_file": multiply_input_file,
"output_file": multiply_output_file,
"x": 4,
}
sort_params = {
"input_file": multiply_output_file,
"output_file": sorted_output_file
}
add_args = f"{sorted_output_file} {add_output_file} --y 10"
job_profile = PegasusProfile(namespace="pegasus",
key="transfer.bypass.input.staging",
value="True")
resources = SlurmResourceRequest.from_parameters(slurm_params)
initialize_vista_pegasus_wrapper(workflow_params)
multiply_job_name = Locator(_parse_parts("jobs/multiply"))
multiply_artifact = ValueArtifact(
multiply_output_file,
depends_on=run_python_on_parameters(
multiply_job_name,
multiply_by_x_main,
multiply_params,
depends_on=[],
job_profiles=[job_profile],
),
locator=Locator("multiply"),
)
multiple_dir = directory_for(multiply_job_name)
assert (multiple_dir / "___run.sh").exists()
assert (multiple_dir / "____params.params").exists()
sort_job_name = Locator(_parse_parts("jobs/sort"))
sort_dir = directory_for(sort_job_name)
sort_artifact = run_python_on_parameters(
sort_job_name,
sort_nums_main,
sort_params,
depends_on=[multiply_artifact],
resource_request=resources,
category="add",
)
assert (sort_dir / "___run.sh").exists()
assert (sort_dir / "____params.params").exists()
add_job_name = Locator(_parse_parts("jobs/add"))
add_dir = directory_for(add_job_name)
run_python_on_args(add_job_name,
"add_job_main.py",
add_args,
depends_on=[sort_artifact])
assert (add_dir / "___run.sh").exists()
dax_file_one = write_workflow_description(tmp_path)
dax_file_two = write_workflow_description()
assert dax_file_one.exists()
assert dax_file_two.exists()
submit_script_one = tmp_path / "submit_script_one.sh"
submit_script_two = tmp_path / "submit_script_two.sh"
build_submit_script(submit_script_one, str(dax_file_one),
experiment_directory())
build_submit_script(submit_script_two, str(dax_file_two),
experiment_directory())
assert submit_script_one.exists()
assert submit_script_two.exists()
site_catalog = workflow_params.existing_directory(
"workflow_directory") / "sites.yml"
assert site_catalog.exists()
replica_catalog = (
workflow_params.existing_directory("workflow_directory") /
"replicas.yml")
assert replica_catalog.exists()
transformations_catalog = (
workflow_params.existing_directory("workflow_directory") /
"transformations.yml")
assert transformations_catalog.exists()
properties_file = (
workflow_params.existing_directory("workflow_directory") /
"pegasus.properties")
assert properties_file.exists()
def test_dax_with_job_in_container(tmp_path):
workflow_params = Parameters.from_mapping({
"workflow_name":
"Test",
"workflow_created":
"Testing",
"workflow_log_dir":
str(tmp_path / "log"),
"workflow_directory":
str(tmp_path / "working"),
"site":
"saga",
"namespace":
"test",
"partition":
"gaia",
"experiment_name":
"fred",
"home_dir":
str(tmp_path),
})
slurm_params = Parameters.from_mapping({
"partition": "gaia",
"num_cpus": 1,
"num_gpus": 0,
"memory": "4G"
})
multiply_input_file = tmp_path / "raw_nums.txt"
random = Random()
random.seed(0)
nums = immutableset(int(random.random() * 100) for _ in range(25))
multiply_output_file = tmp_path / "multiplied_nums.txt"
sorted_output_file = tmp_path / "sorted_nums.txt"
with multiply_input_file.open("w") as mult_file:
mult_file.writelines(f"{num}\n" for num in nums)
multiply_params = Parameters.from_mapping({
"input_file": multiply_input_file,
"output_file": multiply_output_file,
"x": 4
})
sort_params = Parameters.from_mapping({
"input_file": multiply_output_file,
"output_file": sorted_output_file
})
resources = SlurmResourceRequest.from_parameters(slurm_params)
initialize_vista_pegasus_wrapper(workflow_params)
# Add Container
example_docker = add_container("example_container", "docker",
tmp_path / "docker.img")
with pytest.raises(ValueError):
_ = add_container("fake_container", "invalid",
tmp_path / "invalid_docker.img")
multiply_job_name = Locator(_parse_parts("jobs/multiply"))
multiply_artifact = ValueArtifact(
multiply_output_file,
depends_on=run_python_on_parameters(
multiply_job_name,
multiply_by_x_main,
multiply_params,
depends_on=[],
container=example_docker,
),
locator=Locator("multiply"),
)
multiple_dir = directory_for(multiply_job_name)
assert (multiple_dir / "___run.sh").exists()
assert (multiple_dir / "____params.params").exists()
sort_job_name = Locator(_parse_parts("jobs/sort"))
sort_dir = directory_for(sort_job_name)
run_python_on_parameters(
sort_job_name,
sort_nums_main,
sort_params,
depends_on=[multiply_artifact],
resource_request=resources,
container=example_docker,
)
assert (sort_dir / "___run.sh").exists()
assert (sort_dir / "____params.params").exists()
dax_file_one = write_workflow_description()
assert dax_file_one.exists()
site_catalog = workflow_params.existing_directory(
"workflow_directory") / "sites.yml"
assert site_catalog.exists()
replica_catalog = (
workflow_params.existing_directory("workflow_directory") /
"replicas.yml")
assert replica_catalog.exists()
transformations_catalog = (
workflow_params.existing_directory("workflow_directory") /
"transformations.yml")
assert transformations_catalog.exists()
properties_file = (
workflow_params.existing_directory("workflow_directory") /
"pegasus.properties")
assert properties_file.exists()
def test_dax_with_checkpointed_jobs_on_saga(tmp_path):
workflow_params = Parameters.from_mapping({
"workflow_name":
"Test",
"workflow_created":
"Testing",
"workflow_log_dir":
str(tmp_path / "log"),
"workflow_directory":
str(tmp_path / "working"),
"site":
"saga",
"namespace":
"test",
"partition":
"gaia",
"home_dir":
str(tmp_path),
})
slurm_params = Parameters.from_mapping({
"partition": "gaia",
"num_cpus": 1,
"num_gpus": 0,
"memory": "4G"
})
resources = SlurmResourceRequest.from_parameters(slurm_params)
initialize_vista_pegasus_wrapper(workflow_params)
multiply_job_name = Locator(_parse_parts("jobs/multiply"))
multiply_output_file = tmp_path / "multiplied_nums.txt"
multiply_input_file = tmp_path / "raw_nums.txt"
multiply_params = Parameters.from_mapping({
"input_file": multiply_input_file,
"output_file": multiply_output_file,
"x": 4
})
multiple_dir = directory_for(multiply_job_name)
# Create checkpointed file so that when trying to create the job again,
# Pegasus just adds the file to the Replica Catalog
checkpointed_multiply_file = multiple_dir / "___ckpt"
checkpointed_multiply_file.touch()
multiply_output_file.touch()
assert checkpointed_multiply_file.exists()
assert multiply_output_file.exists()
multiply_artifact = ValueArtifact(
multiply_output_file,
depends_on=run_python_on_parameters(multiply_job_name,
multiply_by_x_main,
multiply_params,
depends_on=[]),
locator=Locator("multiply"),
)
sort_job_name = Locator(_parse_parts("jobs/sort"))
sorted_output_file = tmp_path / "sorted_nums.txt"
sort_params = Parameters.from_mapping({
"input_file": multiply_output_file,
"output_file": sorted_output_file
})
run_python_on_parameters(
sort_job_name,
sort_nums_main,
sort_params,
depends_on=[multiply_artifact],
resource_request=resources,
)
write_workflow_description()
site_catalog = workflow_params.existing_directory(
"workflow_directory") / "sites.yml"
assert site_catalog.exists()
replica_catalog = (
workflow_params.existing_directory("workflow_directory") /
"replicas.yml")
assert replica_catalog.exists()
transformations_catalog = (
workflow_params.existing_directory("workflow_directory") /
"transformations.yml")
assert transformations_catalog.exists()
properties_file = (
workflow_params.existing_directory("workflow_directory") /
"pegasus.properties")
assert properties_file.exists()
# Make sure the Replica Catalog is not empty
assert replica_catalog.stat().st_size > 0
def integrated_experiment_entry_point(params: Parameters) -> None:
initialize_vista_pegasus_wrapper(params)
baseline_parameters = params.namespace("integrated_learners_experiment")
pursuit_resource_request_params = params.namespace(
"pursuit_resource_request")
# This code is commented out but may be used in the near future to add language ablation
# Capabilities to this curriculum.
# get the minimum and maximum accuracy of the language with the situation
# min_language_accuracy = params.floating_point("min_language_accuracy", default=0.1)
# max_language_accuracy = params.floating_point("max_language_accuracy", default=0.5)
# num_language_accuracy_increment = params.integer(
# "num_language_accuracy_increment", default=5
# )
# values_for_accuracy = np.linspace(
# min_language_accuracy, max_language_accuracy, num_language_accuracy_increment
# )
# Get if attributes or relations should be included
include_attributes = params.boolean("include_attributes", default=True)
include_relations = params.boolean("include_relations", default=True)
limit_jobs_for_category(
"pursuit_job_limit",
params.integer("num_pursuit_learners_active", default=8))
curriculum_repository_path = params.creatable_directory(
"curriculum_repository_path")
# Job to build desired curriculum(s) which our learners use
curriculum_dependencies = immutableset((
CURRICULUM_NAME_FORMAT.format(
noise=add_noise,
shuffled=shuffle,
relations=include_relations,
attributes=include_attributes,
),
run_python_on_parameters(
Locator(
CURRICULUM_NAME_FORMAT.format(
noise=add_noise,
shuffled=shuffle,
relations=include_relations,
attributes=include_attributes,
).split("-")),
generate_curriculum_script,
baseline_parameters.unify({
"train_curriculum":
Parameters.from_mapping(CURRICULUM_PARAMS).unify(
{
"add_noise": add_noise,
"shuffled": shuffle,
"include_attributes": include_attributes,
"include_relations": include_relations,
}).as_mapping()
}).unify(FIXED_PARAMETERS).unify(
{"curriculum_repository_path": curriculum_repository_path}),
depends_on=[],
),
Parameters.from_mapping(CURRICULUM_PARAMS).unify(
{
"add_noise": add_noise,
"shuffled": shuffle,
"include_attributes": include_attributes,
"include_relations": include_relations,
}),
) for add_noise in (True, False) for shuffle in (True, False))
# jobs to build experiment
for (curriculum_str, curriculum_dep,
curr_params) in curriculum_dependencies:
object_learner_type = params.string(
"object_learner.learner_type",
valid_options=["pursuit", "subset", "pbv"],
default="pursuit",
)
attribute_learner_type = params.string(
"attribute_learner.learner__type",
valid_options=["none", "pursuit", "subset"],
default="pursuit",
)
relation_learner_type = params.string(
"relation_learner.learner_type",
valid_options=["none", "pursuit", "subset"],
default="pursuit",
)
experiment_name_string = EXPERIMENT_NAME_FORMAT.format(
curriculum_name=curriculum_str.replace("-", "+"),
object_learner=object_learner_type,
attribute_learner=attribute_learner_type,
relation_learner=relation_learner_type,
)
experiment_name = Locator(experiment_name_string.split("-"))
# Note that the input parameters should include the root params and
# anything else we want.
experiment_params = baseline_parameters.unify(FIXED_PARAMETERS).unify({
"experiment":
experiment_name_string,
"experiment_group_dir":
directory_for(experiment_name),
"hypothesis_log_dir":
directory_for(experiment_name) / "hypotheses",
"learner_logging_path":
directory_for(experiment_name),
"log_learner_state":
True,
"resume_from_latest_logged_state":
True,
"load_from_curriculum_repository":
curriculum_repository_path,
"train_curriculum":
curr_params,
})
run_python_on_parameters(
experiment_name,
log_experiment_script,
experiment_params,
depends_on=[curriculum_dep],
resource_request=SlurmResourceRequest.from_parameters(
pursuit_resource_request_params) if "pursuit" in [
object_learner_type, attribute_learner_type,
relation_learner_type
] else None,
category="pursuit" if "pursuit" in [
object_learner_type, attribute_learner_type,
relation_learner_type
] else "subset",
use_pypy=True,
)
write_workflow_description()
def object_language_ablation_runner_entry_point(params: Parameters) -> None:
"""This function creates all possible object language ablation param files within a given range"""
initialize_vista_pegasus_wrapper(params)
baseline_parameters = params.namespace("object_language_ablation")
pursuit_resource_request_params = params.namespace(
"pursuit_resource_request")
# get the minimum and maximum number of objects in a scene
min_num_objects = params.integer("min_num_objects", default=1)
max_num_objects = params.integer("max_num_objects", default=7)
# get the minimum and maximum accuracy of the language with the situation
min_language_accuracy = params.floating_point("min_language_accuracy",
default=0.1)
max_language_accuracy = params.floating_point("max_language_accuracy",
default=0.5)
num_language_accuracy_increment = params.integer(
"num_language_accuracy_increment", default=5)
values_for_accuracy = np.linspace(min_language_accuracy,
max_language_accuracy,
num_language_accuracy_increment)
limit_jobs_for_category(
"pursuit", params.integer("num_pursuit_learners_active", default=8))
for num_objects in range(min_num_objects, max_num_objects + 1):
for language_accuracy in values_for_accuracy:
for learner_type in LEARNER_VALUES_TO_PARAMS:
for params_str, learner_params in LEARNER_VALUES_TO_PARAMS[
learner_type]:
experiment_name_string = EXPERIMENT_NAME_FORMAT.format(
num_objects=num_objects,
language_accuracy=language_accuracy,
learner_type=learner_type,
learner_params=params_str,
)
experiment_name = Locator(
experiment_name_string.split("-"))
# Note that the input parameters should include the root params and
# anything else we want.
experiment_params = baseline_parameters.unify(
FIXED_PARAMETERS
).unify({
"experiment":
experiment_name_string,
"experiment_group_dir":
directory_for(experiment_name),
"hypothesis_log_dir":
directory_for(experiment_name) / "hypotheses",
"learner_logging_path":
directory_for(experiment_name),
"log_learner_state":
True,
"resume_from_latest_logged_state":
True,
"train_curriculum": {
"accurate_language_percentage":
float(language_accuracy)
},
"object_learner_type":
learner_type,
"object_learner":
learner_params,
# We subtract one because the target object is a given
"num_noise_objects":
num_objects - 1,
})
run_python_on_parameters(
experiment_name,
log_experiment_script,
experiment_params,
depends_on=[],
resource_request=SlurmResourceRequest.from_parameters(
pursuit_resource_request_params)
if learner_type == "pursuit" else None,
category=learner_type,
)
write_workflow_description()
def explicit_train_dev_test_split(
corpus: KeyValueStore,
*,
train_ids: ValueArtifact[Path],
dev_ids: ValueArtifact[Path],
test_ids: ValueArtifact[Path],
output_locator: Locator,
exhaustive: bool = True,
downsample_to: Optional[int] = None,
) -> DataSplit:
"""
Explicit implementation for handling a train/dev/test split over a `KeyValueStore`
The split is done by a list of keys handed explicitly to the user.
If *exhaustive* is True then an exception will be thrown if a document does not get assigned to
one of the three sets. This is to help prevent accidental omissions in the key lists
*downsample_to* is an optional integer to reduce to the size of the key_value split
for quicker debugging. See `vistautils.scripts.downsample_key_value_store` as the function
which is
See `DataSplit` for the output description.
"""
train_locator = output_locator / "train"
dev_locator = output_locator / "dev"
test_locator = output_locator / "test"
train_zip = directory_for(train_locator) / "train.zip"
dev_zip = directory_for(dev_locator) / "dev.zip"
test_zip = directory_for(test_locator) / "test.zip"
split_job = run_python_on_parameters(
output_locator,
split_entry_point,
parameters={
"input": corpus.input_parameters(),
"explicit_split": {
"train": {
"keys_file": train_ids.value,
"output_file": train_zip
},
"dev": {
"keys_file": dev_ids.value,
"output_file": dev_zip
},
"test": {
"keys_file": test_ids.value,
"output_file": test_zip
},
"must_be_exhaustive": exhaustive,
},
},
depends_on=[corpus],
)
deps = [
corpus.depends_on,
split_job,
train_ids.depends_on,
dev_ids.depends_on,
test_ids.depends_on,
]
train_store = ZipKeyValueStore(train_zip,
locator=train_locator,
depends_on=deps)
dev_store = ZipKeyValueStore(dev_zip, locator=dev_locator, depends_on=deps)
test_store = ZipKeyValueStore(test_zip,
locator=test_locator,
depends_on=deps)
if downsample_to is None:
return DataSplit(train=train_store, dev=dev_store, test=test_store)
else:
return DataSplit(
train=downsample(train_store, limit=downsample_to),
dev=downsample(dev_store, limit=downsample_to),
test=downsample(test_store, limit=downsample_to),
)
