# Copyright 2021 Amazon.com, Inc. or its affiliates. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License").
# You may not use this file except in compliance with the License.
# A copy of the License is located at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# or in the "license" file accompanying this file. This file is distributed
# on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either
# express or implied. See the License for the specific language governing
# permissions and limitations under the License.
from pathlib import Path
from typing import Optional
from syne_tune.constants import ST_METADATA_FILENAME, ST_RESULTS_DATAFRAME_FILENAME
from syne_tune.optimizer.schedulers.random_seeds import (
generate_random_seed,
RANDOM_SEED_UPPER_BOUND,
)
from syne_tune.util import s3_experiment_path, experiment_path
[docs]
def filter_none(a: dict) -> dict:
return {k: v for k, v in a.items() if v is not None}
[docs]
def sync_from_s3_command(experiment_name: str, s3_bucket: Optional[str] = None) -> str:
s3_source_path = s3_experiment_path(
s3_bucket=s3_bucket, experiment_name=experiment_name
)
target_path = str(experiment_path() / experiment_name)
return (
f'aws s3 sync {s3_source_path} {target_path} --exclude "*" '
f'--include "*{ST_METADATA_FILENAME}" '
f'--include "*{ST_RESULTS_DATAFRAME_FILENAME}"'
)
[docs]
def message_sync_from_s3(experiment_tag: str) -> str:
return (
"Launched all requested experiments. Once everything is done, use this "
"command to sync result files from S3:\n"
+ sync_from_s3_command(experiment_name=experiment_tag)
)
[docs]
def combine_requirements_txt(synetune_requirements_file: Path, script: Path) -> Path:
script_requirements_file = script.parent / "requirements.txt"
target_requirements_file = synetune_requirements_file.parent / "requirements.txt"
f1_exists = synetune_requirements_file.exists()
f2_exists = script_requirements_file.exists()
equal_fnames = synetune_requirements_file == target_requirements_file
assert not (f1_exists and f2_exists and equal_fnames), (
f"{synetune_requirements_file} would be overwritten by combined "
"requirements.txt file. Please use a different filename"
)
if f1_exists:
if not f2_exists and equal_fnames:
return target_requirements_file # Nothing to do
target_content = synetune_requirements_file.read_text()
else:
target_content = ""
if f2_exists:
extra_content = script_requirements_file.read_text()
target_content += "\n" + extra_content
if target_content:
target_requirements_file.write_text(target_content)
return target_requirements_file
[docs]
def ERR_MSG(fname: str) -> str:
return (
f"I'll create a safe '{fname}' for you:\n"
" syne-tune[extra]\n"
" tqdm\n"
"You can optimize start times by using restricted dependencies, e.g.:\n"
" syne-tune[gpsearchers,kde]\n"
" tqdm\n"
"If your launcher script requires additional dependencies, add them "
"here as well."
)
[docs]
def find_or_create_requirements_txt(
entry_point: Path, requirements_fname: Optional[str] = None
) -> Path:
if requirements_fname is None:
requirements_fname = "requirements.txt"
files = list(entry_point.parent.glob("requirements*.txt"))
if len(files) == 0:
print(
f"Could not find {entry_point.parent}/requirements*.txt\n"
+ ERR_MSG(requirements_fname)
)
fname = entry_point.parent / requirements_fname
fname.write_text("syne-tune[extra]\ntqdm\n")
elif len(files) > 1:
# Filter out the typical target fname
target_fname = entry_point.parent / "requirements.txt"
files = [fname for fname in files if fname != target_fname]
fname = files[0]
assert (
len(files) == 1
), f"Found more than one {entry_point.parent}/requirements*.txt:\n" + str(
[str(path) for path in files]
)
else:
fname = files[0]
return fname
[docs]
def get_master_random_seed(random_seed: Optional[int]) -> int:
if random_seed is None:
random_seed = generate_random_seed()
print(f"Master random_seed = {random_seed}")
return random_seed
[docs]
def effective_random_seed(master_random_seed: int, seed: int) -> int:
return (master_random_seed + seed) % RANDOM_SEED_UPPER_BOUND