child_poet_master_script.py

# Copyright (c) 2020 Uber Technologies, Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License 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.

"""
Master script for starting training of CHILD Enhanced POET
"""

from argparse import ArgumentParser
import logging
import numpy as np
import mlflow as mlf
from poet_distributed.es import initialize_master_fiber
from poet_distributed.poet_algo import MultiESOptimizer
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)


def run_main(args):

    initialize_master_fiber()

    # set master_seed
    np.random.seed(args.master_seed)

    optimizer_zoo = MultiESOptimizer(args=args)

    optimizer_zoo.optimize(iterations=args.n_iterations,
                           propose_with_adam=args.propose_with_adam,
                           reset_optimizer=True,
                           checkpointing=args.checkpointing,
                           steps_before_transfer=args.steps_before_transfer)


def main():
    parser = ArgumentParser()
    parser.add_argument('log_file')
    parser.add_argument('niche_file')
    parser.add_argument('--mlflow_folder', default='mlruns')
    parser.add_argument('--save_to_dataset', default=False)
    parser.add_argument('--init', default='random')
    parser.add_argument('--learning_rate', type=float, default=0.01)
    parser.add_argument('--lr_decay', type=float, default=0.9999)
    parser.add_argument('--lr_limit', type=float, default=0.001)
    parser.add_argument('--noise_std', type=float, default=0.1)
    parser.add_argument('--noise_decay', type=float, default=0.999)
    parser.add_argument('--noise_limit', type=float, default=0.01)
    parser.add_argument('--l2_coeff', type=float, default=0.01)
    parser.add_argument('--batches_per_chunk', type=int, default=50)
    parser.add_argument('--batch_size', type=int, default=64)
    parser.add_argument('--eval_batch_size', type=int, default=1)
    parser.add_argument('--eval_batches_per_step', type=int, default=50)
    parser.add_argument('--num_workers', type=int, default=20)
    parser.add_argument('--n_iterations', type=int, default=200)
    parser.add_argument('--steps_before_transfer', type=int, default=25)
    parser.add_argument('--max_children', type=int, default=8)
    parser.add_argument('--max_admitted', type=int, default=1)
    parser.add_argument('--master_seed', type=int, default=111)
    parser.add_argument('--mc_lower', type=int, default=25)
    parser.add_argument('--mc_upper', type=int, default=340)
    parser.add_argument('--repro_threshold', type=int, default=200)
    parser.add_argument('--max_num_envs', type=int, default=100)
    parser.add_argument('--normalize_grads_by_noise_std', action='store_true', default=False)
    parser.add_argument('--propose_with_adam', action='store_true', default=False)
    parser.add_argument('--checkpointing', action='store_true', default=False)
    parser.add_argument('--adjust_interval', type=int, default=4)
    parser.add_argument('--returns_normalization', default='normal')
    parser.add_argument('--stochastic', action='store_true', default=False)
    parser.add_argument('--envs', nargs='+')
    parser.add_argument('--start_from', default=None)  # Json file to start from
    # Whether or not to run the CHILD set of features, which predict simulation outcome:
    parser.add_argument('--run_child_poet', action='store_true', default=False)
    # Whether or not to run the part of CHILD that skips simulation when score estimates have been made:
    parser.add_argument('--omit_simulation', action='store_true', default=False)
    parser.add_argument('--child_success_reward', type=float, default=0.5)
    parser.add_argument('--agent_tracker_certainty_threshold', type=float, default=0.8)

    args = parser.parse_args()
    # print(args.run_child_poet)

    logger.info(args)

    mlf.log_params(args.__dict__)

    run_main(args)


if __name__ == "__main__":
    main()