fc_names = ["MFC_{}".format(*v) for v in NOC_FC_SIZES]
# A2C
experiment_title = "A2C_Atari"
variant_levels = list()
variant_levels.append(VariantLevel(game_key, games, game_names)) # Games
variant_levels.append(VariantLevel(fc_key, NOC_FC_SIZES,
fc_names)) # Smaller or larger model
variant_levels.append(VariantLevel(lr_key, A2C_LRS,
lr_names_a2c)) # Learning rates
variants, log_dirs = make_variants(*variant_levels)
run_experiments(
script=path_a2c,
affinity_code=affinity_code,
experiment_title=experiment_title,
runs_per_setting=runs_per_setting,
variants=variants,
log_dirs=log_dirs,
common_args=(default_key, ),
)
# PPO
experiment_title = "PPO_Atari"
variant_levels = list()
variant_levels.append(VariantLevel(game_key, games, game_names)) # Games
variant_levels.append(VariantLevel(fc_key, NOC_FC_SIZES,
fc_names)) # Smaller or larger model
variant_levels.append(VariantLevel(lr_key, PPO_LRS,
lr_names_ppo)) # Learning rates
variants, log_dirs = make_variants(*variant_levels)
run_experiments(
script = "rlpyt/experiments/scripts/atari/pg/train/atari_ff_a2c_gpu.py"
affinity_code = encode_affinity(n_cpu_core=16,
n_gpu=4,
hyperthread_offset=20,
n_socket=2
# cpu_per_run=2,
)
runs_per_setting = 2
experiment_title = "atari_ff_a2c_basic"
variant_levels = list()
games = ["pong", "seaquest", "qbert", "chopper_command"]
values = list(zip(games))
dir_names = ["{}".format(*v) for v in values]
keys = [("env", "game")]
variant_levels.append(VariantLevel(keys, values, dir_names))
variants, log_dirs = make_variants(*variant_levels)
default_config_key = "0"
run_experiments(
script=script,
affinity_code=affinity_code,
experiment_title=experiment_title,
runs_per_setting=runs_per_setting,
variants=variants,
log_dirs=log_dirs,
common_args=(default_config_key, ),
)
runs_per_setting = 2
experiment_title = "example_6"
variant_levels = list()
# Within a variant level, list each combination explicitly.
learning_rate = [7e-4, 1e-3]
batch_B = [16, 32]
values = list(zip(learning_rate, batch_B))
dir_names = ["example6_{}lr_{}B".format(*v) for v in values]
keys = [("algo", "learning_rate"), ("sampler", "batch_B")]
variant_levels.append(VariantLevel(keys, values, dir_names))
games = ["pong", "seaquest"]
values = list(zip(games))
dir_names = ["{}".format(*v) for v in values]
keys = [("env", "game")]
variant_levels.append(VariantLevel(keys, values, dir_names))
# Between variant levels, make all combinations.
variants, log_dirs = make_variants(*variant_levels)
run_experiments(
script="examples/example_6a.py",
affinity_code=affinity_code,
experiment_title=experiment_title,
runs_per_setting=runs_per_setting,
variants=variants,
log_dirs=log_dirs,
)
