聊聊python的gin库的介绍和使用

pip install gin-config

@gin.configurable
def my_network(images, num_outputs, num_layers=3, weight_decay=1e-4):
  ...

@gin.configurable('supernet', denylist=['images'])
def my_network(images, num_outputs, num_layers=3, weight_decay=1e-4):
  ...

gin.bind_parameter('supernet.num_layers', 5)
gin.bind_parameter('supernet.weight_decay', 1e-3)

supernet.num_layers = 5
supernet.weight_decay = 1e-3

num_layers = gin.query_parameter('supernet.num_layers')
weight_decay = gin.query_parameter('supernet.weight_decay')

@gin.configurable
class DNN(object):
  def __init__(self, num_units=(1024, 1024)):
    ...
  def __call__(inputs, num_outputs):
    ...

@gin.configurable(denylist=['data'])
def train_model(network_fn, data, learning_rate, optimizer):
  ...

train_model.network_fn = @DNN()  # An instance of DNN is passed.
train_model.optimizer = @MomentumOptimizer  # The class itself is passed.
train_model.learning_rate = 0.001

DNN.num_units = (2048, 2048, 2048)
MomentumOptimizer.momentum = 0.9

from absl import flags

flags.DEFINE_multi_string(
  'gin_file', None, 'List of paths to the config files.')
flags.DEFINE_multi_string(
  'gin_param', None, 'Newline separated list of Gin parameter bindings.')

FLAGS = flags.FLAGS

gin.parse_config_files_and_bindings(FLAGS.gin_file, FLAGS.gin_param)

flags.DEFINE_multi_string(
  'gin_file', ["example.gin"], 'List of paths to the config files.')

gin.external_configurable(tf.train.MomentumOptimizer)

gin.external_configurable(tf.train.GradientDescentOptimizer)

@gin.configurable(allowlist=['generator_optimizer', 'discriminator_optimizer'])
def gan_trainer(
    generator_loss, generator_vars, generator_optimizer,
    discriminator_loss, discriminator_vars, discriminator_optimizer):
  # Construct the optimizers and minimize w.r.t. the correct variables.
  generator_train_op = generator_optimizer().minimize(
      generator_loss, generator_vars)
  discriminator_train_op = discriminator_optimizer().minimize(
      discriminator_loss, discriminator_vars)
  ...

# Won't work!
gan_trainer.generator_optimizer = @GradientDescentOptimizer
GradientDescentOptimizer.learning_rate = 0.01

gan_trainer.discriminator_optimizer = @GradientDescentOptimizer
# This binding will overwrite the previous one:
GradientDescentOptimizer.learning_rate = 0.001

# This will work! Use scoping to apply different parameter bindings.
gan_trainer.generator_optimizer = @generator/GradientDescentOptimizer
gan_trainer.discriminator_optimizer = @discriminator/GradientDescentOptimizer

generator/GradientDescentOptimizer.learning_rate = 0.01
discriminator/GradientDescentOptimizer.learning_rate = 0.001

my_network(images, gin.REQUIRED, num_layers=5, weight_decay=gin.REQUIRED)

@gin.configurable
def run_training(model_dir=gin.REQUIRED, network=gin.REQUIRED, ...):
  ...

import some.module.spec

include 'path/to/another/file.gin'

@gin.configurable
def macro(value):
  return value

num_layers/macro.value = 10
network.num_layers = @num_layers/macro()

num_layers = 10
network.num_layers = %num_layers

gin.constant('THE_ANSWER', 42)

meaning.of_life = %THE_ANSWER

gin.constant('some.modules.PI', 3.14159)

filegroup(
    name = "gin_files",
    srcs = glob(["*.gin"]),
    visibility = [":internal"],
)

data = ["//path/to/configs:gin_files",]

from absl import flags

flags.DEFINE_multi_string(
  'gin_file', None, 'List of paths to the config files.')
flags.DEFINE_multi_string(
  'gin_param', None, 'Newline separated list of Gin parameter bindings.')

FLAGS = flags.FLAGS

gin.parse_config_files_and_bindings(FLAGS.gin_file, FLAGS.gin_param)

.../run_gin_eval \
  --gin_file=$CONFIGS_PATH/cartpole_balance.gin \
  --gin_file=$CONFIGS_PATH/base_dqn.gin \
  --gin_file=$CONFIGS_PATH/eval.gin \
  --gin_param='evaluate.num_episodes_eval = 10' \
  --gin_param='evaluate.generate_videos = False' \
  --gin_param='evaluate.eval_interval_secs = 60'