def _build_train_spec(self):
train_hooks = [
hooks.LogParametersCountHook(),
hooks.CountersHook(
every_n_steps=self._estimator.config.save_summary_steps,
output_dir=self._estimator.model_dir)]
train_spec = tf.estimator.TrainSpec(
input_fn=self._model.input_fn(
tf.estimator.ModeKeys.TRAIN,
self._config["train"]["batch_size"],
self._config["data"],
self._config["data"]["train_features_file"],
labels_file=self._config["data"]["train_labels_file"],
batch_type=self._config["train"].get("batch_type", "examples"),
batch_multiplier=self._num_devices,
bucket_width=self._config["train"].get("bucket_width", 5),
single_pass=self._config["train"].get("single_pass", False),
num_threads=self._config["train"].get("num_threads"),
sample_buffer_size=self._config["train"].get("sample_buffer_size", 500000),
prefetch_buffer_size=self._config["train"].get("prefetch_buffer_size"),
maximum_features_length=self._config["train"].get("maximum_features_length"),
maximum_labels_length=self._config["train"].get("maximum_labels_length")),
max_steps=self._config["train"].get("train_steps"),
hooks=train_hooks)
return train_spec
def _build_train_spec(self, checkpoint_path):
train_hooks = [
hooks.LogParametersCountHook()]
if checkpoint_path is not None:
train_hooks.append(hooks.LoadWeightsFromCheckpointHook(checkpoint_path))
if self._hvd is not None:
train_hooks.append(self._hvd.BroadcastGlobalVariablesHook(0))
train_steps = self._config["train"].get("train_steps")
if train_steps is not None and self._hvd is not None:
train_steps //= self._hvd.size()
train_spec = tf.estimator.TrainSpec(
input_fn=estimator_util.make_input_fn(
self._model,
tf.estimator.ModeKeys.TRAIN,
self._config["train"]["batch_size"],
features_file=self._config["data"]["train_features_file"],
labels_file=self._config["data"].get("train_labels_file"),
batch_type=self._config["train"]["batch_type"],
batch_multiplier=self._num_devices,
bucket_width=self._config["train"]["bucket_width"],
maximum_features_length=self._config["train"].get("maximum_features_length"),
maximum_labels_length=self._config["train"].get("maximum_labels_length"),
shuffle_buffer_size=self._config["train"]["sample_buffer_size"],
single_pass=self._config["train"].get("single_pass", False),
num_shards=self._hvd.size() if self._hvd is not None else 1,
shard_index=self._hvd.rank() if self._hvd is not None else 0,
num_threads=self._config["train"].get("num_threads"),
prefetch_buffer_size=self._config["train"].get("prefetch_buffer_size"),
return_dataset=False),
max_steps=train_steps,
hooks=train_hooks)
return train_spec
def _build_train_spec(self, checkpoint_path):
train_hooks = [
hooks.LogParametersCountHook()]
if checkpoint_path is not None:
train_hooks.append(hooks.LoadWeightsFromCheckpointHook(checkpoint_path))
train_spec = tf.estimator.TrainSpec(
input_fn=self._model.input_fn(
tf.estimator.ModeKeys.TRAIN,
self._config["train"]["batch_size"],
self._config["data"],
self._config["data"]["train_features_file"],
labels_file=self._config["data"]["train_labels_file"],
batch_type=self._config["train"]["batch_type"],
batch_multiplier=self._num_devices,
bucket_width=self._config["train"]["bucket_width"],
single_pass=self._config["train"].get("single_pass", False),
num_threads=self._config["train"].get("num_threads"),
sample_buffer_size=self._config["train"]["sample_buffer_size"],
prefetch_buffer_size=self._config["train"].get("prefetch_buffer_size"),
maximum_features_length=self._config["train"].get("maximum_features_length"),
maximum_labels_length=self._config["train"].get("maximum_labels_length")),
max_steps=self._config["train"].get("train_steps"),
hooks=train_hooks)
return train_spec
def _build_train_spec(self, checkpoint_path):
train_hooks = [hooks.LogParametersCountHook()]
#if checkpoint_path is not None:
# train_hooks.append(hooks.LoadWeightsFromCheckpointHook(checkpoint_path))
# NEW: loads params based on config.yml ["load_weights"] - see config*.yml
if checkpoint_path is not None and "load_weights" in self._config:
not_restore = []
loadw = self._config["load_weights"]
if loadw is not None:
if not loadw.get("src_embs"):
not_restore.append("encoder/w_embs")
if not loadw.get("tgt_embs"):
not_restore.append("decoder/w_embs")
if not loadw.get("projection"):
not_restore.append("decoder/dense")
if not loadw.get("shared_embs"):
not_restore.append("shared_embeddings/w_embs")
if not loadw.get("encoder"):
not_restore.append("encoder")
if not loadw.get("decoder"):
not_restore.append("decoder")
if not loadw.get(
"optim"
): # if true, also avoids global_step and word_per_sec
not_restore.append("optim")
if not loadw.get("global_step"):
not_restore.append("global_step")
if not loadw.get("words_per_sec"):
not_restore.append("words_per_sec")
tf.logging.info("NOT RESTORING: %s",
json.dumps(not_restore, indent=2, sort_keys=True))
train_hooks.append(
hooks.LoadWeightsFromCheckpointHook(checkpoint_path,
not_restore))
train_spec = tf.estimator.TrainSpec(
input_fn=self._model.input_fn(
tf.estimator.ModeKeys.TRAIN,
self._config["train"]["batch_size"],
self._config["data"],
self._config["data"]["train_features_file"],
labels_file=self._config["data"]["train_labels_file"],
batch_type=self._config["train"]["batch_type"],
batch_multiplier=self._num_devices,
bucket_width=self._config["train"]["bucket_width"],
single_pass=self._config["train"].get("single_pass", False),
num_threads=self._config["train"].get("num_threads"),
sample_buffer_size=self._config["train"]["sample_buffer_size"],
prefetch_buffer_size=self._config["train"].get(
"prefetch_buffer_size"),
maximum_features_length=self._config["train"].get(
"maximum_features_length"),
maximum_labels_length=self._config["train"].get(
"maximum_labels_length")),
max_steps=self._config["train"].get("train_steps"),
hooks=train_hooks)
return train_spec
def _build_train_spec(self):
train_hooks = [
hooks.LogParametersCountHook(),
hooks.CountersHook(
every_n_steps=self._estimator.config.save_summary_steps,
output_dir=self._estimator.model_dir)
]
default_sample_buffer_size = 1000000
if "sample_buffer_size" not in self._config["train"]:
tf.logging.warn(
"You did not set sample_buffer_size. By default, the "
"training dataset is shuffled by chunk of %d examples. "
"If your dataset is larger than this value and eval_delay "
"is shorter than the training time of one epoch, a section "
"of the dataset will be discarded. Consider setting "
"sample_buffer_size to the size of your dataset." %
default_sample_buffer_size)
train_spec = tf.estimator.TrainSpec(
input_fn=self._model.input_fn(
tf.estimator.ModeKeys.TRAIN,
self._config["train"]["batch_size"],
self._config["data"],
self._config["data"]["train_features_file"],
labels_file=self._config["data"]["train_labels_file"],
batch_type=self._config["train"].get("batch_type", "examples"),
batch_multiplier=self._num_devices,
bucket_width=self._config["train"].get("bucket_width", 5),
single_pass=self._config["train"].get("single_pass", False),
num_threads=self._config["train"].get("num_threads"),
sample_buffer_size=self._config["train"].get(
"sample_buffer_size", default_sample_buffer_size),
maximum_features_length=self._config["train"].get(
"maximum_features_length"),
maximum_labels_length=self._config["train"].get(
"maximum_labels_length")),
max_steps=self._config["train"].get("train_steps"),
hooks=train_hooks)
return train_spec
def train(estimator, model, config):
"""Runs training.
Args:
estimator: A `tf.estimator.Estimator`.
model: A `opennmt.models.Model`.
config: The configuration.
"""
batch_size = config["train"]["batch_size"]
prefetch_buffer_size = config["train"].get("prefetch_buffer_size",
batch_size * 1000)
num_parallel_process_calls = config["train"].get(
"num_parallel_process_calls", multiprocessing.cpu_count())
train_hooks = [
hooks.LogParametersCountHook(),
hooks.CountersHook(every_n_steps=estimator.config.save_summary_steps,
output_dir=estimator.model_dir)
]
eval_hooks = []
if config["train"].get("save_eval_predictions", False):
save_path = os.path.join(estimator.model_dir, "eval")
if not os.path.isdir(save_path):
os.makedirs(save_path)
eval_hooks.append(
hooks.SaveEvaluationPredictionHook(
model,
os.path.join(save_path, "predictions.txt"),
post_evaluation_fn=external_evaluation_fn(
config["train"].get("external_evaluators"),
config["data"]["eval_labels_file"],
output_dir=estimator.model_dir)))
elif config["train"].get("external_evaluators") is not None:
tf.logging.warning(
"External evaluators only work when save_eval_predictions is enabled."
)
train_spec = tf.estimator.TrainSpec(input_fn=model.input_fn(
tf.estimator.ModeKeys.TRAIN,
batch_size,
prefetch_buffer_size,
num_parallel_process_calls,
config["data"],
config["data"]["train_features_file"],
labels_file=config["data"]["train_labels_file"],
num_buckets=config["train"].get("num_buckets", 5),
sample_buffer_size=config["train"].get("sample_buffer_size", 1000000),
maximum_features_length=config["train"].get("maximum_features_length"),
maximum_labels_length=config["train"].get("maximum_labels_length")),
max_steps=config["train"].get(
"train_steps"),
hooks=train_hooks)
eval_spec = tf.estimator.EvalSpec(
input_fn=model.input_fn(
tf.estimator.ModeKeys.EVAL,
batch_size,
prefetch_buffer_size,
num_parallel_process_calls,
config["data"],
config["data"]["eval_features_file"],
labels_file=config["data"]["eval_labels_file"]),
steps=None,
hooks=eval_hooks,
exporters=tf.estimator.LatestExporter(
"latest", model.serving_input_fn(config["data"])),
throttle_secs=config["train"].get("eval_delay", 18000))
tf.estimator.train_and_evaluate(estimator, train_spec, eval_spec)
def train(estimator, model, config):
"""Runs training.
Args:
estimator: A `tf.estimator.Estimator`.
model: A `opennmt.models.Model`.
config: The configuration.
"""
if "eval" not in config:
config["eval"] = {}
train_hooks = [
hooks.LogParametersCountHook(),
hooks.CountersHook(every_n_steps=estimator.config.save_summary_steps,
output_dir=estimator.model_dir)
]
eval_hooks = []
if (config["eval"].get("save_eval_predictions", False)
or config["eval"].get("external_evaluators") is not None):
save_path = os.path.join(estimator.model_dir, "eval")
if not os.path.isdir(save_path):
os.makedirs(save_path)
eval_hooks.append(
hooks.SaveEvaluationPredictionHook(
model,
os.path.join(save_path, "predictions.txt"),
post_evaluation_fn=external_evaluation_fn(
config["eval"].get("external_evaluators"),
config["data"]["eval_labels_file"],
output_dir=estimator.model_dir)))
default_sample_buffer_size = 1000000
if "sample_buffer_size" not in config["train"]:
tf.logging.warn(
"You did not set sample_buffer_size. By default, the "
"training dataset is shuffled by chunk of %d examples. "
"If your dataset is larger than this value and eval_delay "
"is shorter than the training time of one epoch, a section "
"of the dataset will be discarded. Consider setting "
"sample_buffer_size to the size of your dataset." %
default_sample_buffer_size)
train_batch_size = config["train"]["batch_size"]
train_batch_type = config["train"].get("batch_type", "examples")
train_prefetch_buffer_size = config["train"].get(
"prefetch_buffer_size",
train_batch_size * (1000 if train_batch_type == "examples" else 50))
train_num_parallel_process_calls = config["train"].get(
"num_parallel_process_calls", multiprocessing.cpu_count())
train_spec = tf.estimator.TrainSpec(input_fn=model.input_fn(
tf.estimator.ModeKeys.TRAIN,
train_batch_size,
train_prefetch_buffer_size,
train_num_parallel_process_calls,
config["data"],
config["data"]["train_features_file"],
labels_file=config["data"]["train_labels_file"],
batch_type=train_batch_type,
bucket_width=config["train"].get("bucket_width", 5),
sample_buffer_size=config["train"].get("sample_buffer_size",
default_sample_buffer_size),
maximum_features_length=config["train"].get("maximum_features_length"),
maximum_labels_length=config["train"].get("maximum_labels_length")),
max_steps=config["train"].get(
"train_steps"),
hooks=train_hooks)
eval_batch_size = config["eval"].get(
"batch_size",
train_batch_size if train_batch_type == "examples" else 30)
eval_prefetch_buffer_size = config["eval"].get("prefetch_buffer_size",
eval_batch_size * 10)
eval_num_parallel_process_calls = config["eval"].get(
"num_parallel_process_calls", train_num_parallel_process_calls)
eval_spec = tf.estimator.EvalSpec(
input_fn=model.input_fn(
tf.estimator.ModeKeys.EVAL,
eval_batch_size,
eval_prefetch_buffer_size,
eval_num_parallel_process_calls,
config["data"],
config["data"]["eval_features_file"],
labels_file=config["data"]["eval_labels_file"]),
steps=None,
hooks=eval_hooks,
exporters=tf.estimator.LatestExporter(
"latest", model.serving_input_fn(config["data"])),
throttle_secs=config["eval"].get("eval_delay", 18000))
tf.estimator.train_and_evaluate(estimator, train_spec, eval_spec)
def _build_train_spec(self, checkpoint_path):
train_hooks = [hooks.LogParametersCountHook()]
#if checkpoint_path is not None:
# train_hooks.append(hooks.LoadWeightsFromCheckpointHook(checkpoint_path))
# TODO: pass what not/to load based on the config "load_weights" boolean info. MOVE OPTS TO LOAD TO UTIL FUNCTION
if checkpoint_path is not None and "load_weights" in self._config:
#not_restore = ['encoder', 'decoder', 'shared_embeddings', 'optim', 'global_step', 'word_per_sec', 'output_layer']
#not_restore = ['optim', 'global_step', 'word_per_sec'] #, 'encoder'
# FOR RUNN THE word_per_sec DOEN'T EXIST THROWS ERROR - HOW TO HANDLE? --checkpoint LOADING? if.startswith("word_per_sec") WAY IS THE ANSE
not_restore = [] # make sure its empty at first call
loadw = self._config[
"load_weights"] # search in the var_list with [0] for optim..., [1] for enc/dec and [1:2] for embs, projection
if loadw is not None:
if not loadw.get("src_embs"):
not_restore.append("encoder/w_embs")
if not loadw.get("tgt_embs"):
not_restore.append("decoder/w_embs")
if not loadw.get("projection"):
not_restore.append("decoder/dense")
if not loadw.get("shared_embs"):
not_restore.append("shared_embeddings/w_embs")
if not loadw.get("encoder"):
not_restore.append("encoder")
if not loadw.get("decoder"):
not_restore.append("decoder")
if not loadw.get(
"optim"): #if avoided the the next two are avoided
not_restore.append("optim")
if not loadw.get("global_step"):
not_restore.append("global_step")
if not loadw.get("words_per_sec"):
not_restore.append("words_per_sec")
tf.logging.info("NOT RESTORING SUB-NETWORKS: %s",
json.dumps(not_restore, indent=2, sort_keys=True))
train_hooks.append(
hooks.LoadWeightsFromCheckpointHook(
checkpoint_path,
not_restore)) #self._config["load_partial_weights"]))
train_spec = tf.estimator.TrainSpec(
input_fn=self._model.input_fn(
tf.estimator.ModeKeys.TRAIN,
self._config["train"]["batch_size"],
self._config["data"],
self._config["data"]["train_features_file"],
labels_file=self._config["data"]["train_labels_file"],
batch_type=self._config["train"]["batch_type"],
batch_multiplier=self._num_devices,
bucket_width=self._config["train"]["bucket_width"],
single_pass=self._config["train"].get("single_pass", False),
num_threads=self._config["train"].get("num_threads"),
sample_buffer_size=self._config["train"]["sample_buffer_size"],
prefetch_buffer_size=self._config["train"].get(
"prefetch_buffer_size"),
maximum_features_length=self._config["train"].get(
"maximum_features_length"),
maximum_labels_length=self._config["train"].get(
"maximum_labels_length")),
max_steps=self._config["train"].get("train_steps"),
hooks=train_hooks)
return train_spec
