def sample_rnn_text_classification_family_cfg(seed):
"""Samples a task config for an RNN classification task.
These configs are nested python structures that provide enough information
to create an instance of the problem.
Args:
seed: int Random seed to generate task from.
Returns:
A nested dictionary containing a configuration.
"""
rng = np.random.RandomState(seed)
cfg = {}
cfg["embed_dim"] = utils.sample_log_int(rng, 8, 128)
cfg["w_init"] = utils.sample_initializer(rng)
cfg["dataset"] = utils.sample_text_dataset(rng)
# TODO(lmetz) trim this if using characters...
cfg["vocab_size"] = utils.sample_log_int(rng, 100, 10000)
cfg["core"] = utils.sample_rnn_core(rng)
cfg["trainable_init"] = bool(rng.choice([True, False]))
cfg["loss_compute"] = rng.choice(["last", "avg", "max"])
return cfg
def sample_word_language_model_family_cfg(seed):
"""Sample a word language model config."""
rng = np.random.RandomState(seed)
cfg = {}
cfg["embed_dim"] = utils.sample_log_int(rng, 8, 128)
cfg["w_init"] = utils.sample_initializer(rng)
cfg["vocab_size"] = utils.sample_log_int(rng, 1000, 30000)
cfg["core"] = utils.sample_rnn_core(rng)
cfg["trainable_init"] = bool(rng.choice([True, False]))
cfg["dataset"] = utils.sample_word_lm_dataset(rng)
return cfg
def sample_char_rnn_language_model_family_cfg(seed):
"""Samples a character NN language modeling task."""
rng = np.random.RandomState(seed)
cfg = {}
cfg["embed_dim"] = utils.sample_log_int(rng, 8, 128)
cfg["w_init"] = utils.sample_initializer(rng)
full_vocab = utils.sample_bool(rng, 0.8)
if full_vocab:
cfg["vocab_size"] = 256
else:
# only operate on some subset of full words.
cfg["vocab_size"] = utils.sample_log_int(rng, 100, 256)
cfg["core"] = utils.sample_rnn_core(rng)
cfg["trainable_init"] = bool(rng.choice([True, False]))
cfg["dataset"] = utils.sample_char_lm_dataset(rng)
return cfg
