from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import numpy as np
import tensorflow as tf
import sys
sys.path.append('chatbot')
sys.path.append('utils')
from models.base_model import BaseModel
import misc_utils as utils
from models import model_helper
utils.check_tensorflow_version(version="1.3.0")
class HierarchicalModel(BaseModel):
"""
Sequence-to-sequence hierarchical model.
This class implements a multi-layer recurrent neural network as encoder,
a multi-layer recurrent neural network as a context encoder
and a multi-layer recurrent neural network decoder.
"""
def _build_encoder(self, hparams):
"""Build an encoder"""
encoder_num_layers = hparams.num_layers
encoder_num_residual_layers = hparams.num_residual_layers
from __future__ import print_function
import os
import tensorflow as tf
import sys
sys.path.append('utils')
#from utils import misc_utils as utils
#from utils import vocab_utils
import misc_utils as utils
import vocab_utils
utils.check_tensorflow_version()
def add_arguments(parser):
"""Build the arguements parser"""
# Register the boolean type. This allows us to use booleans as arguments
parser.register('type', 'bool', lambda v: v.lower == "true")
# Hyperparameters regarding the neural network
parser.add_argument("--num_units", type=int, default=32, help="Network size.")
parser.add_argument("--num_layers", type=int, default=2,
help="Network depth.")
parser.add_argument("--encoder_type", type=str, default="uni", help="""\
uni | bi. For bi, we build num_layers/2 bi-directional layers.""")
parser.add_argument("--residual", type="bool", nargs="?", const=True,
default=False,