def make_encoder(opt, embeddings, embeddings_inter=None):
"""
Various encoder dispatcher function.
Args:
opt: the option in current environment.
embeddings (Embeddings): vocab embeddings for this encoder.
"""
if opt.encoder_type == "transformer":
return TransformerEncoder(opt.enc_layers, opt.rnn_size, opt.dropout,
embeddings)
elif opt.encoder_type == "cnn":
return CNNEncoder(opt.enc_layers, opt.rnn_size, opt.cnn_kernel_width,
opt.dropout, embeddings)
elif opt.encoder_type == "mean":
return MeanEncoder(opt.enc_layers, embeddings)
elif opt.encoder_type == "double_encoder":
print("The double encoder will be needed here")
opt.brnn = True
return DoubleRNNEncoder(opt.rnn_type, opt.brnn, opt.dec_layers,
opt.rnn_size, opt.dropout, embeddings,
embeddings_inter)
else:
# "rnn" or "brnn"
print("The double encoder will be needed here")
opt.brnn = True
return RNNEncoder(opt.rnn_type, opt.brnn, opt.dec_layers, opt.rnn_size,
opt.dropout, embeddings)
def make_encoder(opt, embeddings):
"""
Various encoder dispatcher function.
Args:
opt: the option in current environment.
embeddings (Embeddings): vocab embeddings for this encoder.
"""
if opt.encoder_type == "transformer":
return TransformerEncoder(opt.enc_layers, opt.rnn_size, opt.dropout,
embeddings)
elif opt.encoder_type == "cnn":
return CNNEncoder(opt.enc_layers, opt.rnn_size, opt.cnn_kernel_width,
opt.dropout, embeddings)
elif opt.encoder_type == "mean":
return MeanEncoder(opt.enc_layers, embeddings)
elif opt.encoder_type == "trigramrnn":
return RNNTrigramsEncoder(opt.rnn_type, True, opt.enc_layers,
opt.rnn_size, opt.dropout, embeddings,
opt.bridge)
else:
# NOTE: THIS IS WHAT GETS TRIGGERED BY DEFAULT EXPERIMENT
# "rnn" or "brnn"
print('About to make encoder')
print(f"opt.rnn_type={opt.rnn_type}")
print(f"opt.brnn={opt.brnn}")
print(f"opt.enc_layers={opt.enc_layers}")
print(f"opt.rnn_size ={opt.rnn_size}")
print(f"opt.dropout={opt.dropout}")
print(f"embeddings={embeddings}")
print(f"opt.bridge={opt.bridge}")
return RNNEncoder(opt.rnn_type, opt.brnn, opt.enc_layers, opt.rnn_size,
opt.dropout, embeddings, opt.bridge)
def make_encoder(opt, embeddings, morph_embeddings=None):
"""
Various encoder dispatcher function.
Args:
opt: the option in current environment.
embeddings (Embeddings): vocab embeddings for this encoder.
"""
if opt.encoder_type == "transformer":
return TransformerEncoder(opt.enc_layers, opt.rnn_size, opt.dropout,
embeddings)
elif opt.encoder_type == "cnn":
return CNNEncoder(opt.enc_layers, opt.rnn_size, opt.cnn_kernel_width,
opt.dropout, embeddings)
elif opt.encoder_type == "mean":
return MeanEncoder(opt.enc_layers, embeddings)
elif opt.encoder_type == "gcn":
print('use gates = ', opt.gcn_use_gates)
return GCNEncoder(embeddings, opt.gcn_num_inputs, opt.gcn_num_units,
opt.gcn_num_labels, opt.gcn_num_layers,
opt.gcn_in_arcs, opt.gcn_out_arcs,
opt.gcn_batch_first, opt.gcn_residual,
opt.gcn_use_gates, opt.gcn_use_glus,
morph_embeddings)
else:
# "rnn" or "brnn"
return RNNEncoder(opt.rnn_type, opt.brnn, opt.enc_layers, opt.rnn_size,
opt.dropout, embeddings, opt.bridge)
def make_encoder(opt, embeddings, for_vae=False):
"""
Various encoder dispatcher function.
Args:
opt: the option in current environment.
embeddings (Embeddings): vocab embeddings for this encoder.
"""
if for_vae:
enc_layers = opt.enc_layers
rnn_size = opt.rnn_size_vae
dropout = opt.dropout_vae
else:
enc_layers = opt.enc_layers
rnn_size = opt.rnn_size
dropout = opt.dropout
if opt.encoder_type == "transformer":
return TransformerEncoder(opt.enc_layers, opt.rnn_size, opt.dropout,
embeddings)
elif opt.encoder_type == "cnn":
return CNNEncoder(opt.enc_layers, opt.rnn_size, opt.cnn_kernel_width,
opt.dropout, embeddings)
elif opt.encoder_type == "mean":
return MeanEncoder(opt.enc_layers, embeddings)
else:
# "rnn" or "brnn"
return RNNEncoder(opt.rnn_type, opt.brnn, enc_layers, rnn_size,
dropout, embeddings)
def make_encoder(opt, embeddings):
"""
Various encoder dispatcher function.
Args:
opt: the option in current environment.
embeddings (Embeddings): vocab embeddings for this encoder.
"""
if opt.encoder_type == "transformer":
return TransformerEncoder(opt.enc_layers, opt.rnn_size,
opt.dropout, embeddings)
elif opt.encoder_type == "cnn":
return CNNEncoder(opt.enc_layers, opt.rnn_size,
opt.cnn_kernel_width,
opt.dropout, embeddings)
elif opt.encoder_type == "mean":
return MeanEncoder(opt.enc_layers, embeddings)
elif opt.encoder_type == "trigramrnn":
return RNNTrigramsEncoder(opt.rnn_type, True, opt.enc_layers,
opt.rnn_size, opt.dropout, embeddings,
opt.bridge)
else:
# "rnn" or "brnn"
return RNNEncoder(opt.rnn_type, opt.brnn, opt.enc_layers,
opt.rnn_size, opt.dropout, embeddings,
opt.bridge)
def make_encoder(opt, embeddings):
"""
Various encoder dispatcher function.
Args:
opt: the option in current environment.
embeddings (Embeddings): vocab embeddings for this encoder.
"""
if opt.encoder_type == "transformer":
return TransformerEncoder(opt.enc_layers, opt.rnn_size, opt.dropout,
embeddings)
elif opt.encoder_type == "cnn":
return CNNEncoder(opt.enc_layers, opt.rnn_size, opt.cnn_kernel_width,
opt.dropout, embeddings)
elif opt.encoder_type == "mean":
return MeanEncoder(opt.enc_layers, embeddings)
else:
# "rnn" or "brnn"
return RNNEncoder(opt.rnn_type, opt.brnn, opt.enc_layers, opt.rnn_size,
opt.dropout, embeddings, opt.bridge, opt.elmo,
opt.elmo_size, opt.elmo_options, opt.elmo_weight,
opt.subword_elmo, opt.subword_elmo_size,
opt.subword_elmo_options, opt.subword_weight,
opt.subword_spm_model, opt.node2vec,
opt.node2vec_emb_size, opt.node2vec_weight,
use_gpu(opt))
def make_encoder(opt, embeddings, mmod_imgw=False):
"""
Various encoder dispatcher function.
Args:
opt: the option in current environment.
embeddings (Embeddings): vocab embeddings for this encoder.
"""
if opt.encoder_type == "transformer":
if mmod_imgw:
return multimodal.MultiModalTransformerEncoder(
opt.enc_layers, opt.rnn_size,
opt.img_feat_dim,
opt.dropout, embeddings)
else:
return TransformerEncoder(opt.enc_layers, opt.rnn_size,
opt.dropout, embeddings)
elif opt.encoder_type == "cnn":
return CNNEncoder(opt.enc_layers, opt.rnn_size,
opt.cnn_kernel_width,
opt.dropout, embeddings)
elif opt.encoder_type == "mean":
return MeanEncoder(opt.enc_layers, embeddings)
else:
# "rnn" or "brnn"
return RNNEncoder(opt.rnn_type, opt.brnn, opt.enc_layers,
opt.rnn_size, opt.dropout, embeddings,
opt.bridge)
