File size: 8,820 Bytes
093b0a5 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 | import torch
import torch.nn as nn
import torch.nn.functional as F
from utils.masking import TriangularCausalMask, ProbMask
from layers.encoder import Encoder, EncoderLayer, ConvLayer, EncoderStack
from layers.decoder import Decoder, DecoderLayer
from layers.attn import FullAttention, ProbAttention, AttentionLayer
from layers.embed import DataEmbedding
class Informer(nn.Module):
def __init__(self, config):
# enc_in, dec_in, c_out, seq_len, label_len, out_len,
# factor=5, d_model=512, n_heads=8, e_layers=3, d_layers=2, d_ff=512,
# dropout=0.0, attn='prob', t_embed='fixed', freq='h', activation='gelu',
# output_attention = False, distil=True, mix=True,
# device=torch.device('cuda:0')
super(Informer, self).__init__()
self.pred_len = config.pred_len
self.attn = config.attn
self.output_attention = config.output_attention
# Encoding
self.enc_embedding = DataEmbedding(
config.enc_in, config.d_model, config.t_embed, config.freq, config.dropout_emb
)
self.dec_embedding = DataEmbedding(
config.dec_in, config.d_model, config.t_embed, config.freq, config.dropout_emb
)
# Attention
Attn = ProbAttention if config.attn == "prob" else FullAttention
# Encoder
self.encoder = Encoder(
[
EncoderLayer(
AttentionLayer(
Attn(
False,
config.factor,
attention_dropout=config.dropout,
output_attention=config.output_attention,
),
config.d_model,
config.n_heads,
mix=False,
),
config.d_model,
config.d_ff,
dropout=config.dropout,
activation=config.activation,
)
for l in range(config.e_layers)
],
[ConvLayer(config.d_model) for l in range(config.e_layers - 1)]
if config.distil
else None,
norm_layer=torch.nn.LayerNorm(config.d_model),
)
# Decoder
self.decoder = Decoder(
[
DecoderLayer(
AttentionLayer(
Attn(
True,
config.factor,
attention_dropout=config.dropout,
output_attention=False,
),
config.d_model,
config.n_heads,
mix=config.mix,
),
AttentionLayer(
FullAttention(
False,
config.factor,
attention_dropout=config.dropout,
output_attention=False,
),
config.d_model,
config.n_heads,
mix=False,
),
config.d_model,
config.d_ff,
dropout=config.dropout,
activation=config.activation,
)
for l in range(config.d_layers)
],
norm_layer=torch.nn.LayerNorm(config.d_model),
)
self.projection = nn.Linear(config.d_model, config.c_out, bias=True)
def forward(
self,
x_enc,
x_mark_enc,
x_dec,
x_mark_dec,
enc_self_mask=None,
dec_self_mask=None,
dec_enc_mask=None,
):
enc_out = self.enc_embedding(x_enc, x_mark_enc)
enc_out, attns = self.encoder(enc_out, attn_mask=enc_self_mask)
dec_out = self.dec_embedding(x_dec, x_mark_dec)
dec_out = self.decoder(
dec_out, enc_out, x_mask=dec_self_mask, cross_mask=dec_enc_mask
)
dec_out = self.projection(dec_out)
if self.output_attention:
return dec_out[:, -self.pred_len :, :], attns
else:
return dec_out[:, -self.pred_len :, :] # [B, L, D]
class InformerStack(nn.Module):
def __init__(self, config):
# enc_in, dec_in, c_out, seq_len, label_len, out_len,
# factor=5, d_model=512, n_heads=8, e_layers=[3,2,1], d_layers=2, d_ff=512,
# dropout=0.0, attn='prob', t_embed='fixed', freq='h', activation='gelu',
# output_attention = False, distil=True, mix=True,
# device=torch.device('cuda:0'))
super(InformerStack, self).__init__()
self.pred_len = config.pred_len
self.attn = config.attn
self.output_attention = config.output_attention
assert (
type(config.e_layers) is list
), "For Informer Stack e_layers must be a list"
# Encoding
self.enc_embedding = DataEmbedding(
config.enc_in, config.d_model, config.t_embed, config.freq, config.dropout_emb
)
self.dec_embedding = DataEmbedding(
config.dec_in, config.d_model, config.t_embed, config.freq, config.dropout_emb
)
# Attention
Attn = ProbAttention if config.attn == "prob" else FullAttention
# Encoder
inp_lens = list(
range(len(config.e_layers))
) # [0,1,2,...] you can customize here
encoders = [
Encoder(
[
EncoderLayer(
AttentionLayer(
Attn(
False,
config.factor,
attention_dropout=config.dropout,
output_attention=config.output_attention,
),
config.d_model,
config.n_heads,
mix=False,
),
config.d_model,
config.d_ff,
dropout=config.dropout,
activation=config.activation,
)
for l in range(el)
],
[ConvLayer(config.d_model) for l in range(el - 1)]
if config.distil
else None,
norm_layer=torch.nn.LayerNorm(config.d_model),
)
for el in config.e_layers
]
self.encoder = EncoderStack(encoders, inp_lens)
# Decoder
self.decoder = Decoder(
[
DecoderLayer(
AttentionLayer(
Attn(
True,
config.factor,
attention_dropout=config.dropout,
output_attention=False,
),
config.d_model,
config.n_heads,
mix=config.mix,
),
AttentionLayer(
FullAttention(
False,
config.factor,
attention_dropout=config.dropout,
output_attention=False,
),
config.d_model,
config.n_heads,
mix=False,
),
config.d_model,
config.d_ff,
dropout=config.dropout,
activation=config.activation,
)
for l in range(config.d_layers)
],
norm_layer=torch.nn.LayerNorm(config.d_model),
)
self.projection = nn.Linear(config.d_model, config.c_out, bias=True)
def forward(
self,
x_enc,
x_mark_enc,
x_dec,
x_mark_dec,
enc_self_mask=None,
dec_self_mask=None,
dec_enc_mask=None,
):
enc_out = self.enc_embedding(x_enc, x_mark_enc)
enc_out, attns = self.encoder(enc_out, attn_mask=enc_self_mask)
dec_out = self.dec_embedding(x_dec, x_mark_dec)
dec_out = self.decoder(
dec_out, enc_out, x_mask=dec_self_mask, cross_mask=dec_enc_mask
)
dec_out = self.projection(dec_out)
if self.output_attention:
return dec_out[:, -self.pred_len :, :], attns
else:
return dec_out[:, -self.pred_len :, :] # [B, L, D]
|