04c-mmf_sae.py
5.96 KB
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
# coding: utf-8
# In[2]:
# Import
import gensim
from scipy import sparse
import itertools
from sklearn import preprocessing
from keras.models import Sequential
from keras.optimizers import SGD,Adam
from mlp import *
import mlp
import sklearn.metrics
import shelve
import pickle
from utils import *
import sys
import os
import json
# In[4]:
infer_model=shelve.open("{}".format(sys.argv[2]))
in_dir = sys.argv[1]
if len(sys.argv) > 4 :
features_key = sys.argv[4]
else :
features_key = "LDA"
save_projection = True
#['ASR', 'TRS', 'LABEL']
# In[6]:
json_conf =json.load(open(sys.argv[3]))
sae_conf = json_conf["sae"]
hidden_size= sae_conf["hidden_size"]
input_activation=sae_conf["input_activation"]
output_activation=sae_conf["output_activation"]
loss=sae_conf["loss"]
epochs=sae_conf["epochs"]
batch=sae_conf["batch"]
patience=sae_conf["patience"]
do_do=sae_conf["do"]
try:
k = sae_conf["sgd"]
if sae_conf["sgd"]["name"] == "adam":
sgd = Adam(lr=sae_conf["sgd"]["lr"])
elif sae_conf["sgd"]["name"] == "sgd":
sgd = SGD(lr=sae_conf["sgd"]["lr"])
except :
sgd = sae_conf["sgd"]
name = json_conf["name"]
print name
try:
os.mkdir("{}/{}".format(in_dir,name))
except:
pass
db = shelve.open("{}/{}/ae_model.shelve".format(in_dir,name),writeback=True)
#
mlp_conf = json_conf["mlp"]
mlp_h = mlp_conf["hidden_size"]
mlp_loss = mlp_conf["loss"]
mlp_dropouts = mlp_conf["do"]
mlp_epochs = mlp_conf["epochs"]
mlp_batch_size = mlp_conf["batch"]
mlp_input_activation=mlp_conf["input_activation"]
mlp_output_activation=mlp_conf["output_activation"]
try:
k = mlp_conf["sgd"]
if mlp_conf["sgd"]["name"] == "adam":
mlp_sgd = Adam(lr=mlp_conf["sgd"]["lr"])
elif mlp_conf["sgd"]["name"] == "sgd" :
mlp_sgd = SGD(lr=mlp_conf["sgd"]["lr"])
except :
mlp_sgd = mlp_conf["sgd"]
keys = infer_model[features_key].keys()
db["SAE"] = {}
db["SAEFT"] = {}
for mod in keys :
res_tuple=train_sae(infer_model[features_key][mod]["TRAIN"],infer_model[features_key][mod]["DEV"],
infer_model[features_key][mod]["TEST"],
hidden_size,dropouts=do_do,
patience = "patience",sgd=sgd,input_activation="tanh",
output_activation="tanh",loss=loss,epochs=epochs,
batch_size=batch,verbose=0)
#print len(res), [len(x) for x in res[0]], [ len(x) for x in res[1]]
for i, levels in zip(["SAE","SAEFT"],res_tuple):
mlp_res_by_level = []
for lvl,res in enumerate(levels):
mlp_res_list=[]
for nb,layer in enumerate(res) :
if save_projection:
pd = pandas.DataFrame(layer[0])
col_count= (pd.sum(axis=0) != 0)
pd = pd.loc[:,col_count]
hdffile = "{}/{}/{}_{}_{}_df.hdf".format(in_dir,name,i,lvl,nb,mod)
print hdffile
pd.to_hdf(hdffile,"TRAIN")
pd = pandas.DataFrame(layer[1])
pd = pd.loc[:,col_count]
pd.to_hdf(hdffile,"DEV")
pd = pandas.DataFrame(layer[2])
pd = pd.loc[:,col_count]
pd.to_hdf(hdffile,"TEST")
del pd
mlp_res_list.append(train_mlp(layer[0],infer_model["LABEL"][mod]["TRAIN"],
layer[1],infer_model["LABEL"][mod]["DEV"],
layer[2],infer_model["LABEL"][mod]["TEST"],
mlp_h,loss=mlp_loss,dropouts=mlp_dropouts,
sgd=mlp_sgd,epochs=mlp_epochs,batch_size=mlp_batch_size,
fit_verbose=0))
mlp_res_by_level.append(mlp_res_list)
db[i][mod]=mlp_res_by_level
if "ASR" in keys and "TRS" in keys :
mod = "ASR"
mod2= "TRS"
res_tuple = train_sae(infer_model[features_key][mod]["TRAIN"],
infer_model[features_key][mod]["DEV"],
infer_model[features_key][mod]["TEST"],
hidden_size,dropouts=[0],patience="patience",
sgd=sgd,input_activation=input_activation,output_activation=input_activation,
loss=loss,epochs=epochs,batch_size=batch,
y_train=infer_model[features_key][mod2]["TRAIN"],
y_dev=infer_model[features_key][mod2]["DEV"],
y_test=infer_model[features_key][mod2]["TEST"])
for i , levels in zip(["SAE","SAEFT"],res_tuple):
mlp_res_by_level = []
for lvl,res in enumerate(levels) :
mlp_res_list=[]
for nb,layer in enumerate(res) :
if save_projection:
pd = pandas.DataFrame(layer[0])
col_count= (pd.sum(axis=0) != 0)
pd = pd.loc[:,col_count]
pd.to_hdf("{}/{}/{}_{}_{}_{}_df.hdf".format(in_dir,name,i,lvl,nb,"SPE"),"TRAIN")
pd = pandas.DataFrame(layer[1])
pd = pd.loc[:,col_count]
pd.to_hdf("{}/{}/{}_{}_{}_{}_df.hdf".format(in_dir,name,i,lvl,nb,"SPE"),"DEV")
pd = pandas.DataFrame(layer[2])
pd = pd.loc[:,col_count]
pd.to_hdf("{}/{}/{}_{}_{}_{}_df.hdf".format(in_dir,name,i,lvl,nb,"SPE"),"TEST")
del pd
mlp_res_list.append(train_mlp(layer[0],infer_model["LABEL"][mod]["TRAIN"],
layer[1],infer_model["LABEL"][mod]["DEV"],layer[2],
infer_model["LABEL"][mod]["TEST"],
mlp_h,loss=mlp_loss,dropouts=mlp_dropouts,
sgd=mlp_sgd,epochs=mlp_epochs,batch_size=mlp_batch_size,
fit_verbose=0))
mlp_res_by_level.append(mlp_res_list)
db[i]["SPE"] = mlp_res_by_level
db.sync()
db.close()