问题描述
我正在编写使用三层损失的几个镜头问答模型。我能够训练我的模型，但是当我尝试预测时，我得到了上述错误。
我已将我的代码的以下专家提供以供参考。我已经尝试通过检查数据类型和其他因素但没有运气。
谁能告诉我这里出了什么问题？
import pandas as pd
import numpy as np
df = pd.read_csv(r'/content/IAS16_standard.csv',encoding = "latin1")
rev_df = (df.drop('Questions',axis=1)
.join
(
df.Questions
.str
.split(',',expand=True)
.stack()
.reset_index(drop=True,level=1)
.rename('Questions')
))
train=rev_df[['Questions','Intent']]
train.reset_index(drop=True)
import tensorflow as tf
import tensorflow_hub as hub
module_url = 'https://tfhub.dev/google/universal-sentence-encoder-large/4'
# Import the Universal Sentence Encoder's TF Hub module
embed = hub.load(module_url)
import numpy as np # linear algebra
import pandas as pd # data processing,CSV file I/O (e.g. pd.read_csv)
import tensorflow as tf
import tensorflow_hub as hub
import keras
import keras.backend as K
from keras.layers import *
from keras.callbacks import *
from keras.optimizers import *
from keras import Model
from keras.layers.core import Lambda,Flatten,Dense
from keras.layers import Bidirectional,LSTM
import pickle
import os
input_text1 = Input(shape=(512,))
x = Dense(256,activation='relu')(input_text1)
x = Dropout(0.4)(x)
x = Batchnormalization()(x)
x = Dense(64,activation='relu',kernel_regularizer=keras.regularizers.l2(0.001))(x)
x = Dropout(0.4)(x)
dense_layer = Dense(128,name='dense_layer')(x)
norm_layer = Lambda(lambda x: K.l2_normalize(x,axis=1),name='norm_layer')(dense_layer)
model=Model(inputs=[input_text1],outputs=norm_layer)
model.summary()
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
input_1 (InputLayer) [(None,512)] 0
_________________________________________________________________
dense (Dense) (None,256) 131328
_________________________________________________________________
dropout (Dropout) (None,256) 0
_________________________________________________________________
batch_normalization (BatchNo (None,256) 1024
_________________________________________________________________
dense_1 (Dense) (None,64) 16448
_________________________________________________________________
dropout_1 (Dropout) (None,64) 0
_________________________________________________________________
dense_layer (Dense) (None,128) 8320
_________________________________________________________________
norm_layer (Lambda) (None,128) 0
=================================================================
Total params: 157,120
Trainable params: 156,608
Non-trainable params: 512
from keras import backend as K
from keras.models import Model
from keras.layers import Input,Layer
# Input for anchor,positive and negative images
in_a = Input(shape=(512,))
in_p = Input(shape=(512,))
in_n = Input(shape=(512,))
# Output for anchor,positive and negative embedding vectors
# The nn4_small model instance is shared (Siamese network)
emb_a = model(in_a)
emb_p = model(in_p)
emb_n = model(in_n)
class TripletLossLayer(Layer):
def __init__(self,alpha,**kwargs):
self.alpha = alpha
super(TripletLossLayer,self).__init__(**kwargs)
def triplet_loss(self,inputs):
a,p,n = inputs
p_dist = K.sum(K.square(a-p),axis=-1)
n_dist = K.sum(K.square(a-n),axis=-1)
return K.sum(K.maximum(p_dist - n_dist + self.alpha,0),axis=0)
def call(self,inputs):
loss = self.triplet_loss(inputs)
self.add_loss(loss)
return loss
# Layer that computes the triplet loss from anchor,positive and negative embedding vectors
triplet_loss_layer = TripletLossLayer(alpha=0.4,name='triplet_loss_layer')([emb_a,emb_p,emb_n])
# Model that can be trained with anchor,positive negative images
nn4_small2_train = Model([in_a,in_p,in_n],triplet_loss_layer)
nn4_small2_train.summary()
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_2 (InputLayer) [(None,512)] 0
__________________________________________________________________________________________________
input_3 (InputLayer) [(None,512)] 0
__________________________________________________________________________________________________
input_4 (InputLayer) [(None,512)] 0
__________________________________________________________________________________________________
model (Functional) (None,128) 157120 input_2[0][0]
input_3[0][0]
input_4[0][0]
__________________________________________________________________________________________________
triplet_loss_layer (TripletLoss () 0 model[0][0]
model[1][0]
model[2][0]
==================================================================================================
Total params: 157,608
Non-trainable params: 512
unique_train_label=np.array(train['Intent'].unique().tolist())
labels_train=np.array(train['Intent'].tolist())
map_train_label_indices = {label: np.flatnonzero(labels_train == label) for label in unique_train_label}
def get_triplets(unique_train_label,map_train_label_indices):
label_l,label_r = np.random.choice(unique_train_label,2,replace=True)
a,p = np.random.choice(map_train_label_indices[label_l],replace=True)
n = np.random.choice(map_train_label_indices[label_r])
return a,n
def get_triplets_batch(k,train_set,unique_train_label,map_train_label_indices,embed):
while True:
idxs_a,idxs_p,idxs_n = [],[],[]
for _ in range(k):
a,n = get_triplets(unique_train_label,map_train_label_indices)
idxs_a.append(a)
idxs_p.append(p)
idxs_n.append(n)
a=train_set.iloc[idxs_a].values.tolist()
b=train_set.iloc[idxs_p].values.tolist()
c=train_set.iloc[idxs_n].values.tolist()
a = embed(a)
p = embed(b)
n = embed(c)
# return train_set[idxs_a],train_set[idxs_p],train_set[idxs_n]
yield [a,n],[]
nn4_small2_train.compile(loss=None,optimizer='adam')
nn4_small2_train.fit(get_triplets_batch(128,train['Questions'],embed),epochs=100,steps_per_epoch=10)
X_train = model.predict({'input_1':embed(np.stack(np.array(train['Questions'].values.tolist())))})
y_train = np.array(train['Intent'].values.tolist())
最后一行给出了错误。
解决方法
暂无找到可以解决该程序问题的有效方法，小编努力寻找整理中！
如果你已经找到好的解决方法，欢迎将解决方案带上本链接一起发送给小编。
小编邮箱:dio#foxmail.com (将#修改为@）