Keras混合模型在每个时期都给出相同的结果

如何解决Keras混合模型在每个时期都给出相同的结果

我创建了一个包含文本和图像的混合模型。当我训练模型时,在每个时期我都会得到相同的结果。下面是我的代码。

import tensorflow as tf
import pandas as pd
import numpy as np

base_dir = "D:/Dataset/xxxx/datasets/xxx/xx/xxxxx/"

import os

train_dir = os.path.join(base_dir,"trin.jsonl")
test_dir = os.path.join(base_dir,"tst.jsonl")
dev_dir = os.path.join(base_dir,"dv.jsonl")

df_train = pd.read_json(train_dir,lines=True)
df_test = pd.read_json(test_dir,lines=True)
df_dev = pd.read_json(dev_dir,lines=True)

df_train=df_train.set_index('id')
df_dev=df_dev.set_index('id')
df_test=df_test.set_index('id')

from tensorflow.keras import optimizers
from tensorflow.keras.preprocessing.image import ImageDataGenerator
import re
import spacy
import tensorflow as tf
from tensorflow.keras.preprocessing.text import Tokenizer
from tensorflow.keras.preprocessing.sequence import pad_sequences

nlp = spacy.load('en_core_web_md')

train_datagen = ImageDataGenerator(rescale=1./255)
test_datagen = ImageDataGenerator(rescale=1./255)

label_map = {1:"Hate",0:"No_Hate"}
df_dev['label']=df_dev['label'].map(label_map)
df_train['label']=df_train['label'].map(label_map)

train_generator = train_datagen.flow_from_dataframe(dataframe=df_train,directory=img_path,x_col="img",y_col="label",target_size=(224,224),batch_size=8500,class_mode="binary",shuffle=False)

def spacy_tokenizer(sentence):
    sentence = re.sub(r"[^a-zA-Z0-9]+"," ",sentence)
    sentence_list = [word.lemma_ for word in nlp(sentence) if not (word.is_space or word.is_stop or len(word)==1)]
    return ' '.join(sentence_list)
    
image_files = pd.Series(train_generator.filenames)
image_files = image_files.str.split('/',expand=True)[1].str[:-4]
image_files = list(map(int,image_files))

df_sorted = df_train.reindex(image_files)
df_sorted.head(1)

images,labels = next(train_generator)

tokenizer = Tokenizer(num_words=10000)

tokenizer.fit_on_texts(df_sorted['new_text'].values)
sequences = tokenizer.texts_to_sequences(df_sorted['new_text'].values)
train_padd = pad_sequences(sequences,maxlen=maxlen,padding='post',truncating='post')

from tensorflow.keras.models import Model
from tensorflow.keras import layers
from tensorflow.keras import models
from tensorflow.keras.layers import Embedding,Flatten,Dense
from tensorflow.keras.layers import Dense,LSTM,Embedding,Dropout,SpatialDropout1D,Conv1D,MaxPooling1D,GRU,BatchNormalization
from tensorflow.keras.layers import Input,Bidirectional,GlobalAveragePooling1D,GlobalMaxPooling1D,concatenate,LeakyReLU

def create_nlp():
    sequence_input=Input(shape=(maxlen))
    embedding_layer=Embedding(input_dim=text_embedding.shape[0],output_dim=text_embedding.shape[1],weights=[text_embedding],input_length=maxlen,trainable=False)
    embedded_sequence = embedding_layer(sequence_input)
    l_conv_1=Conv1D(128,5,activation='relu')(embedded_sequence)
    l_pool_1=MaxPooling1D(5)(l_conv_1)
    l_conv_2=Conv1D(128,activation='relu')(l_pool_1)
    l_pool_2=MaxPooling1D(5)(l_conv_2)
    l_flat = Flatten()(l_pool_2)
    model=Model(sequence_input,l_flat)
    return model
    
    
from tensorflow.keras.applications import VGG16
from tensorflow.keras import optimizers

def create_img():
    img_input=Input(shape=(224,224,3))
    conv_base = VGG16(weights='imagenet',include_top=False,input_shape=(224,3))
    conv_base.trainable = False
    conv_l_1=conv_base(img_input)
    flat_l = Flatten()(conv_l_1)
    dense_l = Dense(256,activation='relu')(flat_l)
    model = Model(img_input,dense_l)
    return model

nlp_1=create_nlp()
img_cnn=create_img()
combinedInput = concatenate([nlp_1.output,img_cnn.output])

x = Dense(4,activation="relu")(combinedInput)
x = Dense(1,activation="sigmoid")(x)
model1 = Model(inputs=[nlp_1.input,img_cnn.input],outputs=x)
opt = optimizers.Adam(lr=1e-3,decay=1e-3 / 200)
model1.compile(loss="binary_crossentropy",metrics=['acc'],optimizer=opt)

model1_history = model1.fit([train_padd,images],train_y,epochs=15,batch_size=16)

以下是我的训练结果:

Epoch 1/15
532/532 [==============================] - 104s 196ms/step - loss: 0.6528 - acc: 0.6412
Epoch 2/15
532/532 [==============================] - 103s 193ms/step - loss: 0.6528 - acc: 0.6412
Epoch 3/15
532/532 [==============================] - 103s 195ms/step - loss: 0.6528 - acc: 0.6412
Epoch 4/15
532/532 [==============================] - 103s 194ms/step - loss: 0.6528 - acc: 0.6412
Epoch 5/15
532/532 [==============================] - 103s 194ms/step - loss: 0.6528 - acc: 0.6412
Epoch 6/15
532/532 [==============================] - 103s 194ms/step - loss: 0.6528 - acc: 0.6412
Epoch 7/15
532/532 [==============================] - 103s 194ms/step - loss: 0.6528 - acc: 0.6412
Epoch 8/15
532/532 [==============================] - 104s 195ms/step - loss: 0.6528 - acc: 0.6412
Epoch 9/15
532/532 [==============================] - 106s 200ms/step - loss: 0.6528 - acc: 0.6412
Epoch 10/15
532/532 [==============================] - 109s 204ms/step - loss: 0.6528 - acc: 0.6412
Epoch 11/15
532/532 [==============================] - 104s 196ms/step - loss: 0.6528 - acc: 0.6412
Epoch 12/15
532/532 [==============================] - 103s 194ms/step - loss: 0.6528 - acc: 0.6412
Epoch 13/15
532/532 [==============================] - 103s 194ms/step - loss: 0.6528 - acc: 0.6412
Epoch 14/15
532/532 [==============================] - 104s 195ms/step - loss: 0.6528 - acc: 0.6412
Epoch 15/15
532/532 [==============================] - 103s 193ms/step - loss: 0.6528 - acc: 0.6412

我也在终端中获取以下日志:

分配器(GPU_0_bfc)内存不足,试图分配2.36GiB freed_by_count = 0。呼叫者指出这不是 失败,但可能意味着如果更多,性能可能会提高 内存可用。

解决方法

看看here,您可能只使用了不合适的优化程序。如果那没有帮助,我将尝试使用1作为批处理大小,以查看至少在第一次运行中是否存在更改。同样,学习率可能是个问题,请尝试发挥它的价值,看看准确性是否会发生变化。

版权声明:本文内容由互联网用户自发贡献,该文观点与技术仅代表作者本人。本站仅提供信息存储空间服务,不拥有所有权,不承担相关法律责任。如发现本站有涉嫌侵权/违法违规的内容, 请发送邮件至 dio@foxmail.com 举报,一经查实,本站将立刻删除。

相关推荐


依赖报错 idea导入项目后依赖报错,解决方案:https://blog.csdn.net/weixin_42420249/article/details/81191861 依赖版本报错:更换其他版本 无法下载依赖可参考:https://blog.csdn.net/weixin_42628809/a
错误1:代码生成器依赖和mybatis依赖冲突 启动项目时报错如下 2021-12-03 13:33:33.927 ERROR 7228 [ main] o.s.b.d.LoggingFailureAnalysisReporter : *************************** APPL
错误1:gradle项目控制台输出为乱码 # 解决方案:https://blog.csdn.net/weixin_43501566/article/details/112482302 # 在gradle-wrapper.properties 添加以下内容 org.gradle.jvmargs=-Df
错误还原:在查询的过程中,传入的workType为0时,该条件不起作用 <select id="xxx"> SELECT di.id, di.name, di.work_type, di.updated... <where> <if test=&qu
报错如下,gcc版本太低 ^ server.c:5346:31: 错误:‘struct redisServer’没有名为‘server_cpulist’的成员 redisSetCpuAffinity(server.server_cpulist); ^ server.c: 在函数‘hasActiveC
解决方案1 1、改项目中.idea/workspace.xml配置文件,增加dynamic.classpath参数 2、搜索PropertiesComponent,添加如下 <property name="dynamic.classpath" value="tru
删除根组件app.vue中的默认代码后报错:Module Error (from ./node_modules/eslint-loader/index.js): 解决方案:关闭ESlint代码检测,在项目根目录创建vue.config.js,在文件中添加 module.exports = { lin
查看spark默认的python版本 [root@master day27]# pyspark /home/software/spark-2.3.4-bin-hadoop2.7/conf/spark-env.sh: line 2: /usr/local/hadoop/bin/hadoop: No s
使用本地python环境可以成功执行 import pandas as pd import matplotlib.pyplot as plt # 设置字体 plt.rcParams['font.sans-serif'] = ['SimHei'] # 能正确显示负号 p
错误1:Request method ‘DELETE‘ not supported 错误还原:controller层有一个接口,访问该接口时报错:Request method ‘DELETE‘ not supported 错误原因:没有接收到前端传入的参数,修改为如下 参考 错误2:cannot r
错误1:启动docker镜像时报错:Error response from daemon: driver failed programming external connectivity on endpoint quirky_allen 解决方法:重启docker -> systemctl r
错误1:private field ‘xxx‘ is never assigned 按Altʾnter快捷键,选择第2项 参考:https://blog.csdn.net/shi_hong_fei_hei/article/details/88814070 错误2:启动时报错,不能找到主启动类 #
报错如下,通过源不能下载,最后警告pip需升级版本 Requirement already satisfied: pip in c:\users\ychen\appdata\local\programs\python\python310\lib\site-packages (22.0.4) Coll
错误1:maven打包报错 错误还原:使用maven打包项目时报错如下 [ERROR] Failed to execute goal org.apache.maven.plugins:maven-resources-plugin:3.2.0:resources (default-resources)
错误1:服务调用时报错 服务消费者模块assess通过openFeign调用服务提供者模块hires 如下为服务提供者模块hires的控制层接口 @RestController @RequestMapping("/hires") public class FeignControl
错误1:运行项目后报如下错误 解决方案 报错2:Failed to execute goal org.apache.maven.plugins:maven-compiler-plugin:3.8.1:compile (default-compile) on project sb 解决方案:在pom.
参考 错误原因 过滤器或拦截器在生效时,redisTemplate还没有注入 解决方案:在注入容器时就生效 @Component //项目运行时就注入Spring容器 public class RedisBean { @Resource private RedisTemplate<String
使用vite构建项目报错 C:\Users\ychen\work>npm init @vitejs/app @vitejs/create-app is deprecated, use npm init vite instead C:\Users\ychen\AppData\Local\npm-