Try   HackMD

7.11.2 Seq2Seq 模型的實現

Seq2Seq模型由編碼器和解碼器兩個循環神經網路。

Image Not Showing Possible Reasons
  • The image was uploaded to a note which you don't have access to
  • The note which the image was originally uploaded to has been deleted
Learn More →

教師強制(Teacher Forcing)

每次不使用上一個state的輸出作爲下一個state的輸入，而是直接使用訓練數據的標準答案(ground truth)的對應上一項作爲下一個state的輸入

優點: 收斂速度提高

缺點: 實際泛化能力降低

7.11.3 字元級的Seq2Seq模型

Image Not Showing Possible Reasons
  • The image was uploaded to a note which you don't have access to
  • The note which the image was originally uploaded to has been deleted
Learn More →

1.字元單詞表

2.讀取訓練樣本並建構字元單詞表

3.訓練字元級的Seq2Seq模型

問題

1.Seq2Seq模型的結構為和?
2.Seq2Seq的應用為何?

Last changed by 
 
WingSyncAlgorithm
0
39

Read more

7.1_張昀甯

可用於解決資料之間具有序列依賴關係的問題。

Oct 12, 2024
7.10.2-7.11.1_周哲煒

多層LSTM每個隱含層的隱狀態大小一樣，但為使其能適應不同大小的出–>增加全連接層，以適應不同大小的輸出

Jul 13, 2024
7.11.4_李紘宇

import randomfrom PyQt5.QtWidgets import (QApplication, QMainWindow, QVBoxLayout,QWidget, QSpinBox, QSplitter, QHBoxLayout, QLabel)from PyQt5.QtCore import Qt, QTimerfrom matplotlib.backends.backend_qt5agg import FigureCanvasQTAgg as FigureCanvasfrom matplotlib.figure import Figureimport numpy as np

Jul 6, 2024
7.9.2~7.10.1_元甫

##　LSTM

Jun 29, 2024
Read more from WingSyncAlgorithm
Published on HackMD