第一单元

1.这些机制中有一种是指令流水线技术，它允许CPU在执行一条指令的同时取出另一条指令。

One such mechanism, the instruction pipeline, allows the CPU to fetch one instruction while simultaneously executing another instruction.

 

2.这种机制用硬盘充当计算机存储器的一部分，扩大了计算机的存储空间，而且降低了价格，因为一个硬盘字节的价格比一个RAM字节要便宜的多。

This mechanism uses a hard disk as a part of computer's memory, expanding the computer's memory space while minimizing cost, because a byte of hard disk costs less than a byte of RAM. 

 

第二单元

1.DBA通过书写一系列的定义来创建最初的数据库模式，这些定义被DDL编译器翻译成永久地存储在数据字典中的表集合。

The DBA creates the original database schema by writing a set of definitions that is translated by the DDL compiler to a set of tables that is stored permanently in the data dictionary.

 

2.软件设计者使用专门的应用程序来创建新程序,这种专门程序通常被称为实用程序或开发程序。

Software designers create new programs by using special applications, often called utility programs or development programs.

 

3.程序员还使用另一种称为文本编辑器的程序来编写新程序，新程序是用称为程序设计语言的特殊符号来编写的。

A programmer uses another type of program called a text editor to write the new program in a special notation called a programming language.

 

4.这些磁盘上称为目录的额外信息，可指明盘上各种程序的名称、它们写入盘中的时间以及它们在磁盘介质上的起始位置。

Additional information on such disks, called directories, indicates the names of various programs on the disk, when they were written to the disk, and where the program begins on the disk media.

 

5.软件还可以管理硬件的使用方式，比如如何从存储设备检索信息。输入和输出硬件之间的交互是由称为基本输入输出系统的软件控制的。

Software also governs how the hardware is utilized; for example, how information is retrieved from a storage device. The interaction between the input and output hardware is controlled by software called the Basic Input Output System (BIOS) software.

 

第三单元

1.宏将那些用程序员定义的语言结构编写的代码段翻译成编程语言可以理解的语句。

Macros translate segments of code that have been written in a language structure defined by the programmer into statements that the programming language understands.

 

2.我们提出了一种分层结构的强化学习方法，以解决计划生成连贯的多句故事的视觉故事任务的挑战。

we propose a hierarchically structured reinforcement learning approach to address the challenges of planning for generating coherent multi-sentence stories for the visual storytelling task.

 

3.在我们的框架中，生成一个给定图像序列的故事的任务被划分为两个层次的解码器。高级译码器通过生成语义概念(即，主题)为每个图像序列。

Within our framework, the task of generating a story given a sequence of images is divide across a two-level hierachical decoder. The high-level decoder constructs a plan by generating a semantic concept (i.e.,topic) for each image in sequence.

 

4.底层解码器使用语义组合网络为每个图像生成一个句子，这有效地根据主题确定句子的生成。

The low-level decoder generates a sentence for each image using a semantic compositional network,which effectively grounds the sentece generation conditioned on the topic.

 

5.利用增强学习对两个解码器进行端到端的联合训练。我们在视觉讲故事(VIST)数据集上评估我们的模型。

The two docoder are jointly trianed end-to-end using reinforcement learning. We evaluate our model on the visual storytelling dataset.

 

6.来自自动和人工评估的经验结果表明，与强平面深度强化学习基线相比，提出的层次结构强化训练具有更好的性能。

Empirical results from both automatic and human evaluations demonstrate that the proposed hierarchically structrued reinrorced training achieves significantly better performance compared to a strong flat deep reinforcement learning baseline.

 

第四单元

1. 要使程序中的指令得到执行，计算机必须执行该程序，也就是说，计算机要读取该程序，然后按确切的顺序执行程序中编码的步骤，直至程序结束。

For the instructions to be carried out, a computer must execute a program, that is, the computer reads a program, and then follows the steps encoded in the program in a precise order until completion.

 

2.一个程序可被多次执行，每次执行可能产生不同的结果。这取决于用户提供给计算机的操作和数据。

A program can be executed many different times, with each execution yielding a potentially different result depending upon the operations and data that the user gives the computer.

 

3.今天，汇编语言很少用来编写整个程序，但是常用于程序员需要直接控制计算机某方面功能的情况。

Today, assembly language is rarely used to write an entire program, but is instead most often used when the programmer needs to directly control some aspect of the computer’s function.

 

4.程序经常被编写为一组较小的程序块，每块表示整个应用程序的某个方面。

Programs are often written as a set of smaller pieces, with each piece representing some aspect of the overall application program.

 

5.使计算机如此有用的能力之一是它们能够作出条件判定，并根据正在处理的条件数据值执行不同的指令。

One of the capabilities that make computers so useful is their ability to make conditional decisions and perform different instructions based on the values of data being processed.

 

第五单元

1.通常，在它需要修改之前会使用一段时间。

Usually, it is in use for a while before modifications are necessary.

 

2.由于软件中错误的发现、规则的改变或公司本身的变化都导致软件需要经常修改。

Modification is often needed due to errors found in the software, changes in rules or laws, or changes in the company itself.

 

3.整个开发过程始于分析阶段，这个阶段显示程序包应该做什么。

The development process starts with the analysis phase, which shows what the package should do.

 

4.在这个阶段，系统分析员定义需求，指出系统所要实现的目标。

In this phase, the systems analyst defines requirements that specify what the proposed system is to accomplish.

 

5.设计阶段定义系统怎样完成在分析阶段所定义的需求。

The design phase defines how the system will accomplish what was defined in the analysis phase.

 

6.结构图显示了如何将软件包分解成逻辑步骤.

A structure chart shows how to break your package into logical steps.

 

7.伪代码部分是英文，部分是用精确的算法细节来描述程序将完成什么的程序逻辑。

Pseudocode is part English and part program logic that describes, in precise algorithmic detail, what the program is to do.