五、阅读理解。(2025·江苏宿迁期中)
Recently, people around the world have been very excited about a new AI model called DeepSeek - R1. The company behind it, also called DeepSeek, is a business in Hangzhou, Zhejiang.
The new AI model first appeared on 20 Jan. 2025. In just one week, it beat OpenAI's ChatGPT and became No. 1 free application (应用程序) in the App Store in China.
According to DeepSeek, the new AI model is as good at dealing with difficult activities as other AI models. But it needs much less computing power (算力) to do that. It is also one of the cheapest of its kind.
How does DeepSeek - R1 learn? Instead of using the traditional ways, DeepSeek mainly uses Reinforcement Learning (RL, 强化学习). It helps the AI model learn by working out the right way to do things on its own to get points. This makes DeepSeek - R1 better at reasoning and solving hard problems.
What's more, DeepSeek - R1 is open - source (开放源代码的). Everyone is free to get a copy of its code (代码). They can also change the code to meet their own needs. "Being open - source is very important for sharing knowledge and making technology (技术) better for everyone," said Liang Wenfeng, who set up DeepSeek.
(
A. It's new.
B. It's free.
C. It's special.
D. It's popular.
(
A. Because it changes our needs.
B. Because it helps share knowledge.
C. Because it makes AI cheaper.
D. Because it uses new technology.
(
A. The differences between AI models.
B. The problems faced by DeepSeek - R1.
C. The introduction to DeepSeek - R1.
D. The importance of being open - source.
Recently, people around the world have been very excited about a new AI model called DeepSeek - R1. The company behind it, also called DeepSeek, is a business in Hangzhou, Zhejiang.
The new AI model first appeared on 20 Jan. 2025. In just one week, it beat OpenAI's ChatGPT and became No. 1 free application (应用程序) in the App Store in China.
According to DeepSeek, the new AI model is as good at dealing with difficult activities as other AI models. But it needs much less computing power (算力) to do that. It is also one of the cheapest of its kind.
How does DeepSeek - R1 learn? Instead of using the traditional ways, DeepSeek mainly uses Reinforcement Learning (RL, 强化学习). It helps the AI model learn by working out the right way to do things on its own to get points. This makes DeepSeek - R1 better at reasoning and solving hard problems.
What's more, DeepSeek - R1 is open - source (开放源代码的). Everyone is free to get a copy of its code (代码). They can also change the code to meet their own needs. "Being open - source is very important for sharing knowledge and making technology (技术) better for everyone," said Liang Wenfeng, who set up DeepSeek.
(
D
)1. What does Paragraph 2 mainly say about DeepSeek - R1?A. It's new.
B. It's free.
C. It's special.
D. It's popular.
(
B
)2. Why is being open - source important?A. Because it changes our needs.
B. Because it helps share knowledge.
C. Because it makes AI cheaper.
D. Because it uses new technology.
(
C
)3. What is the passage mainly about?A. The differences between AI models.
B. The problems faced by DeepSeek - R1.
C. The introduction to DeepSeek - R1.
D. The importance of being open - source.
答案：五、1. D 【点拨】根据第二段中的“In just one week, it beat OpenAI's ChatGPT and became No. 1 free application（应用程序）in the App Store in China.”可知，DeepSeek-R1 在短时间内迅速流行并成为中国应用商店的头号免费应用。故选 D。
2. B 【点拨】根据最后一段中的“Being open-source is very important for sharing knowledge and making technology(技术) better for everyone”可知，开源有助于知识共享，使技术惠及更多人，扩大 AI 的使用范围。故选 B。
3. C 【点拨】通读全文可知，本文围绕 DeepSeek-R1 的性能、成本、学习方式和开源特点展开，属于对该 AI 模型的整体介绍。故选 C。
2. B 【点拨】根据最后一段中的“Being open-source is very important for sharing knowledge and making technology(技术) better for everyone”可知，开源有助于知识共享，使技术惠及更多人，扩大 AI 的使用范围。故选 B。
3. C 【点拨】通读全文可知，本文围绕 DeepSeek-R1 的性能、成本、学习方式和开源特点展开，属于对该 AI 模型的整体介绍。故选 C。
解析：
【分析】
做这篇阅读理解时，首先可以先扫读题干划出关键词，再带着问题回原文定位对应内容：第1题是段落主旨题，直接定位第二段，梳理段落核心信息，排除只涉及局部细节的干扰选项，提炼段落核心表达的含义；第2题是细节理解题，定位最后一段关于开源的描述，直接匹配原文给出的原因即可选出答案；第3题是全文主旨题，通读所有段落，梳理每段的核心内容，排除只能覆盖局部段落的片面选项，选出能概括全文的主旨项。
【解析】
1. 定位第二段内容：该段说明DeepSeek-R1在2025年1月20日首次发布，仅一周就击败ChatGPT成为中国区App Store免费应用榜第一名，核心体现它短时间内收获大量用户、十分受欢迎。A选项“它是新的”只是上线时间附带的属性，不是段落核心；B选项“它是免费的”只是榜单的属性，属于局部细节；C选项“它很特别”第二段没有相关表述，因此选D。
2. 定位最后一段中DeepSeek创始人的原话“Being open-source is very important for sharing knowledge and making technology better for everyone”，直接对应“开源有助于知识共享”的描述。A选项“改变我们的需求”是用户修改代码的作用，不是开源重要的原因；C选项“让AI更便宜”是第三段提到的该模型本身的优势，和开源无关；D选项“使用新技术”原文没有相关表述，因此选B。
3. 通读全文可知，文章依次介绍了DeepSeek-R1的研发主体、上线后的火爆表现、低算力低成本的优势、采用强化学习的训练模式、开源的特点，整体是对这款AI模型的全面介绍。A选项“不同AI模型的差异”仅在文中略有提及，不是核心内容；B选项“DeepSeek-R1面临的问题”全文完全没有涉及；D选项“开源的重要性”只是最后一段的局部内容，无法概括全文，因此选C。
【答案】
1. D 2. B 3. C
【知识点】
细节信息查找，段落主旨概括，全文主旨理解
【点评】
这是一篇贴合当下科技热点的说明类阅读，题型以基础细节题和主旨题为主，没有设置复杂的推理类考点，只要学生能准确定位题干对应的原文位置，区分局部细节和整体主旨，就可以顺利选出正确答案，同时也能让学生了解前沿AI产品的相关特点。
【难度系数】
0.8
做这篇阅读理解时，首先可以先扫读题干划出关键词，再带着问题回原文定位对应内容：第1题是段落主旨题，直接定位第二段，梳理段落核心信息，排除只涉及局部细节的干扰选项，提炼段落核心表达的含义；第2题是细节理解题，定位最后一段关于开源的描述，直接匹配原文给出的原因即可选出答案；第3题是全文主旨题，通读所有段落，梳理每段的核心内容，排除只能覆盖局部段落的片面选项，选出能概括全文的主旨项。
【解析】
1. 定位第二段内容：该段说明DeepSeek-R1在2025年1月20日首次发布，仅一周就击败ChatGPT成为中国区App Store免费应用榜第一名，核心体现它短时间内收获大量用户、十分受欢迎。A选项“它是新的”只是上线时间附带的属性，不是段落核心；B选项“它是免费的”只是榜单的属性，属于局部细节；C选项“它很特别”第二段没有相关表述，因此选D。
2. 定位最后一段中DeepSeek创始人的原话“Being open-source is very important for sharing knowledge and making technology better for everyone”，直接对应“开源有助于知识共享”的描述。A选项“改变我们的需求”是用户修改代码的作用，不是开源重要的原因；C选项“让AI更便宜”是第三段提到的该模型本身的优势，和开源无关；D选项“使用新技术”原文没有相关表述，因此选B。
3. 通读全文可知，文章依次介绍了DeepSeek-R1的研发主体、上线后的火爆表现、低算力低成本的优势、采用强化学习的训练模式、开源的特点，整体是对这款AI模型的全面介绍。A选项“不同AI模型的差异”仅在文中略有提及，不是核心内容；B选项“DeepSeek-R1面临的问题”全文完全没有涉及；D选项“开源的重要性”只是最后一段的局部内容，无法概括全文，因此选C。
【答案】
1. D 2. B 3. C
【知识点】
细节信息查找，段落主旨概括，全文主旨理解
【点评】
这是一篇贴合当下科技热点的说明类阅读，题型以基础细节题和主旨题为主，没有设置复杂的推理类考点，只要学生能准确定位题干对应的原文位置，区分局部细节和整体主旨，就可以顺利选出正确答案，同时也能让学生了解前沿AI产品的相关特点。
【难度系数】
0.8