本地部署大语言模型-DeepSeek

DeepSeek 是国内顶尖 AI 团队「深度求索」开发的多模态大模型,具备数学推理、代码生成等深度能力,堪称"AI界的六边形战士"。

Hostease AMD 9950X/96G/3.84T NVMe/1G/5IP/RTX4090 GPU服务器提供多种计费模式。

DeepSeek-R1-32B配置

配置项

规格要求

CPU

16 核以上(如 AMD Ryzen 9950)

内存

96GB

硬盘

960GB

显卡

24GB+ 显存(如 RTX 4090)

# cat cpuinfo |grep name|head -1
model name    : AMD Ryzen 9 9950X 16-Core Processor

# nvidia-smi 
Sun Mar  2 14:51:30 2025       
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.183.01             Driver Version: 535.183.01   CUDA Version: 12.2     |
|-----------------------------------------+----------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |         Memory-Usage | GPU-Util  Compute M. |
|                                         |                      |               MIG M. |
|=========================================+======================+======================|
|   0  NVIDIA GeForce RTX 4090        Off | 00000000:01:00.0 Off |                  Off |
|  0%   27C    P8               5W / 450W |    495MiB / 24564MiB |      0%      Default |
|                                         |                      |                  N/A |
+-----------------------------------------+----------------------+----------------------+
安装步骤(略写)

apt update
apt install nvidia-driver-535 -y
curl -fsSL https://ollama.com/install.sh | sh
ollama run deepseek-r1:32b-qwen-distill-q4_K_M --verbose

创建对话返回结果

>>> which is bigger,9.8 or 9.11?
<think>
Alright, so I'm trying to figure out whether 9.8 is bigger than 9.11 or not. At first glance, it seems like both numbers are pretty close to each 
other since they both start with a 9 and then have some decimal parts. But wait, maybe there's something more to it.Okay, let me break this down step by step. First, I need to understand what these decimals represent. So, 9.8 can be read as nine plus eight 
tenths, right? And 9.11 is nine plus eleven hundredths. Hmm, that makes sense. Now, the question is asking which one is bigger.Maybe it would help if I line them up vertically to compare the digits after the decimal point more easily:```9.809.11
```By adding a zero to 9.8 to make it 9.80, we can see that both numbers have two decimal places now. This should make the comparison straightforward.Now, let's look at the tenths place first:- For 9.80, the tenths digit is 8.
- For 9.11, the tenths digit is 1.Okay, so right here, it seems like 8 is greater than 1. That would mean that 9.80 is bigger because its tenths place is larger than the tenths 
place of 9.11.But wait a second—is there something else I should consider? Maybe rounding or anything like that?Well, since both numbers are already given to two decimal places and we've added an extra zero to make them consistent, I think it's safe to say 
that no further adjustment is necessary.Still, just to be thorough, let me convert these decimals into fractions to see if there's a difference in their values.For 9.80:- 0.80 as a fraction is the same as \( \frac{80}{100} \), which simplifies down to \( \frac{4}{5} \).So, 9.80 is equal to \( 9 + \frac{4}{5} = \frac{45}{5} + \frac{4}{5} = \frac{49}{5} \).For 9.11:- 0.11 as a fraction is approximately \( \frac{11}{100} \).So, 9.11 is equal to \( 9 + \frac{11}{100} = \frac{900}{100} + \frac{11}{100} = \frac{911}{100} \).Now, let's compare these two fractions:- \( \frac{49}{5} = 9.8 \)
- \( \frac{911}{100} = 9.11 \)To see which one is bigger, I can convert both to have the same denominator or find a common basis for comparison.Let's try finding equivalent fractions with the same denominator:The denominators are 5 and 100. Since 100 is a multiple of 5 (specifically, \( \frac{100}{5} = 20 \)), I'll convert both fractions to have a 
denominator of 100.For \( \frac{49}{5} \):- Multiply numerator and denominator by 20: \( \frac{49 \times 20}{5 \times 20} = \frac{980}{100} \).So, \( \frac{49}{5} = \frac{980}{100} \).Now we have:- \( \frac{980}{100} \) (which is 9.8)
- \( \frac{911}{100} \) (which is 9.11)Comparing the numerators directly, since both fractions have the same denominator:- 980 > 911This means that \( \frac{980}{100} > \frac{911}{100} \), or in other words, 9.8 is greater than 9.11.Wait a minute—this contradicts what I thought earlier where the tenths place indicated that 9.8 was bigger because of having an 8 instead of a 1 in 
the tenths place. But when converting to fractions with the same denominator, it's clear that even though both numbers have nine as their whole 
number part, the decimal parts make a difference.But according to my calculations:- 9.8 is larger than 9.11 because \( \frac{49}{5} \) (or \( \frac{980}{100} \)) is greater than \( \frac{911}{100} \).So, this further confirms that 9.8 is bigger.Is there anything else I need to consider? Maybe looking at the hundredths place?Well, since we've already converted both numbers to have the same denominator and confirmed that 9.8 has a larger numerator in the hundredths place 
as well (\( \frac{980}{100} \) vs. \( \frac{911}{100} \)), I think we can confidently say that 9.8 is indeed bigger than 9.11.Additionally, thinking about it on a number line might help visualize this:- If you imagine the numbers from 9 to 10:- 9.8 would be closer to 10.- 9.11 would be much closer to 9.Therefore, 9.8 is further along towards 10 and thus larger than 9.11.But just to ensure there's no mistake in the fraction conversion, let me double-check:- \( \frac{49}{5} = 9.8 \) since \( 49 ÷ 5 = 9.8 \).That seems correct because 5 times 9 is 45, and 49 minus 45 leaves a remainder of 4, so \( \frac{4}{5} = 0.8 \), making it 9.8.And for \( \frac{911}{100} \):- \( 911 ÷ 100 = 9.11 \).Yes, that's accurate because dividing by 100 moves the decimal point two places to the left.So everything checks out mathematically.Is there any alternative way to think about this?Perhaps considering money as an analogy:- If you have $9.80 and $9.11:- You'd obviously have more with $9.80 than with $9.11.This real-world scenario aligns with the mathematical conclusion that 9.8 is bigger.Alternatively, think of it in terms of measurements:- If something measures 9.8 centimeters and another measures 9.11 centimeters:- The first one would be longer than the second one.All these analogies reinforce the idea that 9.8 is indeed larger than 9.11.So, after carefully analyzing through different methods—decimal comparison, fraction conversion, number line visualization, and real-life 
analogies—it becomes clear that 9.8 is bigger than 9.11.
</think>To determine which number is larger between \( 9.8 \) and \( 9.11 \), we can follow these steps:1. **Understand the Numbers**:- \( 9.8 \) is equivalent to \( 9.80 \).- Both numbers have a whole number part of 9.2. **Compare the Decimal Parts**:- The decimal part of \( 9.8 \) (or \( 9.80 \)) is 0.80.- The decimal part of \( 9.11 \) is 0.11.- Since \( 0.80 > 0.11 \), the decimal part of \( 9.8 \) is larger.3. **Conclusion**:- Therefore, \( 9.8 \) is bigger than \( 9.11 \).Final Answer:  
\[
\boxed{9.8}
\]eval count:           501 token(s)
eval duration:        12.478s
eval rate:            40.15 tokens/s>>> 

Web页面返回对话,2个数字比大小结果

本文来自互联网用户投稿,该文观点仅代表作者本人,不代表本站立场。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。如若转载,请注明出处:http://www.mzph.cn/news/896755.shtml

如若内容造成侵权/违法违规/事实不符,请联系多彩编程网进行投诉反馈email:809451989@qq.com,一经查实,立即删除!

相关文章

[AI机器人] Web-AI-Robot机器人前瞻版--比奇堡海之霸凯伦

文章目录 简述开源Web-AI-Robot 项目-比奇堡-海之霸-凯伦 技术架构效果预览 简述 本项目配合前端项目bikini_bottom_karen_ui运行&#xff0c;来源于柒杉工作室&#xff08;截止2025.2&#xff0c;目前我自己&#xff09;。 打造一个只需要在浏览器上运行的AI智能机器人&#…

250302-绿联NAS通过Docker配置SearXNG及适配Open-WebUI的yaml配置

A. 配置Docker中的代理 绿联NAS简单解决docker无法获取镜像-不用软路由 - 哔哩哔哩 B. 下载官网对应的镜像 群晖NAS用docker搭建SearXNG元搜索引擎_哔哩哔哩_bilibili C. 修改默认省略的参数&#xff0c;只配置Base_URL&#xff0c;删除其它默认的空缺项 searxng-docker/REA…

java容器 LIst、set、Map

Java容器中的List、Set、Map是核心数据结构&#xff0c;各自适用于不同的场景 一、List&#xff08;有序、可重复&#xff09; List接口代表有序集合&#xff0c;允许元素重复和通过索引访问&#xff0c;主要实现类包括&#xff1a; ArrayList 底层结构&#xff1a;动态数组…

3471. 找出最大的几近缺失整数

3471. 找出最大的几近缺失整数 class Solution:# 辅助方法&#xff0c;判断第三种情况&#xff0c;只有首位两个元素有可能为最大几近缺失数def f(self,nums,x):return -1 if x in nums else xdef largestInteger(self, nums: List[int], k: int) -> int:n len(nums)if k …

【异常错误】No module named ‘taming.modules.vqvae‘

错误&#xff1a; File "/mnt/d/Pycharm_workspace/text2image/OmniGen-version/OmniGen/latentDiffusion/ldm/models/autoencoder.py", line 6, in <module> from taming.modules.vqvae.quantize import VectorQuantizer2 as VectorQuantizer ModuleNotF…

快检查达梦库怎么了

扁鹊的弟弟来了 要求5分钟定位达梦数据库问题 #!/bin/bash## content 实例个数 告警日志 实例状态 用户连接 活动会话 锁 集群状态 服务状态 磁盘空间 cpu mem 侦听及日志 ## scope 单机、DW、DSC Linux 多实例 ## example 将脚本保存为d.sh&#xff0c;用root用执行&#…

C++20中`constexpr`的显著增强

文章目录 1. **更多标准库函数支持constexpr**2. **支持动态内存分配**3. **支持虚函数和多态**4. **支持try-catch异常处理**5. **更灵活的控制流**6. **支持std::initializer_list**7. **支持修改union活跃成员**8. **允许更多类型的非类型模板参数**总结 C20对 constexpr进…

Tomcat 乱码问题彻底解决

1. 终端乱码问题 找到 tomcat 安装目录下的 conf —> logging.properties .修改ConsoleHandler.endcoding GBK &#xff08;如果在idea中设置了UTF-8字符集&#xff0c;这里就不需要修改&#xff09; 2. CMD命令窗口设置编码 参考&#xff1a;WIN10的cmd查看编码方式&…

以太坊测试网

文章目录 什么是测试网如何使用测试网获取测试以太币 什么是测试网 测试网&#xff08;Testnet&#xff09;是一个模拟以太坊主网&#xff08;Mainnet&#xff09;行为的区块链网络。它允许开发人员和用户在不使用真实资金的情况下测试智能合约和应用程序。虽然测试网上的代币…

算法随笔_62: 买卖股票的最佳时机

上一篇:算法随笔_61:二进制求和-CSDN博客 题目描述如下: 给定一个数组 prices &#xff0c;它的第 i 个元素 prices[i] 表示一支给定股票第 i 天的价格。 你只能选择 某一天 买入这只股票&#xff0c;并选择在 未来的某一个不同的日子 卖出该股票。设计一个算法来计算你所能获…

腾讯混元文生图大模型(Hunyuan-DiT)与Stable Diffusion(SD)对比分析

腾讯混元文生图大模型&#xff08;Hunyuan-DiT&#xff09;与Stable Diffusion&#xff08;SD&#xff09;对比分析 腾讯混元文生图大模型&#xff08;Hunyuan-DiT&#xff09;与Stable Diffusion&#xff08;SD&#xff09;作为当前文生图领域的两大代表模型&#xff0c;各自…

【HTML—前端快速入门】HTML 基础

准备工作 vscode下载 百度网盘 Subline Text 下载 Sublime Text下载 百度网盘 vscode 下载 Sublime Text 是一款轻量好用的文本编辑器&#xff0c;我们在写前端代码时&#xff0c;使用 Sublime Text 打开比使用记事本打开&#xff0c;得到的代码体验更好&#xff0c;比 vscode…

基于单片机的GPS定位系统设计

1 系统硬件 1.1单片机模块 单片机的种类和型号可以说是有成百上千种&#xff0c;很多大的公司和企业都生产开发自己的单片机芯片&#xff0c;并且广泛应用于各种产品。Intel、 philips、 摩托罗拉、凌阳、宏晶等等种类繁多。大体上可以分为51系列单片机和非51系列单片机。 其…

对大模型输出的 logits 进行处理,从而控制文本的生成

对大模型输出的 logits 进行处理&#xff0c;从而控制文本的生成 flyfish 在文本生成任务中&#xff0c;模型输出的 logits 代表了每个词被选为下一个生成词的未归一化概率得分。通过对 logits 进行处理&#xff0c;可以精确地控制文本的生成 基本原理 在每一步生成过程中&…

Reids缓存穿透、缓存雪崩和缓存击穿

Redis缓存中常见的三个问题&#xff1a;缓存穿透、缓存雪崩和缓存击穿。这些问题在使用Redis作为缓存时经常遇到&#xff0c;但通过合理的策略可以有效解决。我会用简单易懂的方式来讲解&#xff0c;帮助你理解这些问题的原理和解决方案。 1. 缓存穿透 1.1 什么是缓存穿透&…

附录-Python — 包下载缓慢,配置下载镜像

1️⃣命令行配置 pip config set global.index-url http://mirrors.aliyun.com/pypi/simple/ pip config set install.trusted-host mirrors.aliyun.com 2️⃣配置文件配置 1、打开文件夹&#xff0c;输入 %APPDATA% 回车 2、打开 %APPDATA% 路径&#xff0c;并在此路径下新建…

VS 2019 免费版 下载与安装 教程说明

推荐大家直接转到第13步&#xff0c;点击链接即可下载VS2019版本 1.VS官网 2.登录账号 3.在搜索栏输入“2019” 4.点击2019这个标题 5.点击“下载” 6.选择合适的版本下载 7.打开下载文件&#xff08;若下载过程总是转圈圈&#xff0c;则换个网络下载即可&#xff09; 8.安…

介绍 torch-mlir 从 pytorch 生态到 mlir 生态

一、引言 The Torch-MLIR project provides core infrastructure for bridging the PyTorch ecosystem and the MLIR ecosystem. For example, Torch-MLIR enables PyTorch models to be lowered to a few different MLIR dialects. Torch-MLIR does not attempt to provide a…

Java并发编程之ConcurrentHashMap的原理和使用

ConcurrentHashMap(CHM)是Java为解决高并发场景下哈希表性能瓶颈而设计的线程安全容器,其核心目标在于: 线程安全‌:避免多线程操作导致的数据不一致问题‌;高吞吐量‌:通过细粒度锁和无锁化设计降低线程竞争‌;动态扩展‌:支持自动扩容与数据结构优化(如链表转红黑树…

AbMole揭秘伤口愈合:IGF-1-SP1-CD248信号通路的新发现

科学家们揭示了一条新的信号通路——IGF-1-SP1-CD248&#xff0c;这一发现为理解伤口愈合障碍提供了新的视角&#xff0c;并为未来的研究开辟了新方向。 研究背景 糖尿病患者的伤口愈合是一个长期存在的挑战。据统计&#xff0c;约15%的糖尿病患者会遭受慢性伤口的困扰&#…