{ "text": " No, wait, let's go adopt the first horse of your life. Wait, wrong one, it's this one. Hey, today I'll guide you through getting started with Hermes Agent and show you step-by-step how to build a cyber tutor that understands you better every day, especially for students, so make sure to like, save, and follow. Hermes is an agent known for its automation capabilities. Compared to OpenCloud, it not only has automated skills but is also easier to use, has fewer bugs, is safer and more stable, and is more cost-effective, which is why its reputation is soaring. At the same time, it's better suited for connecting to Fichu for remote use compared to coding agents, and with its long-term memory, it's the perfect 24-7 personal assistant. It is particularly well-suited for long-term learning. So, we will use the example of a cyber tutor that continuously teaches you AI to show you how to set up Hermes. Also, a big thanks to Dooyin Technology and NVIDIA for inviting us to join the AI Creator Camp. After watching this, if you participate, you might even win an autograph from Jensen Huang. Okay, the first step, downloading and installing, is very simple. Just head to the official Hermes website, download the desktop version, install it, and open it. After opening, let it finish its initial setup, then you can select the model you want to use here, like DeepSeek, and simply enter your DeepSeek API key. If you have a membership or a monthly subscription, it's even more convenient. You can just log in and start using your member credits. For those with a dedicated graphics card, we can also add a local model later as an auxiliary model to save even more tokens. And the biggest benefit of a local model is that sensitive, private data can be processed directly on your own computer, and it even works well offline. I'd also like to add that you really don't need to buy a separate computer or rent a cloud server for an agent like Hermes. I've tried it all, and it's only truly useful when deployed on your own machine, right within your daily work environment. I've deployed it on my personal Windows PC with an NVIDIA 5070 card, allowing me to pair Hermes with a local model to handle high privacy tasks, and I can run 24-7 automated tasks without a worry. I'll show you how to deploy it in a moment, it's super simple. Okay, once you're in, let's first open the settings to select the appearance, then change the language here to simplified Chinese. Head back to the page, and that's it. Actually, Hermes is ready to use now, so let's give it a try. Introduce yourself and, by the way, do something you think you can do for me to demonstrate your overall capabilities. As you can see, it directly chose to give our computer a health checkup, including memory, disk, and security status, and even provided a complete report. It seems my computer isn't in the best health. This is its most basic way of working, working directly on your computer and delivering results straight to you. It's fantastic. Now, if you want to connect to IM channels like Lark or WeChat, you can configure that here in the messaging platform section. However, I recommend having Hermes guide you through the connection process via conversation. For example, ask it to connect to Lark, then just follow its prompts to click and confirm permissions. Okay, so now that it's set up, how can you use it more effectively? The answer is multi-agent. Hermes can have different agents with different identities, memories, and purposes, and they are completely isolated from one another. For example, a life assistant, a work assistant, and a study tutor. And the biggest advantage is that if one fails, you can have another one fix it. Additionally, using multiple models is also necessary. Hermes allows you to set up auxiliary models to handle menial tasks with smaller, especially free local models, which is more cost-effective. You can even set up an agent that only uses local models to handle sensitive, private tasks. Let's get right to it and add a local model. It's very easy to get started. We go directly to the Olama website, download the Olama desktop client, and then just open the desktop app. Here, we can search for a model that fits your VRAM. For example, since I have a 12GB RTX 4070, I'll go with a 9BQEN 2.5. Then, just start a chat with it, and it will download. Once it's downloaded, let's test the speed by asking it who, ChuG2046, is. Wow, look at that token generation speed, and it even invoked search tools in the process. Since NVIDIA GPUs have inference acceleration in Olama, 9B models basically generate tokens instantly. If you need to perform batch tasks, the speed is definitely much faster than cloud-based models. Let's go back to Hermes. We can directly have Hermes add a local Olama model provider to MIS for us, and it will automatically add it into Hermes by modifying the configuration file. Then, when we open the settings and look at the models, our Olama will be right there. As for the auxiliary models below, such as for web extraction, summarization, or context, we can now set them all to be handled by QEN 3.5. You can even download a local vision model in Olama specifically for image analysis. The deployment is simple enough, and your graphics card should be up to the task. Okay, so this main agent can serve as your permanent daily assistant. You can just tell it your identity or habits through chat, and it will remember them. Now, let's customize our second agent, the Cyber Mentor. This Cyber Mentor can create a long-term study plan based on your learning goals, send you daily materials through scheduled tasks, guide your practice, keep records, and even provide a website to track your progress, making everything clear until you finish it all. Let's create a new agent by clicking the plus sign next to the little house icon. Now, let's give the new agent a name, like, AI study, and for the clone source, since we want a brand new agent, we'll just select, blank. As for the SMD, which is its persona, we won't write it yet. We'll let the agent write it itself later, so just select, create. Back here, we can now switch between our two agents at any time. We'll switch to the study agent, send it the document I prepared, and tell it, read this document and transform yourself into the AI learning mentor described here. I have already prepared this requirements document for you all as well. Once you look at it, it immediately lists out a whole set of plans. There's a lot of work to be done as it begins its self-transformation. During this initialization process, it will start interviewing us, like asking what subject we want to learn. We might want to learn coding, so we tell it we want to learn web coding. Then it asks, what is your purpose for learning? You then provide your answer, followed by, what is your current learning background? And, what teaching style do you prefer? You can specify when you want to study each day or when you'd like a weekly summary. Once you provide your preferences, it only takes a short while for the system to set everything up for you. This system creates a folder locally on your machine, containing your daily study logs, your mistakes, and a dashboard to track your overall progress. Inside, you'll find a complete study roadmap, a record of your errors, and a weekly review. Every day moving forward, it will send you a prompt at your chosen time to remind you to study. These daily study sessions are scheduled as automated tasks within Hermes. We can return to the schedule tab in Hermes to view and manage all of these recurring tasks. So, how well does this tutor teach? Let's take a look at this web coding course. First, it explains the theory, breaking down the seven key components of a product, and even creates a real-world workflow diagram for us. After covering the concepts, it will ask questions to verify that we have truly grasped the material. It also requires us to complete exercises and submit the results, which it then grades for us. If we don't finish a lesson on a given day, it will automatically adjust tomorrow's plan based on our actual progress. More importantly, its teaching style can evolve. That's because its instructional method exists in the form of skills. Suppose that during your studies, you find the teaching isn't quite what you want. Perhaps it isn't visual enough. If you want it to include a flowchart every time and use real-life analogies, you can provide that feedback directly and ask it to improve its teaching. Look, it shows here that the teaching method has been updated. When we start the next lesson, the effect is already different. Oh, look, there's an analogy here, and it's included a diagram as well. Over time, it not only keeps track of your learning progress, but it also becomes a teacher perfectly tailored to you. Alright, if you search for AI Creation Camp and participate now, you have a chance to win an RTX 5080 Founders Edition graphics card signed by NVIDIA's Jensen Huang. For those in the RTX AI PC community, I highly recommend trying the combination of a local agent and a local model. It saves tokens, protects privacy, and is better for long-term use. Remember to like, save, and follow on Dooyan. See you next time.", "segments": [ { "start": 0.0, "end": 0.8480000000000001, "text": "你好啊" }, { "start": 0.8480000000000001, "end": 4.24, "text": "来领养你人生的第一匹马吧" }, { "start": 5.28, "end": 7.48, "text": "不对 是这个" }, { "start": 9.54, "end": 10.310312499999998, "text": "嘿嘿 今天呢" }, { "start": 10.310312499999998, "end": 13.545624999999998, "text": "我会带大家轻松地上手Hermes Agent" }, { "start": 13.545624999999998, "end": 15.394375, "text": "并且会手把手带你打造一个" }, { "start": 15.394375, "end": 16.935, "text": "越来越懂你的赛博老师" }, { "start": 16.935, "end": 19.0475, "text": "尤其适合需要学习的同学们" }, { "start": 19.0475, "end": 20.455833333333334, "text": "那点好收藏关注赞" }, { "start": 20.455833333333334, "end": 21.16, "text": "我们上马" }, { "start": 23.0, "end": 24.37818181818182, "text": "Let's Go" }, { "start": 24.37818181818182, "end": 25.042424242424243, "text": "那Hermes呢" }, { "start": 25.042424242424243, "end": 26.62, "text": "是以自进化能力出圈的龙虾类的Agent" }, { "start": 26.62, "end": 30.626, "text": "比起Openclaw 它不仅多了自进化的技能" }, { "start": 30.626, "end": 33.02969230769231, "text": "还更容易上手 Bug也更少" }, { "start": 33.02969230769231, "end": 35.834, "text": "也更安全稳定 还更省Token" }, { "start": 35.834, "end": 39.168499999999995, "text": "所以它最近口碑持续在飙升" }, { "start": 39.168499999999995, "end": 42.8, "text": "那同时呢" }, { "start": 42.8, "end": 44.940571428571424, "text": "它又会比编程类的Agent更适合" }, { "start": 44.940571428571424, "end": 46.546, "text": "去连接微信飞书来远程使用" }, { "start": 46.546, "end": 48.095862068965516, "text": "再加上它有长期记忆" }, { "start": 48.095862068965516, "end": 49.817931034482754, "text": "所以就更适合作为一个" }, { "start": 49.817931034482754, "end": 51.54, "text": "7*24小时的日常助手" }, { "start": 51.54, "end": 56.66, "text": "尤其适合需要长周期的学习" }, { "start": 57.54, "end": 60.51, "text": "所以我们会以一个持续教你学AI的" }, { "start": 60.51, "end": 63.48, "text": "赛博导师为例 带你拉通Hermes" }, { "start": 64.04, "end": 66.84583333333333, "text": "也非常感谢抖音科技和英伟达" }, { "start": 66.84583333333333, "end": 69.22, "text": "邀请我们加入AI创造营" }, { "start": 69.76, "end": 71.696, "text": "大家等会看完之后参与这个活动" }, { "start": 71.696, "end": 75.49303225806452, "text": "还有机会获得黄仁勋亲笔签名的显卡哦" }, { "start": 75.49303225806452, "end": 77.72658064516129, "text": "OK 那第一步 下载安装" }, { "start": 77.72658064516129, "end": 78.62, "text": "非常简单" }, { "start": 79.66, "end": 81.898, "text": "我们去到Hermes的官网" }, { "start": 81.898, "end": 83.369, "text": "直接下载它的桌面端" }, { "start": 83.369, "end": 84.84, "text": "然后安装好 给它打开" }, { "start": 86.44, "end": 89.378, "text": "那打开之后 等它再自己装一会" }, { "start": 89.378, "end": 91.18763265306123, "text": "在这里就可以选择我们去用的模型了" }, { "start": 91.18763265306123, "end": 92.43175510204082, "text": "比如说DeepSeek" }, { "start": 92.43175510204082, "end": 94.92, "text": "那我们就填上这个DeepSeek的API密钥" }, { "start": 94.92, "end": 98.764, "text": "那如果你有这些会员 或者是包月" }, { "start": 98.764, "end": 101.69257142857143, "text": "那就更方便了 你可以直接点击登录" }, { "start": 101.69257142857143, "end": 103.78, "text": "你就可以使用你的会员额度了" }, { "start": 104.86, "end": 106.24947368421053, "text": "那有显卡的朋友呢" }, { "start": 106.24947368421053, "end": 109.02842105263157, "text": "一会我们还可以额外加一个本地模型" }, { "start": 109.02842105263157, "end": 111.46, "text": "作为辅助模型 会更省Token" }, { "start": 112.84, "end": 115.405, "text": "而且本地模型最大的好处 就是隐私" }, { "start": 115.405, "end": 118.654, "text": "敏感资料也可以直接留在自己的电脑上处理" }, { "start": 118.654, "end": 120.24, "text": "并且离线断网了也可以用" }, { "start": 120.24, "end": 121.67307692307692, "text": "哦 另外我想补充一下" }, { "start": 121.67307692307692, "end": 124.53923076923077, "text": "其实大家用Hermes这种Agent" }, { "start": 124.53923076923077, "end": 125.9723076923077, "text": "真的不用去专门给它" }, { "start": 125.9723076923077, "end": 128.52, "text": "买一个单独的电脑 或者是租云服务器" }, { "start": 131.52, "end": 132.43076923076924, "text": "我全部都试过" }, { "start": 132.43076923076924, "end": 134.8594871794872, "text": "但是只有当它部署在我自己的电脑上" }, { "start": 134.8594871794872, "end": 136.52923076923076, "text": "在你平时工作的环境当中" }, { "start": 136.52923076923076, "end": 137.44, "text": "它才会更有用" }, { "start": 137.44, "end": 140.47354545454544, "text": "那我呢就是部署在了自用的Windows电脑" }, { "start": 140.47354545454544, "end": 142.207, "text": "加上英伟达的5070显卡" }, { "start": 142.207, "end": 145.35473333333334, "text": "还能给Hermes配一个本地模型" }, { "start": 145.35473333333334, "end": 148.30573333333334, "text": "专门来解决一些隐私要求高的任务" }, { "start": 148.30573333333334, "end": 151.06, "text": "7*24小时跑定时任务都不心疼" }, { "start": 151.72, "end": 154.9, "text": "那等会也会带大家部署 超级简单" }, { "start": 155.92, "end": 157.2928695652174, "text": "OK 那进来之后呢" }, { "start": 157.2928695652174, "end": 159.867, "text": "我们先打开这边的设置 来选择外观" }, { "start": 159.867, "end": 162.6, "text": "然后把这里 语言可以改成简体中文" }, { "start": 162.6, "end": 165.84, "text": "那再回到页面 这就搞定了" }, { "start": 166.54, "end": 169.08307692307693, "text": "其实Hermes大家这样就已经可以用了" }, { "start": 169.08307692307693, "end": 170.02, "text": "我们可以试一下" }, { "start": 171.44, "end": 172.536, "text": "介绍一下你自己" }, { "start": 172.536, "end": 174.88457142857143, "text": "顺便呢做一件觉得可以为我做的事" }, { "start": 174.88457142857143, "end": 176.92, "text": "以表现一下她自己的综合能力" }, { "start": 177.92, "end": 180.3999090909091, "text": "你看看 她直接选择给我们的电脑" }, { "start": 180.3999090909091, "end": 181.817, "text": "做了一个健康体检" }, { "start": 181.817, "end": 183.74325, "text": "这些内存磁盘安全状态" }, { "start": 183.74325, "end": 186.44, "text": "甚至给我们出了一个完整的报告" }, { "start": 186.8, "end": 188.45117857142858, "text": "只不过我的电脑不太健康" }, { "start": 188.45117857142858, "end": 191.003, "text": "那这呢就是它最基本的一个工作方式了" }, { "start": 191.003, "end": 193.257, "text": "直接在你的电脑上工作" }, { "start": 193.257, "end": 197.2, "text": "直接给你交付成果 非常棒" }, { "start": 197.74, "end": 199.0304285714286, "text": "那如果大家想连接上" }, { "start": 199.0304285714286, "end": 200.751, "text": "飞书微信之类的IM渠道呢" }, { "start": 200.751, "end": 203.88, "text": "在消息平台这里就可以配" }, { "start": 204.9, "end": 206.95846153846153, "text": "不过我更建议大家通过对话" }, { "start": 206.95846153846153, "end": 209.36, "text": "直接让Hermes带你去连接" }, { "start": 209.36, "end": 211.4951724137931, "text": "比如说让它连接飞书" }, { "start": 211.4951724137931, "end": 214.5793103448276, "text": "然后你就跟着它的引导去点击" }, { "start": 214.5793103448276, "end": 216.24, "text": "去确认权限就行" }, { "start": 217.44, "end": 219.54947368421054, "text": "OK 那装好了Hermes" }, { "start": 219.54947368421054, "end": 220.78, "text": "怎么用会更爽呢" }, { "start": 221.74, "end": 223.46, "text": "答案是多Agent" }, { "start": 224.74, "end": 226.91875000000002, "text": "Hermes呢可以有不同的Agent" }, { "start": 226.91875000000002, "end": 227.645, "text": "配不同的身份" }, { "start": 227.645, "end": 229.15329411764705, "text": "有不同的技艺和用途" }, { "start": 229.15329411764705, "end": 230.494, "text": "而且是完全隔离的" }, { "start": 230.494, "end": 234.09619999999998, "text": "比如一个生活助理 一个工作助理" }, { "start": 234.09619999999998, "end": 235.64, "text": "一个学习导师" }, { "start": 236.92, "end": 238.1104, "text": "而且最大的好处是" }, { "start": 238.1104, "end": 240.64, "text": "其中一个挂了 你还能让另外的人去修它" }, { "start": 241.32, "end": 244.34, "text": "另外多个模型也是有必要的" }, { "start": 244.6, "end": 247.3567391304348, "text": "Hermes呢可以设置辅助模型" }, { "start": 247.3567391304348, "end": 248.827, "text": "把杂活都给小模型" }, { "start": 248.827, "end": 250.69873684210526, "text": "尤其是免费的本地模型干" }, { "start": 250.69873684210526, "end": 252.06, "text": "会更省Token" }, { "start": 252.44, "end": 253.90875, "text": "甚至你还可以设置一个" }, { "start": 253.90875, "end": 255.67125, "text": "只用本地模型的Agent" }, { "start": 255.67125, "end": 257.14, "text": "专门处理一些隐私任务" }, { "start": 257.76, "end": 258.32875, "text": "那说干就干" }, { "start": 258.32875, "end": 259.58, "text": "我们给它加一个本地模型" }, { "start": 259.58, "end": 260.529, "text": "非常好上手" }, { "start": 260.529, "end": 264.576, "text": "我们直接来到Ollama的官网" }, { "start": 264.576, "end": 266.9724444444445, "text": "然后下载一个Ollama的桌面端" }, { "start": 266.9724444444445, "end": 268.62, "text": "然后呢就是打开桌面端了" }, { "start": 270.1, "end": 271.82800000000003, "text": "那在这里我们可以搜索一个" }, { "start": 271.82800000000003, "end": 272.98, "text": "适合你显存的模型" }, { "start": 273.44, "end": 276.6164705882353, "text": "比如说我是12G的这个RTX5070" }, { "start": 276.6164705882353, "end": 279.44, "text": "那我就来一个9B的Qwen3.5吧" }, { "start": 279.66, "end": 282.16, "text": "然后直接跟它对个话 它就可以下载了" }, { "start": 282.86, "end": 285.5328, "text": "那下好之后给大家试试速度啊" }, { "start": 285.5328, "end": 288.0, "text": "让它查查秋芝2046是谁" }, { "start": 288.0, "end": 290.361, "text": "哇 这个出Token的速度" }, { "start": 290.361, "end": 292.66, "text": "中间还有搜索工具的调用" }, { "start": 292.9, "end": 294.28875, "text": "那因为英伟达的显卡呢" }, { "start": 294.28875, "end": 296.233, "text": "是有Ollama的推理加速的" }, { "start": 296.233, "end": 299.08, "text": "9B的模型基本上等于秒出Token" }, { "start": 299.6, "end": 301.276, "text": "如果要做批量的任务" }, { "start": 301.276, "end": 304.28, "text": "速度肯定比云端模型要快很多" }, { "start": 304.28, "end": 308.58, "text": "那我们再回到Hermes" }, { "start": 308.88, "end": 310.084775, "text": "我们可以直接让Hermes" }, { "start": 310.084775, "end": 311.28954999999996, "text": "帮我给Hermes添加一个" }, { "start": 311.28954999999996, "end": 312.587, "text": "本地的Ollama模型提供商" }, { "start": 312.587, "end": 314.5601764705882, "text": "那它就可以通过修改配置文件的方式" }, { "start": 314.5601764705882, "end": 316.78, "text": "自己帮我们加到这个Hermes里面了" }, { "start": 316.78, "end": 321.91, "text": "然后我们再打开设置 再看到模型" }, { "start": 321.91, "end": 324.76, "text": "那这里呢 就有刚才我们的Ollama了" }, { "start": 326.36, "end": 328.10521739130434, "text": "那下面这个辅助模型" }, { "start": 328.10521739130434, "end": 331.40173913043475, "text": "比如说网页提取压缩上下文这一趴杂事" }, { "start": 331.40173913043475, "end": 333.34086956521736, "text": "我们就都可以把它改成" }, { "start": 333.34086956521736, "end": 335.28, "text": "用Qwen3.5来做了" }, { "start": 335.28, "end": 337.25166666666667, "text": "甚至你还可以再Ollama" }, { "start": 337.25166666666667, "end": 339.22333333333336, "text": "再去下载一个本地的图片模型" }, { "start": 339.22333333333336, "end": 340.74, "text": "来专门给它做图片分析" }, { "start": 341.38, "end": 344.80146341463416, "text": "反正部署是够简单的 显卡也够用" }, { "start": 344.80146341463416, "end": 347.7341463414634, "text": "OK 那这个主Agent呢" }, { "start": 347.7341463414634, "end": 351.4, "text": "大家可以作为一个常驻的日常助理" }, { "start": 351.86, "end": 353.77, "text": "想让它记住你什么身份什么习惯" }, { "start": 353.77, "end": 355.68, "text": "都可以通过聊天去告诉它就行了" }, { "start": 358.48, "end": 360.08, "text": "那来我们去定制我们的" }, { "start": 360.08, "end": 362.0, "text": "第二个Agent 赛博导师" }, { "start": 362.0, "end": 363.1461935483871, "text": "这个赛博导师呢" }, { "start": 363.1461935483871, "end": 364.9473548387097, "text": "它可以根据你的学习目标" }, { "start": 364.9473548387097, "end": 367.076, "text": "帮你制定一个长期的学习计划" }, { "start": 367.076, "end": 368.9145263157895, "text": "然后通过定时的任务" }, { "start": 368.9145263157895, "end": 370.75305263157895, "text": "每天给你发学习资料" }, { "start": 370.75305263157895, "end": 372.7958596491228, "text": "然后带你练习 帮你记录" }, { "start": 372.7958596491228, "end": 376.064350877193, "text": "还会给你一个这样的学习进度的网站" }, { "start": 376.064350877193, "end": 378.72, "text": "一目了然 直到你把这些都学完" }, { "start": 382.12, "end": 383.8593103448276, "text": "那我们来新建一个Agent" }, { "start": 383.8593103448276, "end": 386.0, "text": "点击这个小房子旁边的加号就可以了" }, { "start": 386.0, "end": 388.27643750000004, "text": "那我们呢 再给新的Agent取一个名字" }, { "start": 388.27643750000004, "end": 390.047, "text": "比如说我们这个AI study" }, { "start": 390.047, "end": 391.5747647058824, "text": "然后克隆来源这里" }, { "start": 391.5747647058824, "end": 394.63029411764705, "text": "我们希望它是一个全新的Agent" }, { "start": 394.63029411764705, "end": 396.54, "text": "所以我们直接选择空白" }, { "start": 398.86, "end": 401.366, "text": "然后soul.md 这是它的人格" }, { "start": 401.366, "end": 402.44, "text": "我们也先不写" }, { "start": 402.82, "end": 404.8257142857143, "text": "等会呢让Agent自己去写" }, { "start": 404.8257142857143, "end": 406.06, "text": "我们选择创建即可" }, { "start": 406.88, "end": 407.65076923076924, "text": "那再回到这里" }, { "start": 407.65076923076924, "end": 410.22, "text": "我们就可以随时切换我们这两个Agent了" }, { "start": 410.22, "end": 412.07203571428573, "text": "我们给它切到学习Agent" }, { "start": 412.07203571428573, "end": 414.209, "text": "直接把我准备好的这个文档发给它" }, { "start": 414.209, "end": 416.3986428571429, "text": "然后跟它说 读这份文档" }, { "start": 416.3986428571429, "end": 419.0262142857143, "text": "把你自已改造成这里面说的" }, { "start": 419.0262142857143, "end": 420.34, "text": "AI学习导师" }, { "start": 421.06, "end": 423.94, "text": "那这个需求文档我也给大家准备好了" }, { "start": 424.12, "end": 426.88, "text": "你看看 它直接一下子列出来了一整套的计划" }, { "start": 427.3, "end": 428.61, "text": "有非常多的活要去干" }, { "start": 428.61, "end": 429.92, "text": "它要开始自我改造了" }, { "start": 430.24, "end": 431.954, "text": "那这个初始化的过程中" }, { "start": 431.954, "end": 433.20552941176476, "text": "它就会开始采访我们" }, { "start": 433.20552941176476, "end": 434.31800000000004, "text": "比如想学什么主题" }, { "start": 434.31800000000004, "end": 436.862, "text": "那我们可能想学Vibe Coding" }, { "start": 436.862, "end": 438.54, "text": "就告诉它想学Vibe Coding" }, { "start": 438.54, "end": 441.02, "text": "哎 那你学习的目的是什么呢" }, { "start": 441.4, "end": 443.32, "text": "那你也给出你的答案" }, { "start": 443.32, "end": 445.24, "text": "那你的学习基础怎么样呢" }, { "start": 445.78, "end": 447.7, "text": "偏向什么教学风格呢" }, { "start": 448.54, "end": 450.856, "text": "你希望每天几点学习" }, { "start": 450.856, "end": 451.3475555555556, "text": "周几给你总结呢" }, { "start": 451.3475555555556, "end": 452.12, "text": "都给出你自己相应的需求" }, { "start": 453.74, "end": 455.16857142857145, "text": "那等答完这些之后" }, { "start": 455.16857142857145, "end": 456.06142857142856, "text": "再过一小会" }, { "start": 456.06142857142856, "end": 458.74, "text": "它自己就把整套的系统给搭起来了" }, { "start": 458.74, "end": 459.8518, "text": "那这个系统呢" }, { "start": 459.8518, "end": 462.446, "text": "会在你本地建一个这样的文件夹" }, { "start": 462.446, "end": 464.88494444444444, "text": "这里面会有你每日的学习记录" }, { "start": 464.88494444444444, "end": 467.6991111111111, "text": "你的错题 以及这个主要给我们看的" }, { "start": 467.6991111111111, "end": 469.2, "text": "一个学习进展网站" }, { "start": 471.68, "end": 474.205, "text": "那里面有我们完整的学习路线图" }, { "start": 474.205, "end": 476.6, "text": "有错题本和每周的复盘" }, { "start": 476.6, "end": 478.28, "text": "那往后的每一天" }, { "start": 478.28, "end": 481.33454545454543, "text": "它也都会按我们选的时间准点发消息" }, { "start": 481.33454545454543, "end": 482.48, "text": "提醒你学习了" }, { "start": 484.64, "end": 485.76, "text": "那这个每天的学习" }, { "start": 485.76, "end": 488.42, "text": "就是Hermes自己设定了一个定时任务" }, { "start": 488.42, "end": 492.5366666666667, "text": "我们回到Hermes 打开这里的这个排程" }, { "start": 492.5366666666667, "end": 496.22, "text": "就可以看到并且管理所有的定时任务了" }, { "start": 498.32, "end": 500.48, "text": "那这个导师的课讲的怎么样呢" }, { "start": 501.08, "end": 503.16, "text": "我们来看一下这个Vibe Coding课程啊" }, { "start": 504.52, "end": 506.147, "text": "首先它会给你先讲理论" }, { "start": 506.147, "end": 508.96349999999995, "text": "告诉我们这是一节说产品的七个部件" }, { "start": 508.96349999999995, "end": 511.78, "text": "甚至还给我们画了一个真实的链路图" }, { "start": 511.78, "end": 515.033, "text": "那说完知识点 它还会提问" }, { "start": 515.033, "end": 517.54, "text": "来检验我们是否真的学会了" }, { "start": 518.1, "end": 519.962487804878, "text": "后续呢还会要求我们练习" }, { "start": 519.962487804878, "end": 523.3701951219513, "text": "并且把练习结果发给它 它给我们批改作业" }, { "start": 523.3701951219513, "end": 525.402, "text": "如果当天我们没有完成课程" }, { "start": 525.402, "end": 526.565, "text": "它也会根据实际的情况" }, { "start": 526.565, "end": 527.728, "text": "自己去调整明天的计划" }, { "start": 527.728, "end": 531.2282105263157, "text": "更重要的是 它的教法也可以进化" }, { "start": 531.2282105263157, "end": 533.9826315789473, "text": "因为它的教学方式呢" }, { "start": 533.9826315789473, "end": 536.94, "text": "是以技能Skill的方式存在的" }, { "start": 536.94, "end": 539.243, "text": "假设我们在学习的过程中" }, { "start": 539.243, "end": 542.42, "text": "你觉得它教的不太符合你的心意" }, { "start": 542.8, "end": 544.18, "text": "比如不够视觉化" }, { "start": 546.9, "end": 549.459, "text": "你希望它每一次都要有流程图演示" }, { "start": 549.459, "end": 551.27134375, "text": "还要用生活中的事物来给你做比喻" }, { "start": 551.27134375, "end": 553.008, "text": "那我们就可以直接把这个给它反馈过去" }, { "start": 553.008, "end": 554.4, "text": "并且要求它改进这个教学" }, { "start": 556.68, "end": 558.8584313725489, "text": "你看看 那它这里就显示了" }, { "start": 558.8584313725489, "end": 562.5529411764705, "text": "它更新了这个教学Skill" }, { "start": 562.5529411764705, "end": 565.3254901960784, "text": "我们再上下一课 效果就已经变了" }, { "start": 565.3254901960784, "end": 567.9, "text": "哎 这里有比喻了 然后也有图了" }, { "start": 568.12, "end": 568.9325, "text": "那长此以往" }, { "start": 568.9325, "end": 571.045, "text": "它不仅随时知道你的学习进度" }, { "start": 571.045, "end": 573.74, "text": "也会变成越来越适合你的老师" }, { "start": 574.06, "end": 577.44, "text": "好了 大家现在去搜索AI创造营参与活动的话" }, { "start": 577.7, "end": 580.5241379310345, "text": "还有机会获得英伟达CEO黄仁勋" }, { "start": 580.5241379310345, "end": 583.16, "text": "亲笔签名的5080公版显卡哦" }, { "start": 583.42, "end": 585.463, "text": "有RTX AI PC的同学" }, { "start": 585.463, "end": 587.096875, "text": "真的建议大家试一下" }, { "start": 587.096875, "end": 589.82, "text": "本地Agent加本地模型的组合" }, { "start": 589.82, "end": 594.1, "text": "省Token 隐私 也更适合长期运行" }, { "start": 594.42, "end": 596.814, "text": "也记得在抖音精选点赞收藏关注" }, { "start": 596.814, "end": 598.06, "text": "我们下次见了" } ], "language": "zh", "refine_applied": true }