hoomjac (@linhoomjac) profile picture

物外之趣

Joined June 2013
Photo by hoomjac (@linhoomjac)Photo by hoomjac (@linhoomjac)Photo by hoomjac (@linhoomjac)Photo by hoomjac (@linhoomjac)Photo by hoomjac (@linhoomjac)Photo by hoomjac (@linhoomjac)Photo by hoomjac (@linhoomjac)Photo by hoomjac (@linhoomjac)
Chat With PDF Is Hard and ChatGPT Sucks At It - Here's Why The most common GPT-4 wrapper is a "chat with a doc/pdf" app. It is one of the killer applications of AI chatbots, as reading a dense document can be tiresome -- it's much simpler to ask the LLM to parse and summarize it for you. Unfortunately, ChatGPT doesn't do a good job, especially regarding PDFS that are> 10 pages. It produces sparse generic summaries and flat-out refuses to elaborate further. One of the reasons this is a non-trivial use case is as follows. OCR - you need a really good OCR that can parse tables and images well. There is no free or commercial OCR tech that does this well. A lot of business and research PDFs have a lot of tables and images. Context - While we currently have 128K context-length LLMs, it's unclear what is deployed as part of ChatGPT. ChatGPT often throws an error if you run OCR on the paper and then feed it the paper text. I suspect a much smaller context length model serves ChatGPT requests. Quick RAG - Implementing a naive RAG that chunks the doc, embeds it, retrieves results, and then passes it to the LLM will likely do the trick, but current ChatBots don't have that feature. Highlighting Doc Sections - The ideal solution should ideally showcase parts of the document where the response is retrieved from. This makes verification super simple. Ideally, chat with PDF should have all these features. It seems like a standalone app in the app store can still make decent revenue if it does a good job of all these features. That being said, I don't think this is a venture-backed start-up. More like a 1-2 person mom-pop thing that can be a good lifestyle business. TLDR: Doing something as simple as "Chat with PDF" is non-trivial and hard to do well
61
96
21
496
186,103
🧩No GPU but wanna create your own LLM on laptop? 🎁Here is a gift for you: QLoRA on CPU, making LLM fine-tuning on client CPU possible! Just give a try. 📔Blog: medium.com/@NeuralCompressor… Kudos to ITREX team! 🎯Code: github.com/intel/intel-exten… #IAmIntel #intelai @intel @huggingface
5
148
5
644
71,097
hoomjac retweeted
最近看到最棒的一篇技术文章: Draggable objects redblobgames.com/making-of/d… 在前端开发中,拖动(Drag)一个图层是非常常见的需求,我相信大家大部分都能写个皮毛,但是这篇文章循序渐进,从最简单的拖动,到兼容手机和 PC,再到各种 edge cases 的修复,最后再到抽象成一个 vue 的组件,我从中学习到非常多。 每一步都配有 demo,讲解的非常详细,是一篇非常好的交互式文章,如果想把拖动完全弄懂,这篇文章就足够了。
15
184
2
728
80,901
hoomjac retweeted
Mr. Ranedeer是我最近看到过的最强大的ChatGPT的Prompt!也适用于GPT的API,但都需要GPT-4。它可以让GPT成为你的导师,指导你学习! 它的用法很简单,在它的项目首页🔗github.com/JushBJJ/Mr.-Raned…🔗有详细说明。 拷贝它的JSON文件内容或者Markdown文件内容,推荐Markdown文件,体积最小。 它有若干选项: - Depth:学习的深度或难度,1代表小学生水平,10代表博士水平 - Learning Styles:学习风格,有: 感知型,视觉型*(需要插件),归纳型,积极型,顺序型,直觉型,语言型,演绎型,反思型,全局型 - Communication:沟通方式,有:随机的、正式的、教科书的、外行的、讲故事的、苏格拉底的、幽默的 - Tone Styles:语气风格,辩论、鼓励、中立、告知、友好 - Reasoning Frameworks:推理框架,演绎法、归纳法、归纳法、类比法、随意法 - Language:语言,默认英语,你可以改成GPT-4支持的任意语言。 你可以略作修改或者直接复制粘贴到ChatGPT,然后你就可以用命令跟GPT对话让它指导你学习了。 它有以下命令: /test :对你当前学习的结果进行测试 /config : 更新AI导师的配置选项 /plan :根据你的喜好,创建一个学习计划,比如你要学英语 /search : 搜索特定信息,需要插件支持 /start : 开始学习 /continue : 如果输出中断或没有结束,可以通过这个命令继续让GPT输出 /self-eval : 让AI导师评估自己的课程 /language : 改变AI导师输出的语言,比如简体中文 /visualize : 使用插件(例如Wolfram)来实现内容的可视化 比如一个参考指令如下: 给我一个中国学术高中生的随机配置 /language 简体中文 /plan 我想学习高级英语语法 测试下来非常强大,唯一的问题是初始的Tokens占用太多,会话消息一多就要重新开始了。 项目地址:github.com/JushBJJ/Mr.-Raned…
89
457
14
1,345
224,374
有些库的用法,看官方文档集成 30 分钟,从开源库里 copy 别人的用法,30秒。优秀的使用范例胜过千言万语😂
45
每当这个时刻,我就很想念 React 的 Immutable Data 的简洁。虽然可能多渲染几次,但数据就是数据,no wrap,no shit。
1
1
28
hoomjac retweeted
Vector databases & embeddings are the current hot thing in AI. Pinecone, a vector DB company, just raised $100M at ~1b valuation. Shopify, Brex, Hubspot and others use them for their AI apps But what are they, how do they work and why are they SO crucial in AI? Let's find out
112
392
56
2,969
1,640,507
medium.com/refactoring-ui/7-… Tailwind 团队写的 7 个界面设计技巧,非常实用,他们还写了《Refactoring UI》这本书,值得阅读。
62
1
258
21,732
Neovim + lazygit 真是太好用啦,无缝丝滑,生产力拉满。 1. 安装 lazygit(github.com/jesseduffield/laz…) 2.安装 vim-floaterm 插件(github.com/voldikss/vim-floa…) 3.设置快捷键,让 lazygit 在任何界面以弹窗形式弹出来:`keymap.set("n", "<leader>lg", ":FloatermNew lazygit<CR>")` 强烈推荐👍
1
913
很有意思的分享👍
1/ 如何通过物理手段获得一组秘密安全的助记词? 我最近一直想做一个关于如何获得秘密安全“助记词(mnemonics)”方面的攻略,正好赶上前几天Do Kwon的段子来了,我觉得有必要给大家分享一下。 一共需要5步: 1.找到助记词列表 2.找点骰子 3.摇掷骰子-转换进制 4.穷举校验位 5.备份
24
一个平时脾气非常nice的同事,午休时候打电话辅导小孩做作业,越教越气,陷入癫狂状态,直接爆粗口…让我们看到了他的另外一面。 “你他妈的告诉我相邻3位数是什么意思!” “我不理解你为什么这个都没法理解!” 不知道下午和他对接业务的同事会不会怕怕的😂
1
280
hoomjac retweeted
人太多了,可能会出现“服务器繁忙”的提示,大家可以自己部署到 vercel 。 代码已开源:github.com/lvwzhen/law-cn-ai
10
31
2
133
41,542