langchain 初体验

最近的AI发展令人又兴奋又焦虑啊啊啊啊啊啊啊啊啊 ?????

早上看medium发来的邮件看到这样一篇文章，感觉很有意思

Create your Document ChatBot with GPT-3 and Langchain

medium的免费额度比较头疼，可以尝试把文章名字放到Google里面搜一下，然后看别的

在这篇文章了解到了langchain

然后去了它的官方文档，看到它可以比较容易地实现对话历史记忆

Adding Memory To an LLMChain

还在huggingface上面看到一个有趣的项目

GPT + WolframAlpha + Whisper

我用下面那个huggingface上面的项目试了试两个例子

(3x/sin2x) 在 x趋近于0的时候，极限是多少，用latex格式写出计算过程

下面这个回答过程很严谨，但是答案错了哈哈哈，是3/2呀

As x x x approaches 0, the limit of 3 x sin ⁡ 2 x \frac{3x}{\sin2x} sin2x3x​ is 3. To calculate this, we start by rewriting the expression as 3 x 2 sin ⁡ x cos ⁡ x \frac{3x}{2\sin x \cos x} 2sinxcosx3x​. Then, as x x x approaches 0, the sine and cosine terms both approach 0, leaving us with 3 2 cos ⁡ x \frac{3}{2\cos x} 2cosx3​. Finally, as x x x approaches 0, the cosine term approaches 1, leaving us with the limit of 3.

(3x/e^x) 在 x趋近于0的时候，极限是多少，用latex格式写出计算过程

As x x x approaches 0, the limit of 3 x e x \frac{3x}{e^x} ex3x​ is 0. To calculate this, we must first take the limit of the fraction as x x x approaches 0. This can be done by dividing both the numerator and denominator by x x x. As x x x approaches 0, the numerator will approach 0, while the denominator will approach e 0 e^0 e0, which is equal to 1. Therefore, the limit of 3 x e x \frac{3x}{e^x} ex3x​ as x x x approaches 0 is 0.

To calculate this limit, we must first consider the fraction 3 x e x \frac{3x}{e^x} ex3x​. As x x x approaches 0, the numerator will approach 0, while the denominator will approach e 0 e^0 e0, which is equal to 1. By dividing both the numerator and denominator by x x x, we can simplify the fraction to 3 e x \frac{3}{e^x} ex3​. As x x x approaches 0, the denominator will approach 1, and the limit of the fraction will be equal to the numerator, which is 3. Therefore, the limit of 3 x e x \frac{3x}{e^x} ex3x​ as x x x approaches 0 is 0.

之后试着实践下langchain的代码~

此外CSDN 也有两篇介绍

大语言模型集成工具 LangChain_kebijuelun的博客-CSDN博客

使用langchain打造自己的大型语言模型(LLMs)_-派神-的博客-CSDN博客

---

what is langchain ？ 以下是 phind.com 的回答

What is langchain

ANSWER

LangChain is a tool that provides many modules to build language model applications [0]. It is a library that can be used to create advanced use cases around large language models (LLMs) by chaining together different components [4]. Here are some of the things that LangChain can help with [3]:

Prompt management: LangChain provides a set of utilities for managing prompts, which are inputs that are passed to LLMs to generate outputs.Chains: LangChain provides a standard interface for chains, which are sequences of calls to LLMs or other utilities. This allows for more complex applications to be built.Data Augmented Generation: This involves chains that interact with external data sources to fetch data to use in the generation step. Examples of this include summarization of long pieces of text and question/answering over specific data sources.Agents: These are LLMs that make decisions about which actions to take, take that action, observe the output, and then repeat the process until done. LangChain provides a standard interface for agents, a selection of agents to choose from, and examples of end-to-end agents.Memory: LangChain provides a standard interface for memory, which allows for state to be persisted between calls of a chain or agent.Generative models: LangChain provides prompts/chains for assisting in the evaluation of generative models.

LangChain can be used to build a variety of language model applications. For example, suppose we want to build a service that generates a company name based on what the company makes. Here’s how we can use LangChain to do this [0]:

We first import the LLM wrapper and initialize it with any arguments. In this example, we probably want the outputs to be more random, so we initialize it with a high temperature.We then construct a prompt using the PromptTemplate utility. We take only the user input describing what the company does, and then format the prompt with that information.We create an LLMChain which takes user input, formats it with the prompt, and then passes the formatted response to the LLM.

LangChain also allows for more complex applications to be built. For example, we can use LangChain to build an agent that uses an LLM to determine which actions to take and in what order [0]. Here’s how we can do this:

We first load the language model we’re going to use to control the agent and some tools to use.We then initialize an agent with the tools, the language model, and the type of agent we want to use. In this example, we use the zero-shot-react-description agent.We can then test out the agent by running it with some input.

LangChain also provides a standard interface for memory, which allows for state to be persisted between calls of a chain or agent [3]. For example, we can use the ConversationChain utility to create a chain that has some concept of “memory”. By default, the ConversationChain has a simple type of memory that remembers all previous inputs/outputs and adds them to the context that is passed. We can use this to build a chatbot that remembers previous messages and uses context from that to have a better conversation [0].