Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/nobleknightt/llm-chatbot
Neo4j-backed Chatbot using Mistral AI, NextJS, TypeScript, TailwindCSS
https://github.com/nobleknightt/llm-chatbot
chatbot mistralai neo4j nextjs tailwindcss typescript
Last synced: 4 days ago
JSON representation
Neo4j-backed Chatbot using Mistral AI, NextJS, TypeScript, TailwindCSS
- Host: GitHub
- URL: https://github.com/nobleknightt/llm-chatbot
- Owner: nobleknightt
- Created: 2024-08-06T14:09:52.000Z (5 months ago)
- Default Branch: main
- Last Pushed: 2024-08-07T11:18:08.000Z (5 months ago)
- Last Synced: 2024-11-03T08:41:58.352Z (about 2 months ago)
- Topics: chatbot, mistralai, neo4j, nextjs, tailwindcss, typescript
- Language: TypeScript
- Homepage: https://graphacademy.neo4j.com/courses/llm-chatbot-typescript
- Size: 275 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Build a Neo4j-backed Chatbot using TypeScript
> [!NOTE]
> I have used [Mistral AI](https://mistral.ai/) (instead of OpenAI) which provides free trail for 15 days.
> During migration from OpenAI to MistralAI, I encountered some issues which I have documented below along with there possible fixes.
This repository accompanies the [Build a Neo4j-backed Chatbot using TypeScript](https://graphacademy.neo4j.com/courses/llm-chatbot-typescript/?ref=github) course on [Neo4j GraphAcademy](https://graphacademy.neo4j.com/?ref=github).
It was originally a [Next.js](https://nextjs.org/) project bootstrapped with [`create-next-app`](https://github.com/vercel/next.js/tree/canary/packages/create-next-app). We added:
* [TailwindCSS](https://tailwindcss.com/docs/guides/nextjs)
* [A `/api/chat` API route for handling chat requests](src/pages/api/)
* [A React hook for calling the `/api/chat` endpoint](src/hooks)
* [Some chat components to display the messages](src/components)
For a complete walkthrough of this repository, [enroll now](https://graphacademy.neo4j.com/courses/llm-chatbot-typescript/?ref=github).
## Setup your Config
To set up the config, create a `.env.local` file with connection details for your Neo4j Sandbox instance and an OpenAI API Key.
You can also configure the name and description of the chatbot, and the initial greeting message.
```env
NEO4J_URI="bolt://123.45.678.90:7687"
NEO4J_USERNAME="neo4j"
NEO4J_PASSWORD="three-word-password"
OPENAI_API_KEY=sk-...
NEXT_PUBLIC_CHATBOT_NAME=Ebert
NEXT_PUBLIC_CHATBOT_DESCRIPTION="The Movie Recommendation Chatbot"
NEXT_PUBLIC_CHATBOT_GREETING="Hello, I'm **Ebert**, your movie recommendation bot! How can I help you today?"
```
## Running the application
To run the application, you must install the dependencies listed in `package.json`.
```sh
npm i
```
Then run the `npm run dev` command to start the app on [http://localhost:3000/](http://localhost:3000/).
```sh
npm run dev
```
## Questions, Comments, Feedback
If you have any questions, experience any problems, or have any general feedback, feel free to open an Issue or you can reach out to us on [Discord](https://dev.neo4j.com/chat) or [Discourse](https://dev.neo4j.com/form).
---
## Issues and Possible Fixes
- `TypeError: A dynamic import callback was invoked without --experimental-vm-modules`
> Change `test` command to `"test": "cross-env DOTENV_CONFIG_PATH=./.env.local NODE_OPTIONS=\"$NODE_OPTIONS --experimental-vm-modules\" jest"` in `package.json`.
> Refer [ECMAScript Modules in Jest](https://jestjs.io/docs/ecmascript-modules) for more.
- `Retrying request on response status: 429 Response: {"message":"Requests rate limit exceeded"}`
> Don't run multiple tests at once. Mistral AI's free trial API allows `1` request per second.
> Or, add `await new Promise((r) => setTimeout(r, 1000));` after each call to Mistral AI's API (e.g. `someChain.invoke(...)`) to add delay of 1 second.
- `The provided embedding function and vector index dimensions do not match.`
> OpenAI embeddings has 1536 dimensions and Mistral AI embeddings has 1024 dimensions.
> Change `dimensions` from `1536` to `1024`.
> ```cypher
> CREATE VECTOR INDEX `moviePlots` IF NOT EXISTS
> FOR (n: Movie) ON (n.embedding)
> OPTIONS {indexConfig: {
> `vector.dimensions`: 1024,
> `vector.similarity_function`: 'cosine'
> }};
> ```
>
> Now, use following cypher to load Mistral AI embeddings instead of OpenAI embedding.
> ```cypher
> LOAD CSV WITH HEADERS
> FROM 'https://raw.githubusercontent.com/nobleknightt/mistralai-embeddings/main/data/mistralai-embeddings.csv'
> AS row
> MATCH (m:Movie {movieId: row.movieId})
> CALL db.create.setNodeVectorProperty(m, 'embedding', apoc.convert.fromJsonList(row.embedding))
> RETURN count(*);
> ```
>
> Refer [github.com/nobleknightt/mistralai-embeddings](https://github.com/nobleknightt/mistralai-embeddings/) for more.
- Test fails for **Challenge: Creating the Agent**
> Use [Tool Calling Agent](https://js.langchain.com/v0.1/docs/modules/agents/agent_types/tool_calling/) instead of OpenAI Functions Agent.