raresmihalache/chatbot_nestjs

Chatbot that could be used to make offers to potential clients for different kinds of applications.

A progressive Node.js framework for building efficient and scalable server-side applications.

Description

Nest framework TypeScript starter repository.

Installation

$ npm install

Running the app

# development
$ npm run start

# watch mode
$ npm run start:dev

# production mode
$ npm run start:prod

Test

# unit tests
$ npm run test

# e2e tests
$ npm run test:e2e

# test coverage
$ npm run test:cov

Support

Nest is an MIT-licensed open source project. It can grow thanks to the sponsors and support by the amazing backers. If you'd like to join them, please read more here.

Stay in touch

• Author - Kamil Myśliwiec
• Website - https://nestjs.com
• Twitter - @nestframework

License

Nest is MIT licensed.

List of dependencies

• axios -> for HTTP calls
• dotenv -> to configure .env file
• openai -> for llm integration (gpt-3.5-turbo)
• tiktoken -> to get to know how many tokens were consumed during requests / responses. This way we can create a policy such that the context window will never be exceeded.
• other dependencies -> used to run common nestjs projects & for typescript configuration

Steps used to train / use the model

There is no need to train the model, since it will be already trained when you run the app. The way this web app works is that after you run the project, 5 offers with the client requirements and the corresponding responses will be put into an array that represents the context window.

Since gtp-3.5-turbo can hold up to 16384 tokens in its context, we need to always keep in mind that we have to remove old messages when we are very close to that threshold.

The model will learn from the system prompt & his previous messages what he needs to do. We will never delete the system prompt (only the user and assistant prompts)

Just run the project, go to localhost:3000 and fill in the input with the client's requirement. After 20-30 seconds you will get the desired output.