feat: add documentation and update readme to reflect llm v2 changes

Author: stackblogger

Readme.md

@@ -1,104 +1,55 @@
-# BitNet.js
+<p align="center">
+  <img src="docs/assets/bitnet-stack.png" alt="BitNet-Stack" />
+</p>
 
-[![License: MIT](https://img.shields.io/badge/license-MIT-purple.svg)](https://opensource.org/licenses/MIT)
-![version](https://img.shields.io/badge/version-1.0.0-purple)
+<h1 align="center">BitNet-Stack</h1>
 
-BitNet.js is the Node.js implementation of Microsoft's [bitnet.cpp](https://github.com/microsoft/BitNet) inference framework. This repository facilitates real-time interaction between a Node.js frontend and the bitnet 1-bit LLM model using **Socket.IO**. The app allows users to send queries to the BitNet LLM (Large Language Model) and receive responses line by line via a web interface.
+[![License: MIT](https://img.shields.io/badge/license-MIT-green.svg)](https://opensource.org/licenses/MIT)
 
-## Working Example
+Run a small BitNet model on local machine with one Docker command, and chat in browser.
 
-Web app will display the results in real-time.
+## What you get
 
-### Docker App Run
+- **One Docker Compose command** — Build and start the LLM server with a single command. The image pulls the **official small BitNet GGUF** from Hugging Face (`BitNet-b1.58-2B-4T-gguf`, `ggml-model-i2_s.gguf`) so you do not manage weights by hand.
+- **Web interface** — Open the app in your browser, start a chat, and talk to the model from a simple page.
+- **Chat history in the browser** — Conversations are stored in **local storage** on your device. They stay until you clear them.
+- **Clear all chats** — Use the control in the UI to remove every saved thread and stop model sessions in one go.
+- **Session context** — Each chat keeps a **conversation session** with the model so follow-up messages stay in context until you start a new chat or clear data.
+- **Responsive replies** — Streaming output and a tuned server path so answers show up quickly as they are generated.
+- **Runs anywhere Docker runs** — On a laptop, a desktop, or a small server: clone the repo, run `docker compose`, open the URL. No extra install steps for Node or Python on the host.
 
-![bitnet js-docker-run](https://github.com/user-attachments/assets/8c439904-fec7-4465-ac7d-5ebfc29d2442)
+## Working example
 
-### Web Interface Interaction
-
-![bitnet js-llm](https://github.com/user-attachments/assets/83882cf5-0fb4-4fe4-9068-6447e26c4283)
-
-
-## Features
-
-- **Web-based interface**: A simple frontend built with HTML and JavaScript to interact with the BitNet model.
-- **Real-time communication**: Uses **Socket.IO** for bi-directional communication between the Node.js app and the Python-based BitNet model.
-- **Dockerized environment**: Both the Node.js app and BitNet model run in separate Docker containers managed by **docker compose**.
-
-## Directory Structure
-
-```bash
-bitnet.js/
-├── apps/
-│   ├── llm/
-│   │   ├── Dockerfile              # Dockerfile for the BitNet model
-│   │   ├── requirements-local.txt  # List of packages that are used in Local Server
-│   │   ├── run_model.py            # Local Python Server to start Socket
-│   ├── web/
-│   │   ├── Dockerfile              # Dockerfile for the Node.js application
-│   │   ├── app.js                  # Node.js app (Socket.IO client)
-│   │   ├── index.html              # Frontend (textarea for input, display for output)
-├── docker-compose.yml              # Compose file to run both containers together
-├── README.md                       # Documentation
-```
+![bitnet-stack](docs/assets/conversational-chat.gif)
 
 ## Prerequisites
 
-- **Docker**: Ensure Docker is installed. [Install Docker](https://docs.docker.com/get-docker/)
+- [Docker](https://docs.docker.com/get-docker/) (with Docker Compose)
 
-## Setup Instructions
+## Quick start
 
-### Step 1: Clone the Repository
+### 1. Clone the repository
 
 ```bash
-git clone https://github.com/stackblogger/bitnet.js.git
-cd bitnet.js
+git clone https://github.com/stackblogger/BitNet-Stack.git
+cd BitNet-Stack
 ```
 
-### Step 2: Build and Run Containers
-
-To build and start both the Node.js app and the BitNet model, use:
+### 2. Start the server
 
 ```bash
 docker compose up --build -d
 ```
 
-This command will build the Docker images for both the **web** (Node.js) and **llm** (Python) containers, and run the services defined in the `docker-compose.yml` file.
-
-### Step 3: Access the Web Application
-
-After starting the containers, access the web interface by navigating to:
-
-`http://localhost:3000`
-
-You'll see a textarea where you can input your queries and send them to the BitNet model.
-
-### Step 4: Interact with the BitNet Model
-
-1. **Enter a query** in the provided textarea (e.g., "Why is the sky blue?").
-2. **Click Send**. Your query is transmitted to the BitNet model via Socket.IO.
-3. The response will be returned **line by line** in real-time.
-
-## How it Works
-
-- The **Node.js application** (in `apps/web`) serves a simple web interface where users can input their queries. The application connects to the **Python backend** using Socket.IO.
-- The **Python app** (in `apps/llm`) runs the BitNet model, processes the queries, and streams the response back to the Node.js client, word by word.
-- Both applications are containerized and communicate through **docker compose**.
-
-### Node.js Frontend (app.js)
-
-The Node.js frontend sends user queries to the backend via a Socket.IO connection. Here's the simplified workflow:
+This builds the **LLM** image and starts one container. The model is downloaded during the image build (first time can take a while).
 
-1. **User Input**: The user enters a query and clicks "Send."
-2. **Socket Emission**: The query is sent to the backend using Socket.IO.
-3. **Receive Data**: The response is received word by word and displayed on the page.
+### 3. Open the chat UI
 
-### Python Backend (BitNet Integration)
+In your browser go to:
 
-The Python backend runs the BitNet model and returns the response:
+**http://localhost:5001**
 
-1. **Receive Query**: The Python app receives a query from the Node.js frontend.
-2. **Model Inference**: The query is processed by the BitNet model.
-3. **Stream Results**: The model's response is streamed word by word back to the Node.js client.
+(Port **5001** is mapped to the app inside the container on **5000**; change it in `docker-compose.yml` if you need another port.)
 
 ## License
 

apps/llm/static/index.html

@@ -450,7 +450,7 @@
 <body>
   <aside>
     <div class="side-head">
-      <div class="logo">BitNet <span>local</span></div>
+      <div class="logo">BitNet <span>stack</span></div>
     </div>
     <button type="button" class="btn btn-new" id="btnNew">New chat</button>
     <div class="history">