inital commit

README.md

@@ -9,84 +9,111 @@ AIxplora is your new personal assistant, an open-source project that opens up un
 It leverages AI and LLMs to understand all types of documents, unrestricted by their length or format.
 
 **Imagine being able to query PDF files, MP3 audio, videos, and other types of documents with equal ease and proficiency. 
-Yes, that's the limitless world AIxplora is inviting you into!***
+Yes, that's the limitless world AIxplora is inviting you into!**
 
 
-## How Can You Contribute?
-There are several ways you can contribute as I just released a PoC of the project:
+## 🚀 Highlighted Features
 
-- Code: Write, refactor, optimize - every line of code matters!
+- **Universal File Integration**: Accepts any file type without restrictions on length.
+- **Open-Source Transparency**: Complete access to the source code, granting unparalleled flexibility and trust.
+- **Flexible Privacy Options**:
+  - Use official OpenAI and ChatGPT models while ensuring data confidentiality.
+  - **Option to utilize open-source models for an added layer of privacy. (Everything will run on your machine, no third party API usage)**
+- **Innovative Summarization**: Harness a unique approach to transform your files into concise summaries.
+- **Interactive File Indexing**: Engage in dynamic conversations with your indexed files, or detach the "AIxplora brain"-interface for a pure ChatGPT experience.
 
-- Documentation: Help us make our project more understandable and user-friendly.
+## 💡 Roadmap
 
-- Testing: Every bug found is a step towards perfection.
+- **AIxplora-Cloud**: Share your knowledge seamlessly, perfect for businesses, friends, or families aiming for collaborative learning.
+- **AIxplora Integration**: Embed your AIxplora brain as a Chat-widget on your website, enabling instant AI-backed responses to user queries.
+- **AIxplora Executable**: Simplified usage for all; install AIxplora just like any standard application, no technical expertise needed.
+- **Stay Tuned!**: More exciting updates are on the horizon.
 
-- Suggest Features: We believe in the power of ideas, no matter where they come from.
+## 🎥 Demo video
 
-- Spread the Word: Share our project within your networks. The more people know about AIxplora, the better it can become!
+https://github.com/grumpyp/aixplora/assets/75830792/7302684f-2c1f-4849-9f10-c6254be1009d
 
-## Demo and introduction videos
-[![Summary function showcase](https://img.youtube.com/vi/8x9HhWjjNtY/hqdefault.jpg)](https://youtu.be/8x9HhWjjNtY)
+
+
+more videos on YouTube:
+
+https://youtu.be/8x9HhWjjNtY
 https://youtu.be/2lNNKLM0o7U
 https://youtu.be/eKLmhJobVvc
 
 
-## How to run locally
+## 🛠 How to Run Locally
 
-1. Clone the repo and Install dependencies
-      ```
-      git clone [email protected]:grumpyp/aixplora.git
-      ```
-2. Install dependencies
-      ```
-      pip install -r backend/requirements.txt
-      cd frontend && npm install
-      cd ..
-      ```
-3. Run the backend and the frontend
-      ```
-      python backend/main.py
-      cd frontend
-      npm start
-      ```
+1. **Clone the Repository & Install Dependencies**
+    ```
+    git clone [email protected]:grumpyp/aixplora.git
+    ```
 
-**Having issues installing frontend? See this [debugging guide](https://github.com/electron-react-boilerplate/electron-react-boilerplate/issues/400)**
+2. **Install Dependencies**
+    ```
+    pip install -r backend/requirements.txt
+    cd frontend && npm install
+    cd ..
+    ```
 
-## How to run using Docker Compose
+3. **Launch the Backend & Frontend**
+    ```
+    python backend/main.py
+    cd frontend
+    npm start
+    ```
+
+🔍 **Troubleshooting**: Encountering frontend installation problems? Consult this [debugging guide](https://github.com/electron-react-boilerplate/electron-react-boilerplate/issues/400).
 
-1. Clone the repo and Install dependencies
+## 🐳 How to Run using Docker Compose
+
+1. **Clone the Repository**
     ```
     git clone [email protected]:grumpyp/aixplora.git
     ```
-2. Build Docker image and run containers
+
+2. **Build Docker Image & Spin Up Containers**
     ```
     install=true docker compose up --build
     ```
-3. When running the above command for the first time, make sure `frontend/node_modules` folder does not exist. The initial build might take some time since it will install all the required dependencies.
 
-4. Once the build and the package installation is finished, it should show an error in the console `app exited with code null, waiting for change to restart it` (We have to work on that issues).
+3. **Initial Build**
+    - Ensure the `frontend/node_modules` folder is absent on the first command execution.
+    - The initial building process might be prolonged due to dependency installation.
 
-5. Navigate to the UI on `http://localhost:1212/`.
+4. **Post-Build Notification**
+    - After the build and package installation concludes, an error might appear in the console: `app exited with code null, waiting for change to restart it`. This is a known issue we're addressing.
 
-6. Next time when starting the app you can simply use the following command
+5. **Access the UI**
+    - Visit `http://localhost:1212/`.
+
+6. **Subsequent Launches**
     ```
     docker compose up
     ```
-    **Note that** 
-      - After adding new packages in `requirements.txt` you'll have to run `docker compose up --build`
-      - After adding new packages in `package.json` you'll have to run `install=true docker compose up` to install the new packages.
-      - If you want to just run frontend run `docker compose up frontend`
-      - If you want to just run backend run `docker compose up backend`
+    📝 **Notes**:
+    - After appending new packages in `requirements.txt`, execute `docker compose up --build`.
+    - Post adding fresh packages in `package.json`, use `install=true docker compose up` for new package installations.
+    - To solely launch the frontend: `docker compose up frontend`.
+    - To solely launch the backend: `docker compose up backend`.
+
+
+
+## 🤝 How Can You Contribute?
 
+With the recent release of a PoC for the project, your involvement is pivotal. Here's how you can be a part of our journey:
 
-## Roadmap
+- **Code**: Dive deep into our codebase! Whether it's writing, refactoring, or optimizing, every line contributes to our collective vision.
+
+- **Documentation**: Illuminate our project's essence. Assist in crafting clearer and more user-centric guidelines and explanations.
+
+- **Testing**: Become our frontline in quality assurance. Each bug identified is a stride towards unparalleled product excellence.
+
+- **Suggest Features**: Your imagination is our canvas. We deeply value ideas, irrespective of their origin.
+
+- **Spread the Word**: Amplify our message. Introduce AIxplora to your network and watch it evolve and flourish with increased collective insight.
 
-- Build a community around the project
-- Release a standalone desktop app
-- Bugfixes and improvements to scale the project
-- Add more features (custom LLMs, more file types, etc.)
-- Cloud deployment
-- Integrations (Google Drive, Dropbox, etc.)
+## Star history
 
 [![Star History Chart](https://api.star-history.com/svg?repos=grumpyp/aixplora&type=Date)](https://star-history.com/#grumpyp/aixplora&Date)
 

backend/embeddings/index_files.py

@@ -7,7 +7,7 @@
 from langchain.schema import Document
 from database.database import Database
 from sqlalchemy import text
-from embeddings.utils import openai_ask
+from embeddings.utils import openai_ask, openai_ask_no_aixplora_brain
 import random
 from qdrant_client import QdrantClient
 from qdrant_client.http import models
@@ -244,24 +244,36 @@ def query(self, query_embedding: List[List[float]] = None, query_texts: str = No
         else:
             relevant_docs = [doc.payload["chunk"] for doc in results]
             meta_data = [doc.payload["metadata"] for doc in results]
-        print("halloooo")
-        print(self.openai_model)
         if not self.openai_model[0].startswith("gpt"):
             print(f"Using local model: {self.openai_model[0]}")
             # TODO: refactor this path to be global
             models_dir = os.path.join(os.getcwd(), "llmsmodels")
             gptj = GPT4All(model_name=self.openai_model[0], model_path=models_dir)
-            messages = [
-                {"role": "user",
-                 "content": f"Answer the following question: {query_texts} based on that context: {relevant_docs},"
-                            " Make sure that the answer of you is in the same language then the question. if you can't just answer: I don't know"}
-            ]
+            if use_brain:
+                messages = [
+                    {"role": "user",
+                     "content": f"Answer the following question: {query_texts} based on that context: {relevant_docs},"
+                                " Make sure that the answer of you is in the same language then the question. if you can't just answer: I don't know"}
+                ]
+            else:
+                messages = [
+                    {"role": "user",
+                     "content": f"{query_texts}"}]
             answer = gptj.chat_completion(messages, streaming=False)["choices"][0]["message"]["content"]
         else:
-            if self.openai_model[0].startswith("gpt"):
-                print(f"Using openai model: {self.openai_model[0]}")
-                answer = openai_ask(context=relevant_docs, question=query_texts, openai_api_key=self.openai_api_key[0],
-                                    openai_model=self.openai_model[0])
+            if use_brain:
+                if self.openai_model[0].startswith("gpt"):
+                    print(f"Using openai model: {self.openai_model[0]}")
+                    answer = openai_ask(context=relevant_docs, question=query_texts, openai_api_key=self.openai_api_key[0],
+                                        openai_model=self.openai_model[0])
+                else:
+                    answer = openai_ask(context=relevant_docs, question=query_texts,
+                                        openai_api_key=self.openai_api_key[0],
+                                        openai_model=self.openai_model[0])
+            else:
+                if self.openai_model[0].startswith("gpt"):
+                    answer = openai_ask_no_aixplora_brain(question=query_texts, openai_api_key=self.openai_api_key[0],
+                                                          openai_model=self.openai_model[0])
         _answer = {"answer": answer, "meta_data": meta_data}
         print(meta_data)
         return _answer

backend/embeddings/utils.py

@@ -12,6 +12,7 @@ def openai_ask(context: str = None, pages: List[int] = None, question: str = Non
     print(pages)
     # TODO: make answer to same language
     if openai_model.startswith("gpt"):
+        openai.api_key = openai_api_key
         completion = openai.ChatCompletion.create(
             model=f"{openai_model}",
             messages=[
@@ -34,3 +35,22 @@ def openai_ask(context: str = None, pages: List[int] = None, question: str = Non
         return gptj.chat_completion(messages=messages, streaming=False)
 
 
+def openai_ask_no_aixplora_brain(question: str, openai_api_key: str = None, openai_model: str = "gpt-3.5-turbo"):
+    if openai_model.startswith("gpt"):
+        openai.api_key = openai_api_key
+        completion = openai.ChatCompletion.create(
+            model=f"{openai_model}",
+            messages=[
+                {"role": "user", "content": f"{question}"}
+            ]
+        )
+        return completion["choices"][0]["message"]["content"]
+    else:
+        print(f"Using local model: {openai_model}")
+        models_dir = os.path.join(os.getcwd(), "llmsmodels")
+        gptj = GPT4All(model_name=openai_model, model_path=models_dir)
+        messages = [
+            {"role": "user",
+                "content": f"{question}"}
+        ]
+        return gptj.chat_completion(messages=messages, streaming=False)

backend/main.py

@@ -1,5 +1,6 @@
 import uvicorn
 from fastapi import FastAPI, File, UploadFile, Request, Header
+from typing import Optional
 from fastapi.middleware.cors import CORSMiddleware
 from sqlalchemy import text
 from typing import List
@@ -206,10 +207,10 @@ def chat(request: Request, question: Question, document: Document):
     apikey = request.headers.get("apikey", False)
     email = request.headers.get("email", False)
     genie = Genie()
-    print(email, apikey)
     if apikey and email:
         genie = Genie(remote_db=True, apikey=apikey, email=email)
     answer = genie.query(query_texts=question.question, specific_doc=document.document)
+
     print(answer)
     return {"question": question.question, "answer": answer["answer"], "meta_data": answer["meta_data"]}
 

frontend/src/renderer/pages/Chat/Chat.tsx

@@ -5,7 +5,7 @@ import {Message} from 'renderer/utils';
 import Question from './components/Question';
 import Answer from './components/Answer';
 import {IconSend, IconTrash, IconArrowUp} from '@tabler/icons-react';
-import {Modal, Select} from '@mantine/core';
+import {Modal, Select, Switch} from '@mantine/core';
 import {useDisclosure} from '@mantine/hooks';
 import Block from './components/Block';
 
@@ -22,6 +22,7 @@ function Chat() {
     const [error, setError] = useState(false);
     const [input, setInput] = useState('');
     const discussionRef = useRef<HTMLDivElement>(null);
+    const [useBrain, setUseBrain] = useState(true);
 
 
     // used for the specific_document chat
@@ -79,6 +80,7 @@ function Chat() {
                 document: {
                     document: selectedFile,
                 },
+                usebrain: useBrain,
             });
             const data = response.data;
 
@@ -146,6 +148,13 @@ function Chat() {
             <div onClick={up} className="arrow_up">
                 <IconArrowUp color="white"/>
             </div>
+            <div className={'switch'}>
+                AIxplora brain
+                <Switch
+                    onLabel="ON" offLabel="OFF" defaultChecked="true"
+                    onClick={() => { setUseBrain(!useBrain) }}
+                />
+                </div>
             <Modal.Root opened={opened} onClose={close}>
                 <Modal.Overlay/>
                 <Modal.Content>

frontend/src/renderer/pages/Chat/chat.css

@@ -185,6 +185,16 @@ p {
   cursor: pointer;
 }
 
+.switch {
+  background: rgba(34, 139, 230, 0.3);
+  position: fixed;
+  top: 15em;
+  left: 0;
+  width: 65px;
+  border-top-right-radius: 10px;
+  border-bottom-right-radius: 10px;
+}
+
 .metadata {
     display: flex;
     flex-direction: row;