https://github.com/astrabert/phiqwenstem
A reasoning assistant for your STEM education
https://github.com/astrabert/phiqwenstem
ai datasets huggingface llm phi-3-5 qdrant qwen react reasoning research scientific stem vite websockets
Last synced: about 1 month ago
JSON representation
A reasoning assistant for your STEM education
- Host: GitHub
- URL: https://github.com/astrabert/phiqwenstem
- Owner: AstraBert
- License: mit
- Created: 2025-01-28T12:57:09.000Z (3 months ago)
- Default Branch: main
- Last Pushed: 2025-02-11T22:40:23.000Z (2 months ago)
- Last Synced: 2025-03-10T15:58:01.312Z (about 1 month ago)
- Topics: ai, datasets, huggingface, llm, phi-3-5, qdrant, qwen, react, reasoning, research, scientific, stem, vite, websockets
- Language: TypeScript
- Homepage: https://pqstem.org
- Size: 1.31 MB
- Stars: 20
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Contributing: CONTRIBUTING.md
- License: LICENSE
Awesome Lists containing this project
README
PhiQwenSTEM
A reasoning assistant for your STEM education
> [!IMPORTANT]
> _The Proof Of Concept live at: https://pqstem.org is **no longer available**_
**PhiQwenSTEM** is an assistant aimed at helping you solve complex STEM questions through reasoning. It is based on **Phi-3.5** and **QwQ-32B-preview** by Microsoft, provided by [HuggingFace](https://huggingface.co) Inference API, and has a vast knowledge base (more than 15,000 STEM questions) managed via [Qdrant](https://qdrant.tech).
## Workflow
## How PhiQwenSTEM works
PhiQwenSTEM operates through three main components:
- **Front-end**: Utilizes Vite to render a landing page and a ChatGPT-like chat interface.
- **Back-end**: Employs a Python-based websocket to process messages from the front-end and send responses.
- **Database**: Uses a vector database built on [Qdrant](https://qdrant.tech) to store data for retrieval-augmented generation and semantic caching.
Once you launch the application, the vector database will ingest more than 15,000 STEM-related questions. Each question is associated with:
- The question itself
- `QwQ-32B-preview` reasoning about the question
The questions span the following domains of science:
- Chemistry (General, Organic, and Biochemistry)
- Physics
- Physical Chemistry
- Quantum Mechanics
- Differential Equations
- Linear Algebra
- Electromagnetism
- Mathematics
- Engineering
- Classical Mechanics
The data comes from the HuggingFace dataset [EricLu/SCP-116K](https://huggingface.co/datasets/EricLu/SCP-116K), made by more than 116,000 STEM-related questions accompanied by the ground truth answer, [`QwQ-32B-preview`](https://huggingface.co/Qwen/QwQ-32B-Preview) reasoning and solution and `o1` reasoning and solution: we selected questions (from the most represented domains in the dataset) in which reasoning by `QwQ-32B-preview` produced the correct answer.
Dense embeddings are obtained using the static text encoder [`tomaarsen/static-retrieval-mrl-en-v1`](https://huggingface.co/tomaarsen/static-retrieval-mrl-en-v1) (embedding size is truncated to 384), while sparse embeddings are generated with [`Qdrant/bm25`](https://huggingface.co/Qdrant/bm25). To speed up retrieval, the medical vector database leverages [binary quantization](https://qdrant.tech/articles/binary-quantization/).
When a user asks a medical question:
1. The backend first checks for similar questions in the semantic cache using [`modernbert-embed-base`](https://huggingface.co/nomic-ai/modernbert-embed-base). If a match is found, the corresponding answer is returned.
2. If no significant match is found, it prompts [`Phi-3.5-mini-instruct`](https://huggingface.co/microsoft/Phi-3.5-mini-instruct) (served on HuggingFace Inference API) to produce a question for searching the vector database.
3. The optimized question prompts a hybrid search within the medical vector database. The top 2 ranking matches for both sparse and dense vectors are retrieved and re-scored by `modernbert-embed-base`.
4. The top-ranking retrieved match (after re-scoring) is retained, and `QwQ-32B-preview`-generating reasoning (from the match payload) is passed on as a "reasoning" context.
5. `QwQ-32B-preview` is prompted to produce an answer based on the reasoning.
> [!NOTE]
> `QwQ-32B-preview` is instructed to assess if the reasoning and the answer provided are valid, relevant to the user's question, and correct. It is also instructed to output an "I don't know" answer when the question is ambiguous and the solution is not completely clear.
## Installation and usage
### 1. Docker
> _Required: [Docker](https://docs.docker.com/desktop/) and [docker compose](https://docs.docker.com/compose/)_
- Clone this repository
```bash
git clone https://github.com/AstraBert/PhiQwenSTEM.git
cd PhiQwenSTEM/docker-workflow/
```
- Add the `hf_token` secret in the [`.env.example`](./docker/.env.example) file and modify the name of the file to `.env`. You can get your HuggingFace token by [registering](https://huggingface.co/join) to HuggingFace and creating a [fine-grained token](https://huggingface.co/settings/tokens) that has access to the Inference API.
```bash
# modify your access token, e.g. hf_token="hf_abcdefg1234567"
mv .env.example .env
```
- Launch the docker application:
```bash
# If you are on Linux/macOS
bash start_services.sh
# If you are on Windows
.\start_services.ps1
```
You will see the application running on http://localhost:8501 and you will be able to use it successfully only after the backend is set up (you can see it from the logs). Depending on your connection and on your hardware, this might take some time (up to 30 mins to set up).
## Local
> _Required: [Docker](https://docs.docker.com/desktop/), [docker compose](https://docs.docker.com/compose/) and [conda](https://anaconda.org/anaconda/conda)_
- Clone this repository
```bash
git clone https://github.com/AstraBert/PhiQwenSTEM.git
cd PhiQwenSTEM/local
```
- If you are on macOS/Linux, you can run:
```bash
bash local_setup.sh
```
- If you are on Windows, running all the commands separately might be optimal:
```bash
# Launch Qdrant
docker compose up -d
# Create conda environment for the backend
conda env create -f ./backend/environment.yml
conda activate backend
# Ingest data
python3 data/toDatabase.py
# Create a semantic cache
python3 data/createCache.py
conda deactivate
# Install necessary dependencies for the UI
cd chatbot-ui/
npm install
# Back to the local folder
cd ..
```
- Once you are done with the set-up, launch the UI:
```bash
cd chatbot-ui/
npm run dev
```
- And, on a separate terminal window, launch the backend:
```bash
conda activate backend
cd backend/
python3 backend.py
```
Head over to http://localhost:8501 and you should see PhiQwenSTEM up and running in less than one minute!
## Contributions
Contributions are more than welcome! See [contribution guidelines](./CONTRIBUTING.md) for more information :)
## Funding
If you found this project useful, please consider to [fund it](https://github.com/sponsors/AstraBert) and make it grow: let's support open-source together!😊
## License and rights of usage
The software is hereby provided under an MIT license and is free to use.