https://github.com/cyb3rward0g/intelragu
Intel Retrieval Augmented Generation (RAG) Utilities
https://github.com/cyb3rward0g/intelragu
Last synced: 3 months ago
JSON representation
Intel Retrieval Augmented Generation (RAG) Utilities
- Host: GitHub
- URL: https://github.com/cyb3rward0g/intelragu
- Owner: Cyb3rWard0g
- Created: 2024-01-28T23:37:57.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2024-01-29T17:55:16.000Z (over 1 year ago)
- Last Synced: 2025-03-21T01:34:25.138Z (7 months ago)
- Language: Jupyter Notebook
- Size: 811 KB
- Stars: 90
- Watchers: 9
- Forks: 5
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Intel Retrieval Augmented Generation (RAG) Utilities
An open-source initiave to document and share experiments to apply Retrieval Augmented Generation (RAG) techniques to Threat Intelligence searching capabilities.
## Build Docker Image
```
docker build . -t rag-chroma
```
## Define .ENV File
Create a `.env` file and define the `OPENAI_API_KEY` variable with your OpenAI Key. This is needed to use the [LangChain's ChatOpenAI module](https://python.langchain.com/docs/integrations/chat/openai). This is not needed to embed the ATT&CK Groups data. This is done with the [all-mpnet-base-v2 sentence-transformers model](https://huggingface.co/sentence-transformers/all-mpnet-base-v2) ;) .
```
OPENAI_API_KEY=XXXXXXXXX
```
## Run Docker Image
```
docker run -it --rm --name rag-chroma --env-file .env -p 8080:8080 rag-chroma
```
After running that command, the container will:
* Download the [all-mpnet-base-v2 sentence-transformers model](https://huggingface.co/sentence-transformers/all-mpnet-base-v2) (~400MB).
* Download the [Hugging Face Cyb3rWard0g/ATTCKGroups dataset](https://huggingface.co/datasets/Cyb3rWard0g/ATTCKGroups) (~846KB).
* Process the dataset by tokenizing and embedding every ATT&CK Group.
* Create the vector database by adding all the embeddings into a local [Chroma Database](https://www.trychroma.com).
## Explore Playground
Browse to `http://127.0.0.1/rag-chroma/playground` and start asking questions.
## References
* https://python.langchain.com/docs/templates/
* https://python.langchain.com/docs/templates/rag-chroma
* https://huggingface.co/datasets/Cyb3rWard0g/ATTCKGroups