|
|
--- |
|
|
language: |
|
|
- en |
|
|
- ur |
|
|
- hi |
|
|
- es |
|
|
tags: |
|
|
- rag |
|
|
- retrieval-augmented-generation |
|
|
- multilingual |
|
|
- faiss |
|
|
- llama |
|
|
- mistral |
|
|
license: apache-2.0 |
|
|
library_name: transformers |
|
|
pipeline_tag: question-answering |
|
|
--- |
|
|
|
|
|
|
|
|
# Multilingual Knowledge RAG Bot – Cross-Lingual Retrieval-Augmented Generation |
|
|
|
|
|
This model is designed for **cross-lingual question answering** using Retrieval-Augmented Generation (RAG). |
|
|
It can take documents in multiple languages — Urdu, Hindi, Spanish, English — and answer in the same or different language. |
|
|
|
|
|
## Key Features |
|
|
- **LLM Used:** Meta-Llama-3-8B-Instruct |
|
|
- **Embedding Model:** `sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2` |
|
|
- **RAG Pipeline:** FAISS-based vector search + context injection |
|
|
- **Training/Processing:** Implemented entirely in Google Colab using open-source tools only |
|
|
- **Zero paid APIs** — 100% free and deployable |
|
|
|
|
|
## Techniques Used |
|
|
- **Vector Database:** FAISS for similarity search |
|
|
- **Cross-Lingual Embeddings:** multilingual sentence transformers |
|
|
- **Prompt Engineering:** Context-aware question answering |
|
|
- **Open-Source Deployment Ready:** Hugging Face Spaces compatible |
|
|
|
|
|
--- |
|
|
|
|
|
### License |
|
|
Apache-2.0 |
|
|
|
