Spaces:
Sleeping
Sleeping
File size: 1,131 Bytes
19c0937 76bd1fc 19c0937 76bd1fc 19c0937 76bd1fc 19c0937 76bd1fc 19c0937 76bd1fc | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 | ---
title: FinAgent Infosys AR QA
emoji: π
colorFrom: blue
colorTo: indigo
sdk: streamlit
sdk_version: 1.58.0
app_file: app.py
pinned: false
license: mit
---
# π FinAgent β Infosys Annual Report Q&A
An agentic **Retrieval-Augmented Generation** app over the Infosys Integrated
Annual Report 2024-25. Hybrid retrieval (BM25 + dense + HyDE) β cross-encoder
reranking β LangGraph multi-agent orchestration β FinBERT sentiment β inline
Self-RAG faithfulness check, served with Streamlit.
## Setup on Hugging Face Spaces
This Space needs **one secret** to answer questions:
- **`GROQ_API_KEY`** β add it under **Settings β Variables and secrets β New secret**.
Get a free key at https://console.groq.com/keys
The UI and models load without it; only the LLM answers require it.
## Stack
Groq `llama-3.3-70b-versatile` Β· LangGraph Β· LangChain Β· ChromaDB (pre-built,
shipped in `finance_db/`) Β· HuggingFace embeddings (`all-mpnet-base-v2`) Β·
cross-encoder (`ms-marco-MiniLM-L-6-v2`) Β· FinBERT.
> First boot downloads ~1 GB of models and takes ~1β2 min. CPU basic (free,
> 16 GB RAM) is sufficient.
|