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: πŸ“Š
colorFrom: blue
colorTo: indigo
sdk: streamlit
sdk_version: 1.58.0
app_file: app.py
pinned: false
license: mit
---

# πŸ“Š 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 ~1–2 min. CPU basic (free,
> 16 GB RAM) is sufficient.