metadata

title: Template Final Assignment
emoji: 🕵🏻‍♂️
colorFrom: indigo
colorTo: indigo
sdk: gradio
sdk_version: 5.25.2
app_file: app.py
pinned: false
hf_oauth: true
hf_oauth_expiration_minutes: 480
short_description: Gemini Agent for GAIA Evaluation

🧠 Gemini Agent for GAIA Evaluation

This project contains a Gemini-powered CodeAgent built with smolagents for use in the GAIA Unit 4 Evaluation on Hugging Face Spaces.

🚀 Features

Uses the Gemini 2.0 Flash model via LiteLLMModel
Equipped with essential tools:
- DuckDuckGoSearchTool for quick lookups
- RunPythonFileTool for executing .py scripts
- ReverseTextTool for decoding reversed questions
- download_server for fetching files from URLs
- Base tools (math, string manipulation, etc.)

📋 Evaluation Strategy

The agent reads questions from the GAIA evaluation endpoint, applies reasoning using a system prompt with strict guidelines, and submits answers back for scoring.

🛠️ Setup

Clone this repository or Space

Set your environment variables:

GEMINI_API_KEY=your_api_key_here
SPACE_ID=your_hf_space_id

Install dependencies:
```
pip install -r requirements.txt
```
Run locally:
```
python app.py
```

Or launch directly via Hugging Face Spaces.

🧪 Evaluation Flow

Log in to Hugging Face through the UI
Click “Run Evaluation & Submit All Answers”
The agent will fetch tasks, solve them, and submit results

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference