Spaces:
Sleeping
Sleeping
| title: GAIA Agent System | |
| emoji: 🤖 | |
| colorFrom: yellow | |
| colorTo: pink | |
| sdk: gradio | |
| sdk_version: 5.25.2 | |
| app_file: ./src/app.py | |
| pinned: false | |
| hf_oauth: true | |
| # optional, default duration is 8 hours/480 minutes. Max duration is 30 days/43200 minutes. | |
| hf_oauth_expiration_minutes: 480 | |
| # Required scopes for Qwen model access via Inference API | |
| hf_oauth_scopes: | |
| - inference-api | |
| short_description: Multi-Agent AI System for GAIA Benchmark Questions | |
| suggested_hardware: cpu-upgrade | |
| models: | |
| - Qwen/Qwen2.5-7B-Instruct | |
| - Qwen/Qwen2.5-32B-Instruct | |
| - Qwen/Qwen2.5-72B-Instruct | |
| tags: | |
| - GAIA | |
| - multi-agent | |
| - LangGraph | |
| - benchmark | |
| - reasoning | |
| - web-search | |
| - file-processing | |
| - question-answering | |
| # 🤖 GAIA Agent System | |
| Advanced Multi-Agent AI System for GAIA Benchmark Questions using LangGraph orchestration. | |
| ## Features | |
| - **Multi-Agent Architecture**: Router, Web Research, File Processing, Reasoning, and Synthesizer agents | |
| - **LangGraph Orchestration**: Intelligent workflow management with state tracking | |
| - **Unit 4 API Integration**: Official GAIA benchmark submission and scoring | |
| - **Smart Model Selection**: Tiered Qwen 2.5 models (7B/32B/72B) for optimal cost/performance | |
| - **Comprehensive Tools**: Wikipedia search, web scraping, mathematical calculations, file analysis | |
| ## Usage | |
| 1. **Official GAIA Evaluation**: Login with HuggingFace and run complete benchmark | |
| 2. **Manual Testing**: Test individual questions with detailed reasoning analysis | |
| 3. **File Processing**: Upload and analyze CSV, images, code, and audio files | |
| Check out the configuration reference at <https://huggingface.co/docs/hub/spaces-config-reference> | |