Responsible AI for Music: Agentic Generation & Discovery
This challenge tasks developers with building an advanced agentic system for responsible music generation and discovery. The system should not only generate novel music snippets based on user prompts but also critically evaluate its own outputs and curated recommendations for adherence to ethical AI principles, originality, and avoiding stylistic biases. This involves orchestrating multiple generative AI models and integrating robust evaluation mechanisms within an agent workflow, ensuring the output is creative, relevant, and ethically sound.
What you are building
The core problem, expected build, and operating context for this challenge.
Build a multi-agent system using /dev/agents and Falcon 180B to responsibly generate music and provide curated discovery, integrating Cleanlab for bias detection.
How work is evaluated
The evaluation will assess the functionality of the multi-agent system, the quality and originality of generated music snippets, and the effectiveness of responsible AI checks. Performance metrics for inference and overall workflow efficiency will also be considered.
Challenge datasets
Public datasets, plus the private datasets you have access to.
What you should walk away with
Optimize model inference for scalable music generation services.
Implement responsible AI checks for bias and ethical considerations in generated and recommended content.
Design and implement a multi-agent system for music generation and discovery.
Integrate and orchestrate generative audio models and large language models.
Reference links and supporting material
Sample data for 2 tasks
Challenge setup
The evaluation will assess the functionality of the multi-agent system, the quality and originality of generated music snippets, and the effectiveness of responsible AI checks. Performance metrics for inference and ov...
Challenge input
{'prompt': 'A mellow jazz track with a hint of lo-fi beats, suitable for studying.'}
Agent execution
No agent tool is listed for this challenge.
Evaluated output
{'music_url': 'https://example.com/track.mp3', 'originality_score': 0.85, 'bias_report': {'genre_bias': 0.1, 'mood_bias': 0.05}}
- Ensures all agentic steps, including generation and ethical review, complete successfully without runtime e...
- Verifies that the `bias_report` contains meaningful analysis from Cleanlab, indicating successful integration.
- OriginalityScore target: 0.8
- Python execution harness
Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.
View technical recipe
Configured tools
No tool records are attached.
Evaluation contract
- Ensures all agentic steps, including generation and ethical review, complete successfully without runtime errors.
- Verifies that the `bias_report` contains meaningful analysis from Cleanlab, indicating successful integration.
Metrics
- OriginalityScore · Target 0.8
- BiasDetectionEfficacy · Target 0.75
- InferenceLatency · Target 1500
Workflow tasks
- Task 1
Input: {'prompt': 'A mellow jazz track with a hint of lo-fi beats, suitable for studying.'}
Output: {'music_url': 'https://example.com/track.mp3', 'originality_score': 0.85, 'bias_report': {'genre_bias': 0.1, 'mood_bias': 0.05}}
Execution harness: Python
- Task 2
Input: {'user_id': 'user123', 'listening_history': ['track_x', 'track_y'], 'desired_mood': 'relaxed'}
Output: {'recommended_tracks': [{'title': 'Track Z', 'artist': 'Artist A', 'url': '...', 'ethical_score': 0.95}], 'ethical_analysis': 'No detected harmful biases.'}
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.