lindalen/generator-evaluator

🔄 Generator-Evaluator Workflow 🔄

🧠 One output. Continuous refinement. This demo showcases the generator-evaluator workflow, where a single output is generated and refined iteratively until it meets clear criteria. Inspired by Anthropic's research on building effective agents, this workflow excels in tasks requiring measurable improvement through iteration.

How It Works

1. 1.The generator creates an initial output.
2. 2.The evaluator provides feedback or refines the output.
3. 3.The process repeats until the output is acceptable.

Use Cases

• —Literary translation: Captures subtle nuances through iterative improvement.
• —Complex searches: Gathers and refines information across multiple iterations.

🚀 Try it Out

Visit the app: 🔄 Generator-Evaluator Workflow 🔄

1. 1.Input a prompt for the generator.
2. 2.Watch as refinement improves the output step by step.
3. 3.See how iteration leads to better results.

📚 Learn More

This workflow is inspired by Anthropic's research on building effective agents.

📜 License

Licensed under the MIT License.