Spaces:
Sleeping
Sleeping
| title: CodeBook | |
| emoji: π | |
| colorFrom: purple | |
| colorTo: red | |
| sdk: gradio | |
| sdk_version: 5.49.1 | |
| app_file: app.py | |
| pinned: true | |
| short_description: Advanced AI workflowing system. | |
| # π€ CodeBook AI Workflow Space: Code or Book Generator with Self-Critique | |
| This Hugging Face Space refines an **idea** into either: | |
| - A *complete code project* (pseudocode β snippet β full code), or | |
| - A *structured story* (Start β Development β Climax β Conclusion β End). | |
| After creation, **two AI critics** rate and explain the output, and the system automatically refines it based on their feedback β a full *AI self-improvement loop*. | |
| ## π§© Workflow | |
| 1. Input your **idea** and choose **Code mode** or **Book mode**. | |
| 2. The AI refines your prompt. | |
| 3. It builds content step-by-step. | |
| 4. Two critics analyze and rate it. | |
| 5. The main model refines the output using that feedback. | |
| ## π οΈ Models Used | |
| - Prompt refiner: `mistralai/Mixtral-8x7B-Instruct-v0.1` | |
| - Code generator: `codellama/CodeLlama-7b-Instruct-hf` | |
| - Book generator: `mistralai/Mixtral-8x7B-Instruct-v0.1` | |
| - Critics: `google/gemma-2-9b-it`, `meta-llama/Meta-Llama-3-8B-Instruct` | |
| You can modify these to lighter or open-source models depending on performance limits. | |
| ## π Deployment | |
| Push this repo to Hugging Face as a Space and select βGradioβ as the SDK. |