How akio-236/dreamweaver_ai works
Dreamweaver AI’s implemented workflow creates structured creative prompts, uses Gemini to generate stories from prompts, and cleans story data into a training-oriented format. The cleaner expects data/raw/dataset.jsonl, while the generator writes dataset.jsonl; the tree does not show the expected raw input or prompt file, so those storage links are not asserted. Fine-tuning and a user-facing web app are roadmap items, not implemented components in the sampled source.
Main entry points: story_generation.py Story generator · prompt_generation.py Prompt generator · data_cleaning.py Data cleaner · cleaned_stories.jsonl Cleaned stories
Components 8
Prompt creation
- Prompt generator
src/prompt_generation.py - Prompt list
Story generation
- Story generator
src/story_generation.py - Generated stories
Dataset preparation
- Data cleaner
src/data_cleaning.py - Cleaned stories
data/processed/cleaned_stories.jsonl
Other components
- Creative director
- Gemini API
Connections 7
- Creative director → Prompt generator: provides ideas
- Prompt generator → Prompt list: writes prompts
- Story generator → Prompt list: loads prompts
- Story generator → Gemini API: requests stories
- Gemini API → Story generator: returns stories
- Story generator → Generated stories: writes stories
- Data cleaner → Cleaned stories: writes cleaned data