Project Summary
R&D sandbox for generative AI experiments — text and image generation, prompt engineering, and fine-tuning prototypes that informed production repos like DocuMind and AutoFlow.
Technical deep dive
The Generative AI repository collects experiments across text generation, image synthesis, prompt engineering, and early transformer workflows — a sandbox where new techniques get tested before graduating into production repos like DocuMind, AutoFlow, and the Google ADK Portfolio. For SEO around generative AI experiments GitHub, prompt engineering portfolio, and deep learning creative AI, this repo shows breadth and continuous learning rather than a single demo.
Experiment categories
- Text generation with transformer models and prompt templates
- Image generation experiments (Stable Diffusion, GAN variants where applicable)
- Fine-tuning and LoRA adaptation prototypes
- Prompt chaining and multi-step generation workflows
- Evaluation notebooks comparing model outputs across temperature and top-p settings
- Utility scripts for dataset curation and output post-processing
From experiments to production
Not every experiment ships to production — that is the point. This repo captures the R&D layer: quick hypothesis tests, failed approaches documented honestly, and patterns that later appear in hardened systems. Local-first inference patterns tested here informed DocuMind's Ollama stack. Multi-step agent behaviors explored here evolved into LangGraph workflows in AutoFlow and Google ADK transfer_to_agent delegation.
Tech stack
| Area | Tools |
|---|---|
| LLM / text | Hugging Face Transformers, LangChain |
| Image gen | Diffusers, various GAN implementations |
| Notebooks | Jupyter Lab |
| Language | Python 3.x |
Explore the repo
git clone https://github.com/cdtalley/Generative-AI
cd Generative-AI
pip install -r requirements.txt
jupyter labKey Features & Capabilities
- Text generation with transformer models and prompt templates
- Image generation experiments with diffusers and GAN variants
- Fine-tuning and LoRA adaptation prototypes
- Evaluation notebooks comparing outputs across generation parameters
Tech Stack & Components
Getting Started
1.Explore notebooks
Clone and launch Jupyter Lab.
git clone https://github.com/cdtalley/Generative-AI
pip install -r requirements.txt
jupyter labFrequently asked questions
- What is in the Generative AI repository?
- A collection of generative AI experiments: text generation, image synthesis, prompt engineering, and fine-tuning prototypes in Jupyter notebooks and Python scripts.
- Are these production-ready systems?
- No — this is an R&D sandbox. Production GenAI systems (DocuMind, AutoFlow, Google ADK Portfolio) have dedicated architecture articles with security models, audit trails, and deployment guides on draketalley.ai/blog.
- Do experiments require cloud GPU?
- Some image generation experiments benefit from GPU. Text experiments often run on CPU with smaller models. Each notebook README notes hardware expectations.
- How does this repo connect to Drake Talley's portfolio?
- It documents the experimentation layer behind seven featured production AI repos. Techniques proven here graduate into repos with FastAPI backends, operator UIs, and full SEO architecture articles.