Update README.md

README.md

@@ -1,371 +1,46 @@
-# 🌟 DReamMachine: Dream A LiL(LLM) Dream
-
-**Multi-Agent LLM Orchestration System for Breakthrough Innovation Discovery via Guided Hallucination**
-
 ---
-
-## 🎯 What is DReamMachine?
-
-DReamMachine is an experimental AI system that uses "controlled hallucination" to discover breakthrough innovations. By orchestrating multiple specialized LLMs through a simulated 100-year creative journey, the system explores ideas from initial discovery through real-world challenges to mass adoption and legacy.
-
-### The Core Concept
-
-Instead of asking LLMs to solve known problems, DReamMachine encourages them to **dream** - to hallucinate creative solutions in a guided, structured way. Multiple AI agents work together:
-
-- **Dreamers** generate wild, creative ideas
-- **Analysts** evaluate feasibility and extract technical details
-- **Curators** score and select the most promising concepts
-- **The System** advances successful ideas through life stages
-
-Ideas that score high on both originality AND feasibility progress through four 25-year life stages, facing different challenges and evolving along the way.
-
+title: DReamMachine
+emoji: 🚀
+colorFrom: purple
+colorTo: yellow
+sdk: gradio
+sdk_version: "4.16.0"
+app_file: app.py
+pinned: false
+# thumbnail: (your image URL here, optional)
+short_description: "Multi-Agent LLM Dream Machine & Idea Orchestrator"
 ---
 
-## 🏗️ Architecture
-
-### The 7-Step Dream Cycle
-
-Each "dream round" executes these steps:
+# DReamMachine
 
-1. **A.1 Setup** - Initialize life stage prompt and constraints
-2. **A.2 Dream & Generate** - 3 creative LLMs generate ideas simultaneously
-3. **A.3 Log & Narrate** - Writer/Logger/Narrator refine outputs into coherent narratives
-4. **A.4 Deep Think & Verify** - Analytical LLM evaluates scientific feasibility
-5. **A.5 Curate & Grade** - Evaluation LLM scores across multiple dimensions
-6. **A.6 Data Storage** - Archive complete session to HuggingFace Dataset
-7. **A.7 Reforge Loop** - Decide: advance to next stage or start new idea?
+A private, modular research framework for automating creative, uncensored invention cycles using the latest open AI models.
 
-### Life Stages (Simulated 100-Year Lifespan)
+## 👤 Credits
 
-| Stage | Age Range | Focus | Challenge |
-|-------|-----------|-------|-----------|
-| **Init** | 1-25 | Foundational Discovery | Bold creativity, breakthrough thinking |
-| **Mid** | 26-50 | Commercialization & Crisis | Real-world market/resource/technical challenges |
-| **Late** | 51-75 | Mass Adoption & Ethics | Societal impact, unintended consequences |
-| **Final** | 76-100 | Legacy & Vision | Long-term civilizational impact, next generation |
-
-### Agent Roles
-
-| Role | Model Type | Temperature | Purpose |
-|------|------------|-------------|---------|
-| **Dreamers (3x)** | Creative/Large (Mixtral, Llama 3) | 0.85-0.9 | Generate radical ideas |
-| **Writer** | Medium (Mistral 7B) | 0.6 | Create compelling narratives |
-| **Logger** | Medium (Mistral 7B) | 0.4 | Extract technical specifications |
-| **Narrator** | Creative (Nous-Hermes) | 0.5 | Present ideas engagingly |
-| **Deep Thinker** | Large/Analytical (Llama 3 70B) | 0.3 | Evaluate feasibility |
-| **Curator** | Large/Analytical (Llama 3 70B) | 0.2 | Score and decide advancement |
-
----
-
-## 🚀 Quick Start
-
-### Prerequisites
-
-- Python 3.9+
-- HuggingFace account with API token
-- (Optional) HuggingFace Spaces Pro account for Zero GPU
-
-### Local Installation
-
-1. **Clone the repository**
-```bash
-git clone <your-repo-url>
-cd DReamMachine
-```
-
-2. **Install dependencies**
-```bash
-pip install -r requirements.txt
-```
-
-3. **Configure environment**
-```bash
-cp .env.example .env
-# Edit .env and add your HF_TOKEN
-```
-
-4. **Run the Gradio interface**
-```bash
-python app.py
-```
-
-5. **Access the interface**
-Open your browser to: `http://localhost:7860`
-
-### HuggingFace Spaces Deployment
-
-1. **Create a new Space**
-   - Go to https://huggingface.co/spaces
-   - Click "Create new Space"
-   - Choose "Gradio" as the SDK
-   - Select "Private" if you want to keep it confidential
-
-2. **Upload files**
-   - Upload all Python files, config.yaml, requirements.txt, and README.md
-   - Or connect your GitHub repository
-
-3. **Set environment variables**
-   - Go to Settings → Repository secrets
-   - Add `HF_TOKEN` with your HuggingFace API token
-
-4. **Enable Zero GPU (Pro users)**
-   - Go to Settings → Hardware
-   - Select "Zero GPU" option
-
-5. **The Space will automatically start!**
-
----
-
-## 📖 Usage Guide
-
-### Running a Single Dream Round
-
-1. Open the "Single Dream Round" tab
-2. Select a life stage (start with `init_1_25`)
-3. Click "🚀 Run Dream Round"
-4. Wait for completion (typically 2-5 minutes)
-5. Review results:
-   - **Session Summary**: Scores and next action
-   - **Narrative Pitch**: The refined idea
-   - **Technical Components**: What's needed to build it
-   - **Feasibility Report**: Scientific evaluation
-   - **Curator Scorecard**: Detailed scoring
-
-### Running Batch Mode
-
-1. Open the "Batch Mode" tab
-2. Set number of rounds (1-50)
-3. Set sleep interval between rounds
-4. Click "🚀 Run Batch Mode"
-5. View aggregate statistics and session IDs
-
-### Viewing Session History
-
-1. Open the "Session History" tab
-2. Click "🔄 Refresh History"
-3. Browse all past sessions with scores and timestamps
-
----
-
-## ⚙️ Configuration
-
-### Model Selection
-
-Edit `config.yaml` to customize models:
-
-```yaml
-models:
-  dreamers:
-    - model_id: "mistralai/Mixtral-8x7B-Instruct-v0.1"
-      temperature: 0.9
-      max_tokens: 1000
-  # ... more models
-```
-
-### Constraints
-
-Customize the constraints that guide dreaming:
-
-```yaml
-constraints:
-  physics: "Must use current or near-future physics (within 50 years)"
-  ethics: "Must solve a global humanitarian problem"
-  feasibility: "Must be achievable with existing materials"
-  scope: "Must impact at least 1 million people"
-```
-
-### Scoring Thresholds
-
-Adjust when ideas advance to the next stage:
-
-```yaml
-orchestration:
-  auto_advance_threshold:
-    feasibility_min: 7  # 1-10 scale
-    originality_min: 5  # 1-10 scale
-```
-
-### Prompt Detail Level
-
-Choose how detailed prompts are:
-
-```yaml
-prompt_detail_level: "full"  # Options: "simple", "moderate", "full"
-```
-
----
-
-## 📊 Output & Logging
-
-### Local JSON Logs
-
-All sessions are saved to `./logs/` directory:
-- Individual session files: `session_YYYYMMDD_HHMMSS_N.json`
-- Chunked archives: `chunk_YYYYMMDD_HHMMSS.json` (every 100 sessions)
-
-### HuggingFace Dataset
-
-Sessions are automatically logged to a private HuggingFace Dataset:
-- Dataset name: `{username}/dreammachine-logs`
-- Format: Structured dataset with columns for all outputs
-- Access: Private by default (configurable)
-
-### Sample Session Data Structure
-
-```json
-{
-  "session_id": "session_20250114_123045_0",
-  "timestamp": "2025-01-14T12:30:45",
-  "life_stage": "init_1_25",
-  "dream_outputs": ["dream1", "dream2", "dream3"],
-  "pitch_narrative": "The Solar Fiber Blanket...",
-  "technical_components": "1. Photovoltaic nano-fibers...",
-  "feasibility_report": "Scientific Validity: 8 - ...",
-  "curator_scorecard": {
-    "originality": 9,
-    "feasibility": 8,
-    "global_impact": 9,
-    "narrative_coherence": 8,
-    "reforge_flag": true
-  },
-  "next_action": {
-    "type": "advance",
-    "next_stage": "mid_26_50"
-  }
-}
-```
-
----
-
-## 🧪 Advanced Usage
-
-### Running from CLI
-
-You can create a CLI script to run DReamMachine:
-
-```python
-# run_dream.py
-import os
-from orchestrator import DreamOrchestrator
-
-# Initialize
-orchestrator = DreamOrchestrator(
-    config_path="config.yaml",
-    hf_token=os.getenv('HF_TOKEN')
-)
-
-# Run single round
-result = orchestrator.run_dream_round()
-
-# Or run batch
-results = orchestrator.run_batch_mode(num_rounds=10, sleep_between=30)
-
-# Or run scheduled mode (for HF Spaces)
-orchestrator.run_scheduled_mode()  # Runs until max_runtime
-```
-
-### Programmatic Access
-
-```python
-from llm_agent import LLMAgent
-from prompt_manager import PromptManager
-from data_logger import DataLogger
-
-# Initialize components individually
-agent = LLMAgent(config_path="config.yaml")
-prompts = PromptManager(config_path="config.yaml")
-logger = DataLogger(config_path="config.yaml")
-
-# Run custom workflows
-prompt = prompts.get_life_stage_prompt('init_1_25')
-dreams = agent.run_parallel_dreamers(prompt)
-# ... process as needed
-```
-
----
-
-## 🎨 Customization
-
-### Adding New Life Stages
-
-Edit `prompt_manager.py` to add new stages:
-
-```python
-def _get_custom_phase_prompt(self, previous_context):
-    return """
-    # Your custom life stage prompt here
-    """
-```
-
-### Changing Scoring Criteria
-
-Modify the curator prompt in `prompt_manager.py`:
-
-```python
-def get_curator_prompt(self, narrative, feasibility_report):
-    # Add custom scoring dimensions
-    # Modify JSON schema
-```
-
-### Using Different Models
-
-Update `config.yaml` with any HuggingFace model:
-
-```yaml
-models:
-  dreamers:
-    - model_id: "your-org/your-model"
-      temperature: 0.8
-      max_tokens: 1500
-```
-
----
-
-## 🛠️ Troubleshooting
-
-### Common Issues
-
-**"HF_TOKEN not found"**
-- Set the `HF_TOKEN` environment variable
-- Or add it to `.env` file
-- Or pass it directly: `DreamOrchestrator(hf_token="your-token")`
-
-**"Model timeout or rate limit"**
-- HuggingFace Inference API has rate limits
-- Increase `sleep_between` in batch mode
-- Consider using smaller models
-- Upgrade to HF Pro for higher limits
-
-**"JSON parsing error in curator"**
-- The curator sometimes returns malformed JSON
-- System will use default scores and continue
-- Check logs for raw output
-- Consider adjusting curator temperature
-
-**"Dataset creation failed"**
-- Ensure HF_TOKEN has write permissions
-- Check dataset name doesn't already exist
-- Set `save_to_hf_dataset: false` to disable
+- Architect/Designer: Dave Roby / DRStudios
+- Worksheet/Buildsheet Author: Gemini Flash
+- Core Collaborator: Claude (model agent)
+- Technical Assistant & Workflow Coordinator: Comet AI Assistant (Perplexity)
 
 ---
 
 ## 📁 Project Structure
 
-```
 DReamMachine/
-├── app.py                    # Gradio interface (HuggingFace Spaces entry point)
-├── orchestrator.py           # Main orchestration engine (7-step cycle)
-├── llm_agent.py             # HuggingFace API interaction layer
-├── prompt_manager.py        # Prompt templates and life stages
-├── data_logger.py           # Logging to JSON and HF Dataset
-├── config.yaml              # Configuration (models, settings, constraints)
-├── requirements.txt         # Python dependencies
-├── .env.example            # Environment variable template
-├── .gitignore              # Git ignore rules
-├── LICENSE                 # MIT License
-├── README.md               # This file
-└── logs/                   # Local JSON logs (created at runtime)
-```
+├── app.py # Gradio interface (HuggingFace Spaces entry point)
+├── orchestrator.py # Main orchestration engine (7-step cycle)
+├── llm_agent.py # HuggingFace API interaction layer
+├── prompt_manager.py # Prompt templates and life stages
+├── data_logger.py # Logging to JSON and HF Dataset
+├── config.yaml # Configuration (models, settings, constraints)
+├── requirements.txt # Python dependencies
+├── .env.example # Environment variable template
+├── .gitignore # Git ignore rules
+├── LICENSE # MIT License
+├── README.md # This file
+└── logs/ # Local JSON logs (created at runtime)
+
+text
 
 ---
 
@@ -391,24 +66,6 @@ DReamMachine/
 
 ---
 
-## 📝 License
-
-MIT License - see [LICENSE](LICENSE) file
-
----
-
-## 🙏 Credits
-
-**Created by**: Dave Roby / DRStudios
-
-**Concept Inspired by**: Conversations with Gemini 2.5 Flash about controlled hallucination, multi-agent systems, and guided creativity
-
-**Built with**: Claude Sonnet 4.5 via Claude Code
-
-**Original Vision**: The idea of making LLMs "dream" came from wanting to discover innovations that don't exist yet but COULD exist - by guiding creative hallucination through simulated lifespans with multiple specialized agents.
-
----
-
 ## 🔮 Future Enhancements
 
 - [ ] Add visualization dashboard for session analytics
@@ -427,8 +84,8 @@ MIT License - see [LICENSE](LICENSE) file
 
 For questions, suggestions, or collaboration:
 - Open an issue on GitHub
-- Contact: Dave Roby / DRStudios
+- Contact: Dave Roby / DRStudios: Drdrproduction@gmail.com
 
 ---
 
-**"Dream a little dream... let the LLMs imagine the future."** 🌟
+**"Dream a little dream... let the LLMs imagine the future."** 🌟