🚀 Step 10: Quick Start - First Experiment

Run your first experiment in 5 minutes!

---

🔄 Experiment Lifecycle

Each experiment follows this lifecycle:

---

✅ Prerequisites Check

Before starting, ensure you've completed:

• [x] Hardware detection (scripts/detect_hardware.py)
• [x] Environment detection (scripts/detect_environment.py)
• [x] Database setup (scripts/setup_fresh_db.py)
• [x] Model configuration (API keys in .env)
• [x] LLM verification (test_llm_setup.py)

---

⚡ Quick Start Commands

1. List Available Tasks

python -m core.execution.tests.run_experiment --list-tasks

2. Run a Simple Local Experiment

python -m core.execution.tests.run_experiment \
    --tasks gsm8k_basic \
    --repetitions 1 \
    --providers local \
    --save-db \
    --verbose

3. View Results

# Check the database
sqlite3 data/experiments.db "SELECT run_id, workflow_type, dynamic_energy_uj/1e6 as energy_j FROM runs LIMIT 5;"

# Launch the GUI
streamlit run streamlit_app.py

4. Generate a PDF Report

1. Open browser at http://localhost:8501
2. Navigate to Session Analysis
3. Find your experiment
4. Click Generate PDF Report

---

📊 What You'll See

Console Output:

📊 Progress: 1/1 runs
  Linear: 1.2043 J
  Agentic: 2.5945 J
  Tax: 2.15x
✅ Pair 1 saved (linear: 1, agentic: 2)

GUI Dashboard: - Energy comparison charts - Orchestration tax analysis - Thermal profiles - Sustainability metrics

---

🧪 Try Another Task

python -m core.execution.tests.run_experiment \
    --tasks gsm8k_multi_step \
    --repetitions 3 \
    --providers local \
    --save-db

---

🔧 Quick Troubleshooting

Issue	Solution
Permission denied	sudo ./scripts/fix_permissions.sh
No results in DB	Did you use --save-db?
Blank GUI	Check streamlit is installed
No diagrams	Run python scripts/tools/generate_diagrams.py

---

✅ Next Steps

• Understanding Metrics
• Running Batch Experiments
• Using the GUI

---

Congratulations! You've successfully run your first A-LEMS experiment! 🎉