shootstuff's picture
Create a web app called 'AI Phenomenology Tester' using Python backend with Streamlit. Include: - User input for selecting an HF model (e.g., Llama-2 via Transformers). - An agentic workflow: Use a simple AI agent (like LangChain or HF Agents) to plan and run tests. - Phenomenology test: Prompt the model with scenarios (e.g., 'What is consciousness?') and analyze emergent behaviors (e.g., entropy or pattern matching). - Eigenmode test: Extract activations, compute eigenvalues/eigenvectors with SciPy, visualize dominant modes (e.g., for physical signals like vibrations). - Dashboard to display results, charts (Matplotlib), and export options. - Integrate Hugging Face API for model loading. Make it deployable on HF Spaces.
b59fe0b verified
raw
history blame contribute delete
805 Bytes
@import url('https://fonts.googleapis.com/css2?family=Inter:wght@300;400;500;600;700&display=swap');
body {
font-family: 'Inter', sans-serif;
transition: background-color 0.3s ease;
}
/* Dark mode toggle transition */
.dark-mode-transition {
transition: background-color 0.3s ease, color 0.3s ease;
}
/* Custom scrollbar */
::-webkit-scrollbar {
width: 8px;
height: 8px;
}
::-webkit-scrollbar-track {
background: #f1f1f1;
border-radius: 10px;
}
::-webkit-scrollbar-thumb {
background: #888;
border-radius: 10px;
}
::-webkit-scrollbar-thumb:hover {
background: #555;
}
.dark ::-webkit-scrollbar-track {
background: #2d3748;
}
.dark ::-webkit-scrollbar-thumb {
background: #4a5568;
}
.dark ::-webkit-scrollbar-thumb:hover {
background: #718096;
}