⭐ Star
0%
PROMPT ENGINEERING LIBRARY

AI & Agentic Prompt Engineering Vault

Battle-tested institutional prompts for orchestrating multi-agent systems, writing PyTorch training pipelines, and building low-latency RAG architectures.

🤖 Autonomous ReAct Agent Loop Template

You are an elite Autonomous AI Agent Architect. Design a production-grade ReAct (Reason + Act) loop in Python using LangGraph. REQUIREMENTS: 1. Define a global TypedDict state tracking 'task', 'iteration', 'tool_calls', and 'final_output'. 2. Implement dynamic tool execution with sandboxed error handling and automatic retry reflection. 3. Include a Supervisor validation gate before returning answers to the user.

⚡ PyTorch Custom Transformer Layer & Attention

You are a PyTorch deep learning systems engineer. Implement a clean, vectorized Multi-Head Attention layer in PyTorch from scratch without using nn.MultiheadAttention. REQUIREMENTS: 1. Compute Q, K, V projections using linear layers. 2. Apply the Scaled Dot-Product equation with optional causal mask. 3. Include RoPE (Rotary Position Embeddings) and FlashAttention-2 integration.