Update index.html

index.html

@@ -1185,7 +1185,7 @@ const SAMPLES = [
   { content: "Retrieval-Augmented Generation (RAG) is an AI framework that combines retrieval of relevant documents with generative language models. It allows LLMs to access external knowledge bases to produce more accurate, up-to-date, and factually grounded responses. RAG reduces hallucination by grounding responses in retrieved context.", source: "rag_guide.md", category: "technical", metadata: { topic: "RAG", author: "AI Team" }},
   { content: "Vector databases store data as high-dimensional vectors (embeddings) that capture semantic meaning. Popular vector DBs include LanceDB, ChromaDB, Qdrant, Milvus, and Pinecone. They enable efficient similarity search using algorithms like HNSW, IVF, or brute-force cosine similarity for nearest neighbor retrieval.", source: "vector_db_comparison.md", category: "documentation", metadata: { topic: "VectorDB", author: "Data Team" }},
   { content: "Cross-encoder reranking improves retrieval accuracy by scoring query-document pairs directly. Unlike bi-encoders that compute query and document embeddings independently, cross-encoders process both together through a transformer, capturing fine-grained interactions. The ms-marco-MiniLM-L-6-v2 model is a lightweight yet effective reranker.", source: "reranking_guide.md", category: "technical", metadata: { topic: "Reranking", author: "ML Team" }},
-  { content: "Neural networks are computing systems inspired by biological neural networks. They consist of interconnected nodes (neurons) organized in layers: input, hidden, and output. Deep learning uses networks with many hidden layers to learn hierarchical representations of data, enabling breakthroughs in image recognition, NLP, and game playing.", source: "neural_networks.md", category: "technical", meta { topic: "Deep Learning", author: "Research Team" }},
+  { content: "Neural networks are computing systems inspired by biological neural networks. They consist of interconnected nodes (neurons) organized in layers: input, hidden, and output. Deep learning uses networks with many hidden layers to learn hierarchical representations of data, enabling breakthroughs in image recognition, NLP, and game playing.", source: "neural_networks.md", category: "technical", metadata: { topic: "Deep Learning", author: "Research Team" }},
   { content: "Modern web development involves frameworks like React, Vue, Angular, and Svelte for frontend, and Node.js, Python, Go, or Rust for backend. Static site generators (SSG) like Next.js, Astro, and Hugo enable fast, SEO-friendly websites. WebAssembly (WASM) allows running compiled languages in the browser for near-native performance.", source: "webdev_overview.md", category: "documentation", metadata: { topic: "WebDev", author: "Frontend Team" }},
   { content: "Cybersecurity best practices include: implementing zero-trust architecture, using multi-factor authentication, regular security audits, encrypting data at rest and in transit, keeping software updated, conducting penetration testing, and following the principle of least privilege. OWASP Top 10 provides guidance on common web vulnerabilities.", source: "security_guide.md", category: "faq", metadata: { topic: "Security", author: "SecOps Team" }}
 ];