Build, buy or fine-tune? How companies choose their AI approach
Off-the-shelf assistants, API integrations, retrieval-augmented generation, fine-tuning or your own model: the trade-offs in cost, control and time to value.
AI Desk3 min read
Tag
#Rag
Every The Latent Daily story, tool, model and resource tagged “Rag”.
7 items
How retrieval-augmented generation (RAG) works, step by step
RAG lets AI answer from your documents. Each stage explained: ingestion, chunking, embeddings, retrieval, reranking, prompting and evaluation, with practical tips.
AI Desk3 min read
Vector databases explained: how they work and how to choose one
Embeddings, similarity search, indexes like HNSW and the trade-offs between pgvector, Pinecone, Weaviate, Qdrant, Milvus and Chroma for your RAG project.
AI Desk3 min read
How to launch an AI customer support chatbot: a step-by-step guide
Plan, build, test and launch an AI support assistant: choosing use cases, preparing the knowledge base, setting guardrails and escalation, and measuring results.
AI Desk4 min read
Quick take: bigger context windows are not always better
Models now accept huge inputs, but stuffing everything into the prompt can reduce accuracy and raise costs. When long context helps, and when retrieval works better.
AI Desk3 min read
Models tagged #Rag
BAAIJan 30, 2024
BGE-M3
Open multilingual embedding model from the Beijing Academy of AI supporting dense, sparse and multi-vector retrieval in over 100 languages.
- Type
- Embedding
- Context
- 8K
- Weights
- Open
OpenAIJan 25, 2024
text-embedding-3-large
OpenAI's large embedding model producing vectors of up to 3,072 dimensions for search, retrieval and clustering, with adjustable dimensions.
- Type
- Embedding
- Context
- 8K
- Weights
- Closed