Hitung vektor dokumen dan kueri dengan openai/text-embedding-3-large
Ambil Top-K dari toko vektor (FAISS / Milvus / pgvector)
Peringkat ulang dengan cohere/rerank-english-v3
Jahit potongan Top-N ke dalam prompt dan biarkan cohere/command-r-plus atau Claude menghasilkan jawaban akhir
Penyematan
python
resp = client.embeddings.create(
model="openai/text-embedding-3-large",
input=[chunk.text for chunk in chunks],
)
vectors = [d.embedding for d in resp.data]
Templat cepat
text
You are an expert assistant. Answer based ONLY on context.
If the answer is not in the context, say "I don't know".
[Context]
{retrieved_chunks}
[Question]
{user_question}