Embeddings multilingual-e5-large per un RAG in italiano

multilingual-e5-large per un RAG in italiano: 1024 dimensioni, 512 token, prefissi query:/passage: obbligatori, distanza coseno. Licenza MIT.

Notes

I prefissi «query: » e «passage: » sono obbligatori: senza, la ricerca peggiora molto.
Al massimo 512 token per testo: dividi i documenti in frammenti di circa 350 parole.
Normalizza i vettori e usa la distanza coseno.
Modello pubblicato da Microsoft (intfloat) con licenza MIT.

Estimated cost

Input tokens per call : ≈ 244

Input cost only, excluding the model's answer. Providers' public standard prices (Mistral AI, DeepSeek, OpenAI, Anthropic) converted from USD to euros at the ECB rate of September 29, 2026. Estimate: 1 token ≈ 3.6 characters.

Identifier

contextetech--multilingual-e5-large-rag-italiano

Type

Embeddings

File

multilingual-e5-large-rag-italiano.embedding.json

Format

JSON (application/json)

Size

679 bytes

Encoding

UTF-8

Estimated tokens

≈ 189

Language

Italian

License

MIT

Commercial use

Allowed

Personal data

None

Version

1.0

Published on

October 3, 2026 at 5:20 AM

Updated on

October 3, 2026 at 5:20 AM

Storage

in database

Uses

0

Likes

0

multilingual-e5-large-rag-italiano.embedding.json

{
  "format": "contextetech/embedding/v1",
  "id": "contextetech/multilingual-e5-large-rag-italiano",
  "description": "multilingual-e5-large per un RAG in italiano: 1024 dimensioni, 512 token, prefissi query:/passage: obbligatori, distanza coseno. Licenza MIT.",
  "tags": [
    "rag",
    "multilingue",
    "ricerca-semantica"
  ],
  "license": "MIT",
  "language": "it",
  "model": "intfloat/multilingual-e5-large",
  "dimensions": 1024,
  "maxTokens": 512,
  "distance": "cosine",
  "languages": [
    "it",
    "en",
    "fr",
    "de",
    "es",
    "pt"
  ],
  "prefixes": {
    "query": "query: ",
    "passage": "passage: "
  },
  "chunkSize": 350,
  "benchmarks": []
}

Embed (Python)

from sentence_transformers import SentenceTransformer

model = SentenceTransformer("intfloat/multilingual-e5-large")
q = model.encode(["query: your input here"], normalize_embeddings=True)
docs = model.encode(["passage: …"], normalize_embeddings=True)
print(q.shape)  # (1, 1024)
print(q @ docs.T)  # similarité