Close Menu
    Facebook X (Twitter) Instagram
    • Privacy Policy
    • Terms Of Service
    • Social Media Disclaimer
    • DMCA Compliance
    • Anti-Spam Policy
    Facebook X (Twitter) Instagram
    Bytecore News
    • Home
    • Crypto News
      • Bitcoin
      • Ethereum
      • Altcoins
      • Blockchain
      • DeFi
    • AI News
    • Stock News
    • Learn
      • AI for Beginners
      • AI Tips
      • Make Money with AI
    • Reviews
    • Tools
      • Best AI Tools
      • Crypto Market Cap List
      • Stock Market Overview
      • Market Heatmap
    • Contact
    Bytecore News
    Home»AI News»How to Build a Matryoshka-Optimized Sentence Embedding Model for Ultra-Fast Retrieval with 64-Dimension Truncation
    How to Build a Matryoshka-Optimized Sentence Embedding Model for Ultra-Fast Retrieval with 64-Dimension Truncation
    AI News

    How to Build a Matryoshka-Optimized Sentence Embedding Model for Ultra-Fast Retrieval with 64-Dimension Truncation

    February 12, 20265 Mins Read
    Share
    Facebook Twitter LinkedIn Pinterest Email
    synthesia


    In this tutorial, we fine-tune a Sentence-Transformers embedding model using Matryoshka Representation Learning so that the earliest dimensions of the vector carry the most useful semantic signal. We train with MatryoshkaLoss on triplet data and then validate the key promise of MRL by benchmarking retrieval quality after truncating embeddings to 64, 128, and 256 dimensions. At the end, we save the tuned model and demonstrate how to load it with a small truncate_dim setting for fast and memory-efficient vector search. Check out the FULL CODES here.

    !pip -q install -U sentence-transformers datasets accelerate

    import math
    import random
    import numpy as np
    import torch

    from datasets import load_dataset
    from torch.utils.data import DataLoader

    notion

    from sentence_transformers import SentenceTransformer, InputExample
    from sentence_transformers import losses
    from sentence_transformers.util import cos_sim

    def set_seed(seed=42):
    random.seed(seed)
    np.random.seed(seed)
    torch.manual_seed(seed)
    torch.cuda.manual_seed_all(seed)

    set_seed(42)

    We install the required libraries and import all the necessary modules for training and evaluation. We set a deterministic seed, so our sampling and training behavior stay consistent across runs. We also ensure PyTorch and CUDA RNGs are aligned when a GPU is available. Check out the FULL CODES here.

    @torch.no_grad()
    def retrieval_metrics_mrr_recall_at_k(
    model,
    queries,
    corpus,
    qrels,
    dims_list=(64, 128, 256, None),
    k=10,
    batch_size=64,
    ):
    device = “cuda” if torch.cuda.is_available() else “cpu”
    model.to(device)

    qids = list(queries.keys())
    docids = list(corpus.keys())

    q_texts = [queries[qid] for qid in qids]
    d_texts = [corpus[did] for did in docids]

    q_emb = model.encode(q_texts, batch_size=batch_size, convert_to_tensor=True, normalize_embeddings=True)
    d_emb = model.encode(d_texts, batch_size=batch_size, convert_to_tensor=True, normalize_embeddings=True)

    results = {}

    for dim in dims_list:
    if dim is None:
    qe = q_emb
    de = d_emb
    dim_name = “full”
    else:
    qe = q_emb[:, :dim]
    de = d_emb[:, :dim]
    dim_name = str(dim)
    qe = torch.nn.functional.normalize(qe, p=2, dim=1)
    de = torch.nn.functional.normalize(de, p=2, dim=1)

    sims = cos_sim(qe, de)

    mrr_total = 0.0
    recall_total = 0.0

    for i, qid in enumerate(qids):
    rel = qrels.get(qid, set())
    if not rel:
    continue

    topk = torch.topk(sims[i], k=min(k, sims.shape[1]), largest=True).indices.tolist()
    topk_docids = [docids[j] for j in topk]

    recall_total += 1.0 if any(d in rel for d in topk_docids) else 0.0

    rr = 0.0
    for rank, d in enumerate(topk_docids, start=1):
    if d in rel:
    rr = 1.0 / rank
    break
    mrr_total += rr

    denom = max(1, len(qids))
    results[dim_name] = {f”MRR@{k}”: mrr_total / denom, f”Recall@{k}”: recall_total / denom}

    return results

    def pretty_print(results, title):
    print(“\n” + “=” * 80)
    print(title)
    print(“=” * 80)
    for dim, metrics in results.items():
    print(f”dim={dim:>4} | ” + ” | “.join([f”{k}={v:.4f}” for k, v in metrics.items()]))

    We implement a lightweight retrieval evaluator that encodes queries and documents, computes cosine similarity, and reports MRR@10 and Recall@10. We re-normalize embeddings after truncation so smaller prefixes remain comparable in cosine space. We also added a compact printer to make before/after comparisons easy to read. Check out the FULL CODES here.

    DATASET_ID = “sentence-transformers/msmarco-co-condenser-margin-mse-sym-mnrl-mean-v1”
    SUBSET = “triplet-hard”
    SPLIT = “train”

    TRAIN_SAMPLES = 4000
    EVAL_QUERIES = 300

    stream = load_dataset(DATASET_ID, SUBSET, split=SPLIT, streaming=True)

    train_examples = []
    eval_queries = {}
    eval_corpus = {}
    eval_qrels = {}

    doc_id_counter = 0
    qid_counter = 0

    for row in stream:
    q = (row.get(“query”) or “”).strip()
    pos = (row.get(“positive”) or “”).strip()
    neg = (row.get(“negative”) or “”).strip()

    if not q or not pos or not neg:
    continue

    train_examples.append(InputExample(texts=[q, pos, neg]))

    if len(eval_queries) < EVAL_QUERIES:
    qid = f”q{qid_counter}”
    qid_counter += 1

    pos_id = f”d{doc_id_counter}”; doc_id_counter += 1
    neg_id = f”d{doc_id_counter}”; doc_id_counter += 1

    eval_queries[qid] = q
    eval_corpus[pos_id] = pos
    eval_corpus[neg_id] = neg
    eval_qrels[qid] = {pos_id}

    if len(train_examples) >= TRAIN_SAMPLES and len(eval_queries) >= EVAL_QUERIES:
    break

    print(len(train_examples), len(eval_queries), len(eval_corpus))

    We stream a mined MS MARCO triplet dataset and build both a training set (queries, positives, negatives) and a tiny IR benchmark set. We map each query to a relevant positive document and include a negative document to make retrieval meaningful. We stop early to keep the run Colab-friendly while still large enough to show truncation effects.

    MODEL_ID = “BAAI/bge-base-en-v1.5”

    device = “cuda” if torch.cuda.is_available() else “cpu”
    model = SentenceTransformer(MODEL_ID, device=device)
    full_dim = model.get_sentence_embedding_dimension()

    baseline = retrieval_metrics_mrr_recall_at_k(
    model,
    queries=eval_queries,
    corpus=eval_corpus,
    qrels=eval_qrels,
    dims_list=(64, 128, 256, None),
    k=10,
    )
    pretty_print(baseline, “BEFORE”)

    We load a strong base embedding model and record its full embedding dimension. We run the baseline evaluation across 64/128/256/full dimensions to see how truncation behaves before any training. We print the results so we can later compare whether MRL improves the early-dimension quality.

    batch_size = 16
    epochs = 1
    warmup_steps = 100

    train_loader = DataLoader(train_examples, batch_size=batch_size, shuffle=True, drop_last=True)

    base_loss = losses.MultipleNegativesRankingLoss(model=model)

    mrl_dims = [full_dim, 512, 256, 128, 64] if full_dim >= 768 else [full_dim, 256, 128, 64]
    mrl_loss = losses.MatryoshkaLoss(
    model=model,
    loss=base_loss,
    matryoshka_dims=mrl_dims
    )

    model.fit(
    train_objectives=[(train_loader, mrl_loss)],
    epochs=epochs,
    warmup_steps=warmup_steps,
    show_progress_bar=True,
    )

    after = retrieval_metrics_mrr_recall_at_k(
    model,
    queries=eval_queries,
    corpus=eval_corpus,
    qrels=eval_qrels,
    dims_list=(64, 128, 256, None),
    k=10,
    )
    pretty_print(after, “AFTER”)

    out_dir = “mrl-msmarco-demo”
    model.save(out_dir)

    m64 = SentenceTransformer(out_dir, truncate_dim=64)
    emb = m64.encode(
    [“what is the liberal arts?”, “liberal arts covers humanities and sciences”],
    normalize_embeddings=True
    )
    print(emb.shape)

    We create a MultipleNegativesRankingLoss and wrap it with MatryoshkaLoss using a descending list of target prefix dimensions. We fine-tune the model on the triplets, then re-run the same truncation benchmark to measure the improvement in retention. Also, we save the model and reload it with truncate_dim=64 to confirm practical usage for compact retrieval.

    In conclusion, we successfully trained a Matryoshka-optimized embedding model that maintains strong retrieval performance even when we truncate vectors to small prefix dimensions, such as 64. We verified the effect by comparing baseline versus post-training retrieval metrics across multiple truncation sizes and the full embedding. With the saved model and the truncate_dim loading pattern, we now have a clean workflow for building smaller, faster vector indexes while keeping the option to rerank with full-dimensional embeddings.

    Check out the FULL CODES here. Also, feel free to follow us on Twitter and don’t forget to join our 100k+ ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.



    Source link

    quillbot
    Share. Facebook Twitter Pinterest LinkedIn Tumblr Email
    CryptoExpert
    • Website

    Related Posts

    How to Build and Evolve a Custom OpenAI Agent with A-Evolve Using Benchmarks, Skills, Memory, and Workspace Mutations

    March 31, 2026

    MIT researchers use AI to uncover atomic defects in materials | MIT News

    March 30, 2026

    When product managers ship code: AI just broke the software org chart

    March 29, 2026

    RPA matters, but AI changes how automation works

    March 28, 2026
    Add A Comment
    Leave A Reply Cancel Reply

    notion
    Latest Posts

    What Does ETH Need to Break Out of Consolidation?

    April 1, 2026

    Ripple’s RLUSD Stablecoin Sits On $1.57 Billion In Reserves: Audit Firm

    April 1, 2026

    Crypto Market‑Structure Bill Now A Long Shot — TD Cowen Puts 2026 Approval At One‑Third

    April 1, 2026

    CoinShares Stock Debuts on Nasdaq After $1.2B SPAC Deal

    April 1, 2026

    Ethereum price approaches $2,200 as Iran signals willingness to end war

    April 1, 2026
    murf
    LEGAL INFORMATION
    • Privacy Policy
    • Terms Of Service
    • Social Media Disclaimer
    • DMCA Compliance
    • Anti-Spam Policy
    Top Insights

    FREE AI Tools To Create Videos & Images 😳🔥 (Full Beginner Tutorial 2026)

    April 1, 2026

    Crypto-Revenge ‘On Demand’ – Why Are Rogue Groups Taking Justice On Their Own Hands?

    April 1, 2026
    aistudios
    Facebook X (Twitter) Instagram Pinterest
    © 2026 BytecoreNews.com - All rights reserved.

    Type above and press Enter to search. Press Esc to cancel.