Model reference · open weights

neodictabert-bilingual-embed

Embeddings dicta-il Embeddings 1 build Open weights 3k dl/mo

neodictabert-bilingual-embed is an open-weight embedding model from dicta-il. neodictabert-bilingual-embed (BF16) weighs 725 MB; the smallest configuration that runs it is RTX 3060 12 GB.

For self-hosting, compare this build's requirements below with the GPU you plan to rent.

  • neodictabert-bilingual-embed is a 363M parameter sentence-transformers model by dicta-il designed for sentence-similarity tasks in Hebrew and English.
  • It maps text to a 768-dimensional vector space for semantic search and clustering, and achieved rank 10 in the private phase of the Hebrew Semantic Retrieval National Challenge.
  • The model is released under the CC BY 4.0 license.

Summary of the dicta-il/neodictabert-bilingual-embed model card, 2026-10-01

What it is

Released bydicta-il
Released2026-02-02
Parameters363M
VRAM725 MB for the weights

What it runs on

Memory and cards for neodictabert-bilingual-embed (BF16)

725 MBweights, file size
1.1 GBruntime overhead
CardRunsMemory
RTX 3060 12 GBfits11.6 GB
RTX 4060 Ti 16 GBfits15.4 GB
RTX 3090 24 GBfits23.4 GB
RTX 4090 24 GBfits23.4 GB
RTX 5090 32 GBfits31.0 GB
L40S 48 GBfits44.0 GB
A100 80 GBfits78.2 GB
H100 80 GBfits78.1 GB
RTX PRO 6000 Blackwell 96 GBfits93.8 GB
DGX Spark (GB10) 128 GB unifiedfits107 GB
H200 141 GBfits138 GB
B200 180 GBfits176 GB

From the model card

What dicta-il says about neodictabert-bilingual-embed

Read the model card

This is a sentence-transformers model finetuned from dicta-il/neodictabert-bilingual on the he dataset. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.

This model achieved #10 on the private phase of the Hebrew Semantic Retrieval National Challenge.

Usage

Direct Usage (Sentence Transformers)

First install the Sentence Transformers library:

pip install -U sentence-transformers

Then you can load this model and run inference.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("dicta-il/neodictabert-bilingual-embed", trust_remote_code=True)
# Run inference
queries = [
    "query: מתכון למיונז ביתי (חלמון, שמן, חרדל, לימון) + הוראות הכנה",
]

documents = [
    "מיונז ביתי. מרכיבים: חלמון בטמפרטורת החדר, חרדל דיז'ון, מיץ לימון/חומץ, מלח, שמן ניטרלי. הכנה: טורפים חלמון+חרדל+מלח+לימון, מזלפים שמן בהדרגה תוך טריפה עד להסמכה (אמולסיה).",
    "ים המלח. עובדות: זהו המקום הנמוך ביותר על פני היבשה, המליחות בו גבוהה בהרבה מהאוקיינוס ולכן אנשים צפים בקלות. בוץ עשיר במינרלים משמש גם לקוסמטיקה.",
    "כתב יתדות. היסטוריה: מסופוטמיה/שומר, חריתה בלוחות טיט בעזרת קנה. התפתח מאידיאוגרמות לייצוג פונטי והאפשר ניהול ביורוקרטי ושימור חוקים וידע.",
    "פסטה ברוטב עגבניות. מרכיבים: פסטה, עגבניות, שום, שמן זית, מלח. הכנה: מבשלים פסטה ומכינים רוטב עגבניות.",
]

query_embeddings = model.encode_query(queries)
document_embeddings = model.encode_document(documents)
print(query_embeddings.shape, document_embeddings.shape)
# [1, 768] [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(query_embeddings, document_embeddings)
print(similarities)
# tensor([[ 0.2235,  0.0164,  0.0822, -0.0282]])

Citation

If you use NeoDictaBERT in your research, please cite NeoDictaBERT: Pushing the Frontier of BERT models for Hebrew

BibTeX:

@misc{shmidman2025neodictabertpushingfrontierbert,
      title={NeoDictaBERT: Pushing the Frontier of BERT models for Hebrew},
      author={Shaltiel Shmidman and Avi Shmidman and Moshe Koppel},
      year={2025},
      eprint={2510.20386},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2510.20386},
}

License

Shield: CC BY 4.0

This work is licensed under a Creative Commons Attribution 4.0 International License.

CC BY 4.0

Quoted from the model card on Hugging Face. The full card is behind the Hugging Face link above.

© 2026 AxForge · EU-hosted AI infrastructure Pricing Docs Trust Privacy Terms