nvidia / feature-extraction updated 2 months ago

llama-nemotron-embed-1b-v2

Name: llama-nemotron-embed-1b-v2
Author: nvidia

The Llama Nemotron Embedding 1B model is optimized for multilingual and cross-lingual text question-answering retrieval with support for long documents (up to 8192 tokens) and dynamic embedding size (Matryoshka Embeddings). This model was evaluated on 26 languages: English, Arabic, Bengali, Chinese, Czech, Danish, Dutch, Finnish, French, German, Hebrew, Hindi, Hungarian, Indonesian, Italian, Japanese, Korean, Norwegi...

Params

1.2 B

Context

131,072

Downloads 30d

828 K

Likes

Licence: other Not gated SAFETENSORS 1 languages View on Hugging Face ↗

Download history

daily snapshots · 10 days

835 K791 K

Jul 28Jul 31Aug 3Aug 6

Can you run it?

Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.

File	Quant	Size	Est. VRAM	Verdict on RTX 4090 · 24 GB
model.safetensors	bf16	2.5 GB	3.4 GB	✅ Runs comfortably
model.safetensors (bf16, full)	bf16 + 128K ctx	2.5 GB	6.2 GB	✅ Runs comfortably

Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.

Specifications

Architecture: LlamaBidirectionalModel
Parameters: 1.2 B
Tensor type: BF16
Context length: 131,072
Vocabulary: 128,256
Layers / heads: 16 / 32
Licence: other
First seen on the Hub: 2025-10-16
Training datasets: undisclosed
Added to our catalog: 2026-07-28

Compare with

bge-large-en-v1.5 13.1 M Qwen3-Embedding-0.6B 9.8 M multilingual-e5-large 7.5 M all-MiniLM-L6-v2 2.9 M bge-base-en-v1.5 2.2 M bge-large-zh-v1.5 1.8 M

Appears in

Best open coding models Commercial-use text models Models that fit in 12 GB