---
license: apache-2.0
base_model: Alibaba-NLP/gte-modernbert-base
library_name: onnxruntime
pipeline_tag: sentence-similarity
tags:
  - onnx
  - sentence-similarity
  - embeddings
  - feature-extraction
  - code-search
  - ziv
  - code-aware
language:
  - en
---

# Ziv Embedder — Code Aware (ONNX)

This is an ONNX export of **[Alibaba-NLP/gte-modernbert-base](https://huggingface.co/Alibaba-NLP/gte-modernbert-base)**, prepared for **[Ziv](https://github.com/Muhd-Uwais/ziv)** — a local semantic code search engine for Python repositories.

Ziv uses this embedder to improve code-aware search quality in version **0.4.0**. The model is optimized for local inference with **onnxruntime**, making it lightweight, fast, and practical for offline developer workflows.

## Why this model?

Ziv needs embeddings that work well for code search and code understanding while staying fully local. This model is designed to support that goal with:

- **Code-aware semantic search**
- **Fast local inference**
- **No cloud dependency**
- **No API keys**
- **ONNX runtime compatibility**

Compared to a standard Python-based embedding stack, this setup is easier to ship and more efficient to run inside a local developer tool.

## Usage with Ziv

```bash
ziv init --model code
ziv start
```

## Model details

| Property | Value |
|---|---|
| Base model | Alibaba-NLP/gte-modernbert-base |
| Model type | Text embedding |
| Embedding dimension | 768 |
| Max sequence length | 8192 |
| Runtime | onnxruntime |
| Primary use | Semantic code search / code understanding |

## Files

| File | Description |
|---|---|
| `model.onnx` | ONNX model weights and graph |
| `tokenizer.json` | Tokenizer vocabulary and rules |
| `tokenizer_config.json` | Tokenizer settings |
| `config.json` | Model architecture config |
| `1_Pooling/config.json` | Pooling configuration |

## Relation to the original model

This model is based on **[Alibaba-NLP/gte-modernbert-base](https://huggingface.co/Alibaba-NLP/gte-modernbert-base)**, developed by **Tongyi Lab, Alibaba Group**.

This repository does **not** claim ownership of the original model weights or training recipe. It provides an ONNX-exported runtime version tailored for Ziv and local inference.

The original model and its concepts should be credited to:
- **Tongyi Lab, Alibaba Group**
- **The gte-modernbert model authors**
- **The broader Sentence Transformers ecosystem**

## License

This model is released under the **Apache 2.0 License**, consistent with the upstream model license.

Original model: **[Alibaba-NLP/gte-modernbert-base](https://huggingface.co/Alibaba-NLP/gte-modernbert-base)**

## Citation

If you use this model or the upstream base model in your work, please cite the original paper:

```bibtex
@inproceedings{zhang2024mgte,
  title={mGTE: Generalized Long-Context Text Representation and Reranking Models for Multilingual Text Retrieval},
  author={Zhang, Xin and Zhang, Yanzhao and Long, Dingkun and Xie, Wen and Dai, Ziqi and Tang, Jialong and Lin, Huan and Yang, Baosong and Xie, Pengjun and Huang, Fei and others},
  booktitle={Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: Industry Track},
  pages={1393--1412},
  year={2024}
}

@article{li2023towards,
  title={Towards general text embeddings with multi-stage contrastive learning},
  author={Li, Zehan and Zhang, Xin and Zhang, Yanzhao and Long, Dingkun and Xie, Pengjun and Zhang, Meishan},
  journal={arXiv preprint arXiv:2308.03281},
  year={2023}
}
```