---
license: mit
language: [multilingual]
tags: [embeddings, gguf, ggml, text-embeddings, xlm-r, crispembed]
pipeline_tag: feature-extraction
base_model: intfloat/multilingual-e5-large
---

# multilingual-e5-large GGUF

GGUF format of [intfloat/multilingual-e5-large](https://huggingface.co/intfloat/multilingual-e5-large) for use with [CrispEmbed](https://github.com/CrispStrobe/CrispEmbed).

Multilingual E5 Large. 100+ languages, 1024-dimensional mean-pooled. Top MTEB multilingual scorer. Use prefix: "query: " / "passage: ".

## Files

| File | Quantization | Size |
|------|-------------|------|
| [multilingual-e5-large-q4_k.gguf](https://huggingface.co/cstr/multilingual-e5-large-GGUF/resolve/main/multilingual-e5-large-q4_k.gguf) | Q4_K | 429 MB |
| [multilingual-e5-large-q8_0.gguf](https://huggingface.co/cstr/multilingual-e5-large-GGUF/resolve/main/multilingual-e5-large-q8_0.gguf) | Q8_0 | 574 MB |
| [multilingual-e5-large.gguf](https://huggingface.co/cstr/multilingual-e5-large-GGUF/resolve/main/multilingual-e5-large.gguf) | F32 | 2141 MB |


## Quick Start

```bash
# Download
huggingface-cli download cstr/multilingual-e5-large-GGUF multilingual-e5-large-q4_k.gguf --local-dir .

# Run with CrispEmbed
./crispembed -m multilingual-e5-large-q4_k.gguf "Hello world"

# Or with auto-download
./crispembed -m multilingual-e5-large "Hello world"
```

## Model Details

| Property | Value |
|----------|-------|
| Architecture | XLM-R |
| Parameters | 560M |
| Embedding Dimension | 1024 |
| Layers | 24 |
| Pooling | mean |
| Tokenizer | SentencePiece |
| Base Model | [intfloat/multilingual-e5-large](https://huggingface.co/intfloat/multilingual-e5-large) |

## Verification

Verified bit-identical to HuggingFace sentence-transformers (cosine similarity >= 0.999 on test texts).

## Usage with CrispEmbed

CrispEmbed is a lightweight C/C++ text embedding inference engine using ggml.
No Python runtime, no ONNX. Supports BERT, XLM-R, Qwen3, and Gemma3 architectures.

```bash
# Build CrispEmbed
git clone https://github.com/CrispStrobe/CrispEmbed
cd CrispEmbed
cmake -S . -B build && cmake --build build -j

# Encode
./build/crispembed -m multilingual-e5-large-q4_k.gguf "query text"

# Server mode
./build/crispembed-server -m multilingual-e5-large-q4_k.gguf --port 8080
curl -X POST http://localhost:8080/v1/embeddings \
    -d '{"input": ["Hello world"], "model": "multilingual-e5-large"}'
```

## Credits

- Original model: [intfloat/multilingual-e5-large](https://huggingface.co/intfloat/multilingual-e5-large)
- Inference engine: [CrispEmbed](https://github.com/CrispStrobe/CrispEmbed) (ggml-based)
- Conversion: `convert-bert-embed-to-gguf.py`