File	Quantization	Size
granite-embedding-107m-multilingual-q4_k.gguf	Q4_K	109 MB
granite-embedding-107m-multilingual-q8_0.gguf	Q8_0	114 MB
granite-embedding-107m-multilingual.gguf	F32	413 MB

Quick Start

bash

# Download
huggingface-cli download cstr/granite-embedding-107m-multilingual-GGUF granite-embedding-107m-multilingual-q4_k.gguf --local-dir .

# Run with CrispEmbed
./crispembed -m granite-embedding-107m-multilingual-q4_k.gguf "Hello world"

# Or with auto-download
./crispembed -m granite-embedding-107m-multilingual "Hello world"

Model Details

Property	Value
Architecture	XLM-R
Parameters	107M
Embedding Dimension	384
Layers	6
Pooling	CLS
Tokenizer	SentencePiece
Base Model	ibm-granite/granite-embedding-107m-multilingual

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 granite-embedding-107m-multilingual-q4_k.gguf "query text"

# Server mode
./build/crispembed-server -m granite-embedding-107m-multilingual-q4_k.gguf --port 8080
curl -X POST http://localhost:8080/v1/embeddings \
    -d '{"input": ["Hello world"], "model": "granite-embedding-107m-multilingual"}'

Credits

Original model: ibm-granite/granite-embedding-107m-multilingual
Inference engine: CrispEmbed (ggml-based)
Conversion: convert-bert-embed-to-gguf.py

Provenance and EU AI Act Art. 53 note

Upstream model: ibm-granite/granite-embedding-107m-multilingual — published by ibm-granite.
Upstream licence: apache-2.0. This repository redistributes under the same terms; it grants no rights the upstream licence does not.
What was done here: format conversion and/or quantisation only (GGUF/GGML). No training, no fine-tuning, no merging, no distillation, no change to architecture, vocabulary or capability. Only the numeric representation of the upstream weights differs.
Training data: documented — where it is documented at all — by the upstream provider; see the upstream model card. No training data was used, added or selected by this repository.
Provider status: under Regulation (EU) 2024/1689 the upstream authors remain the provider of this model. Converting the serialisation format does not make this repository the provider of a new general-purpose AI model, and no such claim is made. Questions about training content, copyright policy or model capability belong upstream.