CoolFace
Modelpublic

cstr/granite-embedding-311m-multilingual-r2-GGUF

sourceHugging Faceapache-2.0updated 2mo agoView on Hugging Face
0likes200downloads
Model Card

granite-embedding-311m-multilingual-r2 GGUF

GGUF format of ibm-granite/granite-embedding-311m-multilingual-r2 for use with CrispEmbed.

IBM Granite Embedding 311M R2. Multilingual (50+ languages), 768-dimensional CLS-pooled, 8192-token context. No query or document prefix.

Files

FileQuantizationSize
granite-embedding-311m-multilingual-r2-f16.ggufF161142 MB
granite-embedding-311m-multilingual-r2-q8_0.ggufQ8_0331 MB

Parity vs HuggingFace reference

Cosine similarity vs the upstream sentence-transformers reference on a fixed test set (text):

QuantText
f161.0000
q8_00.9998

Quick Start

bash
# Download
huggingface-cli download cstr/granite-embedding-311m-multilingual-r2-GGUF granite-embedding-311m-multilingual-r2-f16.gguf --local-dir .

# Run with CrispEmbed
./crispembed -m granite-embedding-311m-multilingual-r2-f16.gguf "Hello world"

# Or with auto-download
./crispembed -m granite-embedding-311m-multilingual-r2 "Hello world"

Model Details

PropertyValue
ArchitectureModernBERT (RoPE, local/global attention, GeGLU, pre-LN)
Parameters311M
Embedding Dimension768
Layers22
PoolingCLS
TokenizerSentencePiece BPE (262k)
Base Modelibm-granite/granite-embedding-311m-multilingual-r2

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-311m-multilingual-r2-f16.gguf "query text"

# Server mode
./build/crispembed-server -m granite-embedding-311m-multilingual-r2-f16.gguf --port 8080
curl -X POST http://localhost:8080/v1/embeddings \
    -d '{"input": ["Hello world"], "model": "granite-embedding-311m-multilingual-r2"}'

Credits