Instructions to use cstr/jina-v5-small-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use cstr/jina-v5-small-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="cstr/jina-v5-small-GGUF",
	filename="jina-v5-small-classification.gguf",
)

output = llm(
	"Once upon a time,",
	max_tokens=512,
	echo=True
)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use cstr/jina-v5-small-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf cstr/jina-v5-small-GGUF:Q8_0
# Run inference directly in the terminal:
llama-cli -hf cstr/jina-v5-small-GGUF:Q8_0

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf cstr/jina-v5-small-GGUF:Q8_0
# Run inference directly in the terminal:
llama-cli -hf cstr/jina-v5-small-GGUF:Q8_0

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf cstr/jina-v5-small-GGUF:Q8_0
# Run inference directly in the terminal:
./llama-cli -hf cstr/jina-v5-small-GGUF:Q8_0

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf cstr/jina-v5-small-GGUF:Q8_0
# Run inference directly in the terminal:
./build/bin/llama-cli -hf cstr/jina-v5-small-GGUF:Q8_0

Use Docker

docker model run hf.co/cstr/jina-v5-small-GGUF:Q8_0

LM Studio
Jan
Ollama
How to use cstr/jina-v5-small-GGUF with Ollama:
```
ollama run hf.co/cstr/jina-v5-small-GGUF:Q8_0
```

Unsloth Studio

How to use cstr/jina-v5-small-GGUF with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for cstr/jina-v5-small-GGUF to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for cstr/jina-v5-small-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for cstr/jina-v5-small-GGUF to start chatting

Atomic Chat new
Docker Model Runner
How to use cstr/jina-v5-small-GGUF with Docker Model Runner:
```
docker model run hf.co/cstr/jina-v5-small-GGUF:Q8_0
```

Lemonade

How to use cstr/jina-v5-small-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull cstr/jina-v5-small-GGUF:Q8_0

Run and chat with the model

lemonade run user.jina-v5-small-GGUF-Q8_0

List all available models

lemonade list

cstr commited on May 17

Commit

7354204

verified ·

1 Parent(s): c6d7fb5

Update README: v2 Ollama format, retrieval LoRA merged

Browse files

Files changed (1) hide show

README.md +27 -57

README.md CHANGED Viewed

@@ -1,70 +1,40 @@
 ---
-license: cc-by-nc-4.0
-language: [multilingual]
-tags: [embeddings, gguf, ggml, text-embeddings, qwen3, crispembed, ollama]
 pipeline_tag: feature-extraction
 base_model: jinaai/jina-embeddings-v5-text-small
 ---
-# jina-v5-small GGUF
-GGUF format of [jinaai/jina-embeddings-v5-text-small](https://huggingface.co/jinaai/jina-embeddings-v5-text-small) for use with [CrispEmbed](https://github.com/CrispStrobe/CrispEmbed) and [Ollama](https://ollama.com).
-## Files
-| File | Quantization | Size |
-|------|-------------|------|
-| [jina-v5-small-q4_k.gguf](https://huggingface.co/cstr/jina-v5-small-GGUF/resolve/main/jina-v5-small-q4_k.gguf) | Q4_K | 0 MB |
-| [jina-v5-small-q8_0.gguf](https://huggingface.co/cstr/jina-v5-small-GGUF/resolve/main/jina-v5-small-q8_0.gguf) | Q8_0 | 0 MB |
-| [jina-v5-small.gguf](https://huggingface.co/cstr/jina-v5-small-GGUF/resolve/main/jina-v5-small.gguf) | F32 | 0 MB |
-**Recommended:** Q8_0 for quality (cos vs HF: L2=1.0), Q4_K for size (L2=1.0).
-## Quick Start
-### CrispEmbed
-```bash
-./crispembed -m jina-v5-small "Hello world"
-./crispembed-server -m jina-v5-small --port 8080
-```
-### Ollama (with [CrispStrobe fork](https://github.com/CrispStrobe/ollama/tree/feat/xlmr-embedding))
-```bash
-echo "FROM jina-v5-small-q8_0.gguf" > Modelfile
-ollama create jina-v5-small -f Modelfile
-curl http://localhost:11434/api/embed -d '{"model":"jina-v5-small","input":["Hello world"]}'
-```
-### Python (CrispEmbed)
-```python
-from crispembed import CrispEmbed
-model = CrispEmbed("jina-v5-small-q8_0.gguf")
-vectors = model.encode(["Hello world", "Goodbye world"])
-```
-## Model Details
-| Property | Value |
-|----------|-------|
-| Architecture | Qwen3 |
-| Parameters | 600M |
-| Embedding Dimension | 1024 |
-| Layers | 28 |
-| Pooling | last-token |
-| Tokenizer | BPE |
-| Language | multilingual |
-| Q8_0 vs HuggingFace | L2=1.0 |
-| Q4_K vs HuggingFace | L2=1.0 |
-## Server API
-CrispEmbed server supports four API dialects:
-- `POST /embed` -- native
-- `POST /v1/embeddings` -- OpenAI-compatible
-- `POST /api/embed` -- Ollama-compatible
-- `POST /api/embeddings` -- Ollama legacy
-## Credits
-- Original model: [jinaai/jina-embeddings-v5-text-small](https://huggingface.co/jinaai/jina-embeddings-v5-text-small)
-- Inference: [CrispEmbed](https://github.com/CrispStrobe/CrispEmbed) (MIT, ggml-based)

 ---
+license: apache-2.0
+language:
+- multilingual
+tags:
+- embeddings
+- gguf
+- text-embeddings
+- jina
+- crispembed
 pipeline_tag: feature-extraction
 base_model: jinaai/jina-embeddings-v5-text-small
 ---
+# jina-embeddings-v5-text-small GGUF
+GGUF format of [jinaai/jina-embeddings-v5-text-small](https://huggingface.co/jinaai/jina-embeddings-v5-text-small) for use with [CrispEmbed](https://github.com/CrispStrobe/CrispEmbed) and Ollama-compatible runtimes.
+**Note:** These GGUFs have the `retrieval` LoRA adapter merged into the base weights. The original model supports 4 task-specific adapters (retrieval, text-matching, clustering, classification); this GGUF uses the retrieval adapter which is the most common use case.
+## Files
+| File | Quantization | Size | Parity (cos vs HF) |
+|------|-------------|------|---------------------|
+| jina-v5-small.gguf | F32 | ~2.3 GB | 0.9999 |
+| jina-v5-small-q8_0.gguf | Q8_0 | ~631 MB | 0.9995 |
+| jina-v5-small-q5_k.gguf | Q5_K | ~489 MB | 0.9926 |
+| jina-v5-small-q4_k.gguf | Q4_K | ~419 MB | 0.9725 |
+## Architecture
+- **Base:** Qwen3-style transformer (28 layers, 1024 dims)
+- **Embedding dimension:** 1024
+- **Pooling:** Last-token + L2 normalize
+- **Context length:** 8,192 tokens
+- **License:** Apache 2.0
+## Notes
+Ollama-compatible format (`qwen3.*` namespace). Bidirectional attention (`is_bidirectional=1`).