Instructions to use roller777/qwen3vl-4b-text-encoder-gguf with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use roller777/qwen3vl-4b-text-encoder-gguf with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M

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 roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M

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 roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M

Use Docker

docker model run hf.co/roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M

LM Studio
Jan
Ollama
How to use roller777/qwen3vl-4b-text-encoder-gguf with Ollama:
```
ollama run hf.co/roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M
```
Unsloth Desktop

How to use roller777/qwen3vl-4b-text-encoder-gguf with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use roller777/qwen3vl-4b-text-encoder-gguf with Docker Model Runner:
```
docker model run hf.co/roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M
```

Lemonade

How to use roller777/qwen3vl-4b-text-encoder-gguf with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M

Run and chat with the model

lemonade run user.qwen3vl-4b-text-encoder-gguf-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use roller777/qwen3vl-4b-text-encoder-gguf with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use roller777/qwen3vl-4b-text-encoder-gguf with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "roller777/qwen3vl-4b-text-encoder-gguf:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Qwen3-VL-4B text encoder (GGUF)

Krea 2 / Mage-Flow が条件付けに使う Qwen3-VL-4B のテキスト塔だけを GGUF 化したものです。 VRAM にテキストエンコーダが乗り切らない環境向け(bf16 8.9GB → q4 で 2.4GB)。

ファイル	サイズ	HF bf16 との隠れ状態誤差
`qwen3vl-4b-q8_0.gguf`	4.3GB	約3%
`qwen3vl-4b-q4_k_m.gguf`	2.4GB	約17%

数値誤差はありますが、Mage-Flow-Turbo での同一シード生成画像は bf16 条件付けと見分けがつきませんでした(q8/q4 とも実測)。この系統の拡散モデルの条件付けは量子化にかなり頑健です。

使い方(llama.cpp — 動作確認済み)

llama-server -m qwen3vl-4b-q4_k_m.gguf --embeddings --pooling none -c 1024

/embedding に トークンID配列で POST({"content": [ids], "embd_normalize": -1})し、先頭34トークン(システム/ユーザーテンプレート部)を剥がすと Krea/Mage の last_hidden_state 相当(トークン毎 2560 次元)が得られます。文字列で渡すとトークン化が HF と数トークンずれるので ID 渡し推奨。テンプレートは Krea/Mage 共通:

<|im_start|>system
Describe the image by detailing the color, shape, size, texture, quantity, text, spatial relationships of the objects and background:<|im_end|>
<|im_start|>user
{プロンプト}<|im_end|>
<|im_start|>assistant

翻訳などの通常のチャット生成(/v1/chat/completions)も同じサーバでそのまま動きます。

ComfyUI について(未検証)

ComfyUI-GGUF の CLIPLoader (gguf) で読める可能性がありますが未検証です。まずは公式の qwen3vl_4b_fp8_scaled.safetensors(5.24GB、Comfy-Org/Krea-2)を試し、それでも足りない場合の選択肢としてどうぞ。

変換メモ

変換元: Mage-Flow 配布の qwen3vl_4b_bf16.safetensors(Krea 2 と同一重み)
llama.cpp b10194 convert_hf_to_gguf.py。config.json の architectures を Qwen3VLForConditionalGeneration に書き換えると視覚塔を自動スキップしてテキスト塔だけが変換されます
q4_k_m は q8_0 からの再量子化(llama-quantize --allow-requantize)

Downloads last month: 555

GGUF

Model size

4B params

Architecture

qwen3vl

Hardware compatibility

4-bit

8-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for roller777/qwen3vl-4b-text-encoder-gguf

Base model

Qwen/Qwen3-VL-4B-Instruct

Quantized

(113)

this model