Instructions to use HelloSun/qwen3-0.6b-ssd-streaming with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use HelloSun/qwen3-0.6b-ssd-streaming 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 HelloSun/qwen3-0.6b-ssd-streaming:Q8_0
# Run inference directly in the terminal:
llama cli -hf HelloSun/qwen3-0.6b-ssd-streaming:Q8_0

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf HelloSun/qwen3-0.6b-ssd-streaming:Q8_0
# Run inference directly in the terminal:
llama cli -hf HelloSun/qwen3-0.6b-ssd-streaming: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 HelloSun/qwen3-0.6b-ssd-streaming:Q8_0
# Run inference directly in the terminal:
./llama-cli -hf HelloSun/qwen3-0.6b-ssd-streaming: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 HelloSun/qwen3-0.6b-ssd-streaming:Q8_0
# Run inference directly in the terminal:
./build/bin/llama-cli -hf HelloSun/qwen3-0.6b-ssd-streaming:Q8_0

Use Docker

docker model run hf.co/HelloSun/qwen3-0.6b-ssd-streaming:Q8_0

LM Studio
Jan

vLLM

How to use HelloSun/qwen3-0.6b-ssd-streaming with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "HelloSun/qwen3-0.6b-ssd-streaming"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "HelloSun/qwen3-0.6b-ssd-streaming",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/HelloSun/qwen3-0.6b-ssd-streaming:Q8_0

Ollama
How to use HelloSun/qwen3-0.6b-ssd-streaming with Ollama:
```
ollama run hf.co/HelloSun/qwen3-0.6b-ssd-streaming:Q8_0
```
Unsloth Desktop

How to use HelloSun/qwen3-0.6b-ssd-streaming with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf HelloSun/qwen3-0.6b-ssd-streaming:Q8_0

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": "HelloSun/qwen3-0.6b-ssd-streaming:Q8_0"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use HelloSun/qwen3-0.6b-ssd-streaming with Docker Model Runner:
```
docker model run hf.co/HelloSun/qwen3-0.6b-ssd-streaming:Q8_0
```

Lemonade

How to use HelloSun/qwen3-0.6b-ssd-streaming with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull HelloSun/qwen3-0.6b-ssd-streaming:Q8_0

Run and chat with the model

lemonade run user.qwen3-0.6b-ssd-streaming-Q8_0

List all available models

lemonade list

Hermes Agent

How to use HelloSun/qwen3-0.6b-ssd-streaming with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf HelloSun/qwen3-0.6b-ssd-streaming:Q8_0

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 HelloSun/qwen3-0.6b-ssd-streaming:Q8_0

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use HelloSun/qwen3-0.6b-ssd-streaming with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf HelloSun/qwen3-0.6b-ssd-streaming:Q8_0

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 "HelloSun/qwen3-0.6b-ssd-streaming:Q8_0" \
  --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"

Qwen3-0.6B：SSD Streaming CPU 推理

這是使用 llama.cpp 純 CPU 執行 Qwen3-0.6B-Q8_0.gguf 的低記憶體推論專案。模型權重約 639 MB，KV cache 使用 SSD 檔案 mmap；所有程式、建置資料與驗證結果都留在本 repo 目錄中。

本專案是 SSD Streaming + KV SSD Paging，不是 MoE expert paging，也不是 EPQ。它適合用來驗證「小模型在極低 RAM 下，以 SSD 映射換取可執行性」的路徑。

使用方法

git clone https://huggingface.co/HelloSun/qwen3-0.6b-ssd-streaming
cd qwen3-0.6b-ssd-streaming

# 編譯、下載模型、驗證並進入 chat
python3 demo.py

# 只做一次推論驗證
python3 demo.py --validate-only

# 多 prompt benchmark
python3 benchmark.py

BENCH_REPEATS 與 BENCH_MAX_TOKENS 可調整 benchmark 次數與生成長度。

實測條件與結果

目前驗證主機使用 8 threads、1024 context、權重與 K/V 都以檔案映射位於 NVMe。 8 次請求平均生成速度為 1.58 tokens/s，實測範圍 1.55–1.61 tokens/s；峰值 RSS 約 125.82 MiB，server process VmSwap=0。

但該主機原本啟用了約 2 GiB system swap，因此這些數字是 SSD offload 效能記錄， 不是整機 swap=0 的通過證據。請以 validate/validation.json 中的主機狀態與驗證結果為準。

記憶體與 SSD 設定

權重：mmap，搭配 LLAMA_MMAP_RELEASE=1 背景釋放頁面。
KV：K/V 使用 q8_0，LLAMA_KV_SSD=1，放在 models/kv-ssd.cache。
page cache：定期對 GGUF 與 KV 檔呼叫 POSIX_FADV_DONTNEED。
RAM：activations、compute buffer、OS 與 llama.cpp 執行狀態仍需 RAM。
設定：LLAMA_CONTEXT、LLAMA_THREADS、MEMORY_LIMIT_GB 可覆寫預設值。

swap=0 必須在實際推論主機上成立；程式會分開記錄 process VmSwap 與 system swap，不會假裝替使用者關閉 swap。

證據

validate/validation.json：smoke test 與驗證 verdict
validate/performance.json：多 prompt 效能
validate/memory.json：RSS / VmSwap 取樣
benchmark.py：可重現 benchmark

參考

Downloads last month: 41

GGUF

Model size

0.6B params

Architecture

qwen3

Hardware compatibility

8-bit