Instructions to use AtomicChat/SenseNova-Loop-9B-Preview-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 AtomicChat/SenseNova-Loop-9B-Preview-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 AtomicChat/SenseNova-Loop-9B-Preview-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf AtomicChat/SenseNova-Loop-9B-Preview-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 AtomicChat/SenseNova-Loop-9B-Preview-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf AtomicChat/SenseNova-Loop-9B-Preview-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf AtomicChat/SenseNova-Loop-9B-Preview-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf AtomicChat/SenseNova-Loop-9B-Preview-GGUF:Q4_K_M

Use Docker

docker model run hf.co/AtomicChat/SenseNova-Loop-9B-Preview-GGUF:Q4_K_M

Llama new
LM Studio
Jan

vLLM

How to use AtomicChat/SenseNova-Loop-9B-Preview-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "AtomicChat/SenseNova-Loop-9B-Preview-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AtomicChat/SenseNova-Loop-9B-Preview-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/AtomicChat/SenseNova-Loop-9B-Preview-GGUF:Q4_K_M

Ollama
How to use AtomicChat/SenseNova-Loop-9B-Preview-GGUF with Ollama:
```
ollama run hf.co/AtomicChat/SenseNova-Loop-9B-Preview-GGUF:Q4_K_M
```
Unsloth Desktop

How to use AtomicChat/SenseNova-Loop-9B-Preview-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf AtomicChat/SenseNova-Loop-9B-Preview-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": "AtomicChat/SenseNova-Loop-9B-Preview-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use AtomicChat/SenseNova-Loop-9B-Preview-GGUF with Docker Model Runner:
```
docker model run hf.co/AtomicChat/SenseNova-Loop-9B-Preview-GGUF:Q4_K_M
```

Lemonade

How to use AtomicChat/SenseNova-Loop-9B-Preview-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull AtomicChat/SenseNova-Loop-9B-Preview-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.SenseNova-Loop-9B-Preview-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use AtomicChat/SenseNova-Loop-9B-Preview-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 AtomicChat/SenseNova-Loop-9B-Preview-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 AtomicChat/SenseNova-Loop-9B-Preview-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use AtomicChat/SenseNova-Loop-9B-Preview-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf AtomicChat/SenseNova-Loop-9B-Preview-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 "AtomicChat/SenseNova-Loop-9B-Preview-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"

Memory needed per quant (file sizes + looped KV cache estimate)

by localllmrig - opened 8 days ago

Discussion

localllmrig

8 days ago

Quick fit table from the published file sizes, in case it helps people pick a quant once the patched llama.cpp lands.

Need ≈ file + KV cache at 32K context + ~0.8 GB overhead (estimate). One thing the loop changes: from config.json, 8 of the 32 stored layers are full attention, but 6 of those (layers 7, 11, 15, 19, 23, 27) sit inside the looped region 4–27 and each visit keeps its own KV, so there are 14 KV layer-visits rather than 8. At 14 × 4 KV heads × 256 head_dim × 2 (K+V) × fp16, that's 57,344 bytes/token: ~1.9 GB at 32K, ~15 GB at the full 262K. The Gated-DeltaNet recurrent state is ~0.1 GB (fp32), so it barely matters.

Quant	File	Need @32K (est.)	Fits (est.)
Q4_K_M	5.6 GB	~9 GB	12 GB GPU / 16 GB Mac
Q5_K_M	6.5 GB	~10 GB	12 GB GPU / 16 GB Mac
Q6_K	7.4 GB	~11 GB	12 GB GPU (tight)
Q8_0	9.5 GB	~13 GB	16 GB GPU / 32 GB Mac
BF16	17.9 GB	~21 GB	24 GB GPU / 32 GB Mac (tight)

(Sizes here are GB = 10^9 bytes from the file list; the card's table looks like GiB, e.g. 5.2 GiB = 5.6 GB.)

So the "~~8 GB-class" note for Q4_K_M holds at short context (~~7 GB at 8K), but at 32K a 12 GB card is the comfortable floor, because the looped KV costs 1.75× what the layer count suggests. Corrections welcome, especially if the patched build shares KV across loop visits.

More per-VRAM numbers: https://localllmrig.com/vram/12gb?utm_source=huggingface&utm_medium=community&utm_campaign=sensenova-loop-9b

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment