Instructions to use lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-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 lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-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 lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_M
# Run inference directly in the terminal:
llama cli -hf lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_M
# Run inference directly in the terminal:
llama cli -hf lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_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 lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_M

Use Docker

docker model run hf.co/lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_M

Llama new
LM Studio
Jan

vLLM

How to use lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-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": "lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_M

Ollama
How to use lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF with Ollama:
```
ollama run hf.co/lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_M
```
Unsloth Desktop

How to use lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_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": "lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF with Docker Model Runner:
```
docker model run hf.co/lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_M
```

Lemonade

How to use lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_M

Run and chat with the model

lemonade run user.Clover-35B-A3B-Preview-GSQ-RCO-GGUF-IQ2_M

List all available models

lemonade list

Hermes Agent

How to use lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-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 lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_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 lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_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 "lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF:IQ2_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"

Clover-35B-A3B-Preview — GSQ-RCO-style GGUF

Per-tensor mixed-precision GGUF builds of Diffuspace/Clover-35B-A3B-Preview, made for the Clover-capable llama.cpp branch.

What this model is

Clover is a text-only, 35B-parameter mixture-of-experts model with about 3B active parameters per token, derived from Qwen3.6-35B-A3B. It has 40 layers, 256 routed experts per layer, top-8 routing, and a 131,072-token context. Upstream reference: Clover model card. The model and its Qwen base are Apache-2.0.

Important architecture limitation

These files preserve Clover's causal/autoregressive path only. The converter filters out the bidirectional GDN tensors used by the upstream block-diffusion denoiser, so these GGUFs do not support block-diffusion generation. Use them for ordinary left-to-right causal decoding with a Clover-aware runtime; do not treat them as a complete export of the original diffusion model.

Vision and speculative heads

Clover is text-only; the source checkpoint has no vision tower or image projector. The source safetensors index also contains no standalone MTP/draft head tensors. Upstream draft-self is self-speculative decoding, not a separate MTP model; it relies on the block-diffusion path and is not available in this AR-only GGUF export.

Files

File	Size	Target BPW
`quant/Clover-35B-A3B-Preview-GSQ-RCO-IQ2_XXS.gguf`	9.43 GiB	2.30
`quant/Clover-35B-A3B-Preview-GSQ-RCO-IQ2_XS.gguf`	10.09 GiB	2.50
`quant/Clover-35B-A3B-Preview-GSQ-RCO-IQ2_S.gguf`	10.91 GiB	2.70
`quant/Clover-35B-A3B-Preview-GSQ-RCO-IQ2_M.gguf`	11.82 GiB	2.90
`quant/Clover-35B-A3B-Preview-GSQ-RCO-IQ3_XXS.gguf`	12.12 GiB	3.00
`quant/Clover-35B-A3B-Preview-GSQ-RCO-IQ3_XS.gguf`	13.12 GiB	3.25
`quant/Clover-35B-A3B-Preview-GSQ-RCO-IQ3_S.gguf`	14.13 GiB	3.50
`quant/Clover-35B-A3B-Preview-GSQ-RCO-IQ3_M.gguf`	15.14 GiB	3.75

Source and provenance

Diffuspace/Clover-35B-A3B-Preview BF16 checkpoint (revision 4b0402b65e523d0183a14c431955438714cb16ff)
   → BF16 GGUF using Clover converter revision ed26f2f7621306a8c393dd5193c18b3c3cea7940
   → local per-tensor GSQ-RCO-style allocation + llama.cpp importance-matrix quantization

Calibration source: bundled/provided text corpus quantization_calibration.txt.

“GSQ-RCO-style” describes a format-compatible local allocation workflow, not a literal reproduction of the upstream GSQ/RCO algorithms. Tensor rules were assigned by this workspace's bundled allocator and applied with llama.cpp's importance-matrix quantizer. Target BPW values are budget inputs; final file sizes and achieved BPW can vary by architecture. Conversion and quantization were CPU-only.

Usage

Use a llama.cpp build with Clover support (for example, the converter/runtime branch at revision ed26f2f7621306a8c393dd5193c18b3c3cea7940). These artifacts are for causal decoding only; bidirectional block denoising is not present.

quant/Clover-35B-A3B-Preview-GSQ-RCO-IQ3_S.gguf

Downloads last month: 982

GGUF

Model size

35B params

Architecture

clover_moe

Hardware compatibility

2-bit

3-bit

Model tree for lavendercomputing/Clover-35B-A3B-Preview-GSQ-RCO-GGUF

Base model

Qwen/Qwen3.6-35B-A3B

Finetuned

Diffuspace/Clover-35B-A3B-Preview

Quantized

(21)

this model