Instructions to use unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-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 unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_M
# Run inference directly in the terminal:
llama cli -hf unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_M
# Run inference directly in the terminal:
llama cli -hf unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-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 unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-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 unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_M

Use Docker

docker model run hf.co/unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_M

LM Studio
Jan

vLLM

How to use unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-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": "unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_M

SGLang

How to use unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF with Ollama:
```
ollama run hf.co/unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_M
```
Unsloth Desktop

How to use unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-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": "unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF with Docker Model Runner:
```
docker model run hf.co/unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_M
```

Lemonade

How to use unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_M

Run and chat with the model

lemonade run user.NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF-UD-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-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 unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-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 unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-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 "unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-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"

Could you upload imatrix_unsloth.gguf for Lightning? (+ two questions about the MTP layer)

by pirola - opened Aug 16

Discussion

pirola

Aug 16

•

edited Aug 16

Hi — thanks for getting Lightning quants out so fast.

The imatrix file seems to be missing from the repo. Your own quant metadata references it:

quantize.imatrix.file = NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF/imatrix_unsloth.gguf
quantize.imatrix.dataset = unsloth_calibration_NVIDIA-Nemotron-3.5-Lightning-30B-A3B.txt
quantize.imatrix.chunks_count = 80
quantize.imatrix.entries_count = 185

but there's no imatrix_unsloth.gguf in the file listing (you shipped one for Nemotron-3-Nano-30B-A3B). Was it just missed in the upload? Having it would let people reproduce or re-mix quants without recomputing calibration.

Does the imatrix cover blk.52 (the MTP/nextn layer)? With entries_count = 185 I suspect it doesn't. Using a third-party imatrix, llama-quantize bails with:

Missing importance matrix for tensor blk.52.ffn_down_exps.weight in a very low-bit quantization

I notice your files put blk.52.ffn_{up,down}_exps at Q5_0 — a block-32 type that needs no importance data — while the trunk experts get the imatrix-guided types. Was that a deliberate workaround for the same issue, or does your imatrix actually include blk.52?

Is there any way to get importance data for the MTP head? As far as I can tell llama-imatrix never exercises the nextn path (it's not part of the normal forward pass), so those tensors can't get statistics by construction. Is there a flag or procedure that works, or is high-precision-for-blk.52 simply the right answer?
Does MTP speculative decoding actually work for this model? i.e. --spec-type draft-mtp — have you measured a speedup, and does it need anything beyond keeping blk.52 in the file? You kept the whole MTP head at high precision, which suggests it's meant to be usable.

Context, in case it's useful: I'm quantizing this model with zero-padded expert tensors (1856→2048, 2688→2816) so block-256 quants actually apply — the Nemotron-3 family's expert dims aren't 256-divisible, which is why (for example) your UD-IQ3_XXS ends up with the 46 trunk expert tensors as IQ4_NL rather than IQ3_XXS, and lands at 19.8 GB. Padded, the same recipe gives genuine IQ3_XXS experts at ~13.7 GB. Happy to share details if that's interesting to you.

I did it successfully with nemotron 3 nano: pirola/Nemotron-3-Nano-30B-A3B-pirola-IQ2_XXS-XS-GGUF and pirola/Nemotron-3-Nano-30B-A3B-pirola-IQ3_XXS-GGUF
i did the pirola/Nemotron-3.5-Lightning-30B-A3B-pirola-IQ3_XXS-GGUF but would rather use your imatrix

pirola changed discussion title from imatrix please? to Could you upload imatrix_unsloth.gguf for Lightning? (+ two questions about the MTP layer) Aug 16

pirola

Aug 17

@danielhanchen could you please help me here?

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