Instructions to use majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M",
	filename="Nemotron-3-Nano-30B-A3B-TurboQuant-Q4_K_M.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M 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 majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M: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 majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M: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 majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M:Q4_K_M

Use Docker

docker model run hf.co/majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M:Q4_K_M

LM Studio
Jan

vLLM

How to use majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M:Q4_K_M

Ollama
How to use majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M with Ollama:
```
ollama run hf.co/majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M:Q4_K_M
```

Unsloth Studio

How to use majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M to start chatting

How to use majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M: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 majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M:Q4_K_M

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M with Docker Model Runner:
```
docker model run hf.co/majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M:Q4_K_M
```

Lemonade

How to use majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M:Q4_K_M

Run and chat with the model

lemonade run user.Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M-Q4_K_M

List all available models

lemonade list

majentik commited on Apr 15

Commit

32b6d20

verified ·

1 Parent(s): a5c7bdc

Update model card with accurate fork requirements and ecosystem status

Browse files

Files changed (1) hide show

README.md +107 -42

README.md CHANGED Viewed

@@ -1,5 +1,7 @@
 ---
-library_name: gguf
 base_model: nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16
 tags:
   - gguf
@@ -12,75 +14,138 @@ tags:
   - moe
   - llama-cpp
   - quantized
-license: other
-license_name: nvidia-open-model-license
-license_link: https://developer.download.nvidia.com/licenses/nvidia-open-model-license-agreement-june-2024.pdf
 ---
 # Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M
-GGUF Q4_K_M weight-quantized variant of [nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16](https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16) with **TurboQuant** KV cache compression for efficient inference with llama.cpp, Ollama, and LM Studio. Features a hybrid Mamba-2 + Transformer MoE architecture with 30.7B total parameters (3.2B active per token) and up to 1M context length.
 ## Overview
-This model combines two compression techniques:
-- **GGUF Q4_K_M weight quantization** -- reduces model size from ~60 GB to ~14 GB
-- **TurboQuant KV cache compression** -- block-diagonal rotations (Clifford algebra) for 3-bit KV cache, 5.3x faster prefill
 ## Quickstart
-### llama.cpp
 ```bash
-llama-cli -m Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M.gguf \
-  --cache-type-k q4_0 --cache-type-v q4_0 \
   -p "Explain quantum computing"
 ```
-### Ollama
 ```bash
-ollama run majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M
 ```
-### LM Studio
-Download the GGUF file and load in LM Studio. Enable TurboQuant KV cache in advanced settings.
 ## Specifications
 | Property | Value |
 |----------|-------|
-| Base Model | nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 |
-| Parameters | 30.7B (3.2B active, Mamba-2 + Transformer MoE) |
-| Context Length | 1,048,576 tokens (1M) |
-| Weight Quantization | GGUF Q4_K_M |
-| KV Cache | TurboQuant 3-bit (planar/iso) |
-| File Size | ~14 GB |
-| License | NVIDIA Open Model License (commercial use OK) |
-| Compatible | llama.cpp, Ollama, LM Studio, koboldcpp |
 ## What is TurboQuant?
-TurboQuant applies block-diagonal rotations (Clifford algebra) for KV cache compression. When used with llama.cpp's `--cache-type-k q4_0 --cache-type-v q4_0` flags:
-| Metric | TurboQuant | TurboQuant |
-|--------|-----------|-----------|
-| Prefill Speed | 3,822 tok/s | 722 tok/s |
-| Decode Speed | 119 tok/s | 93 tok/s |
-| Perplexity | 6.91 | 7.07 |
-## GGUF Quant Variants
-| Quant | File Size | Quality | Variant |
-|-------|-----------|---------|---------|
-| Q2_K | ~9 GB | Lowest | [Q2_K](https://huggingface.co/majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q2_K) |
-| Q3_K_M | ~11 GB | Low | [Q3_K_M](https://huggingface.co/majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q3_K_M) |
-| IQ4_XS | ~13 GB | Medium-Low | [IQ4_XS](https://huggingface.co/majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-IQ4_XS) |
-| **Q4_K_M** | **~14 GB** | **Medium (recommended)** | **This model** |
-| Q5_K_M | ~17 GB | Medium-High | [Q5_K_M](https://huggingface.co/majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q5_K_M) |
-| Q8_0 | ~27 GB | High | [Q8_0](https://huggingface.co/majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q8_0) |
 ## See Also
-- [nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16](https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16) -- Base model
-- [majentik/Nemotron-3-Nano-30B-A3B-TurboQuant](https://huggingface.co/majentik/Nemotron-3-Nano-30B-A3B-TurboQuant) -- TurboQuant KV-cache (transformers)
-- [majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-MLX-4bit](https://huggingface.co/majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-MLX-4bit) -- MLX 4-bit variant
-- [TurboQuant GitHub](https://github.com/scrya-com/turboquant)

 ---
+license: other
+license_name: nvidia-open-model-license
+license_link: https://developer.download.nvidia.com/licenses/nvidia-open-model-license-agreement-june-2024.pdf
 base_model: nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16
 tags:
   - gguf
   - moe
   - llama-cpp
   - quantized
+library_name: gguf
+pipeline_tag: text-generation
 ---
 # Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M
+GGUF Q4_K_M weight-quantized variant of [nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16](https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16) optimised for use with **TurboQuant** KV cache compression via a dedicated llama.cpp fork.
+> **Important:** TurboQuant KV cache types (`planar3`, `iso3`) are **not** available in upstream llama.cpp, standard Ollama, or LM Studio.
+> They require a [specific llama.cpp fork](https://github.com/johndpope/llama-cpp-turboquant/tree/feature/planarquant-kv-cache).
+> The GGUF file itself is a standard GGUF and works with any llama.cpp-compatible runtime using normal KV cache types (f16, q8_0, q4_0, etc.).
 ## Overview
+This model combines two independent compression techniques:
+| Technique | What it does | Requirement |
+|-----------|-------------|-------------|
+| **GGUF Q4_K_M weight quantization** | Reduces model size from ~60 GB (BF16) to ~16.2 GB | Any llama.cpp-compatible runtime |
+| **TurboQuant KV cache compression** — random rotation + Lloyd-Max scalar quantization (`--cache-type-k planar3 --cache-type-v planar3`) | Block-diagonal rotations / random rotation for compressed KV cache | [llama-cpp-turboquant fork](https://github.com/johndpope/llama-cpp-turboquant/tree/feature/planarquant-kv-cache) only |
 ## Quickstart
+### Option A — With TurboQuant KV cache (fork required)
+You must build from the TurboQuant-enabled llama.cpp fork:
 ```bash
+# Clone and build the fork
+git clone https://github.com/johndpope/llama-cpp-turboquant.git
+cd llama-cpp-turboquant && git checkout feature/planarquant-kv-cache
+# CUDA (Windows/Linux)
+cmake -B build -DGGML_CUDA=ON -DCMAKE_BUILD_TYPE=Release && cmake --build build -j
+# Metal (Apple Silicon)
+cmake -B build -DGGML_METAL=ON -DGGML_METAL_EMBED_LIBRARY=ON -DCMAKE_BUILD_TYPE=Release && cmake --build build -j
+# Run with TurboQuant KV cache
+./build/bin/llama-cli -m Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M.gguf \
+  --cache-type-k planar3 --cache-type-v planar3 \
+  -ngl 99 -fa \
   -p "Explain quantum computing"
+# Or run as a server
+./build/bin/llama-server -m Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M.gguf \
+  --cache-type-k planar3 --cache-type-v planar3 \
+  -ngl 99 -fa --jinja
 ```
+### Option B — With standard llama.cpp / LM Studio / Ollama
+The GGUF works as a normal quantised model. You won't get TurboQuant-specific KV cache benefits, but standard KV cache quantization (q8_0, q4_0) still reduces VRAM significantly.
+**llama.cpp (upstream)**
 ```bash
+llama-cli -m Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M.gguf \
+  --cache-type-k q8_0 --cache-type-v q8_0 \
+  -ngl 99 -fa \
+  -p "Explain quantum computing"
 ```
+**LM Studio**
+1. Download the GGUF file and load in LM Studio.
+2. Enable **Developer Mode** (Settings → Developer).
+3. In the model loader's advanced settings, set **Flash Attention** to ON.
+4. Set **K Cache Quantization** and **V Cache Quantization** to `q8_0` (or `q4_0` for more aggressive VRAM savings).
+5. Note: LM Studio does not currently support TurboQuant's `planar3` cache types. Track [this feature request](https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1719) for updates.
+**Ollama**
+```bash
+# Standard Ollama does not support TurboQuant cache types.
+# Use with default or q8_0 KV cache via OLLAMA_KV_CACHE_TYPE=q8_0
+OLLAMA_KV_CACHE_TYPE=q8_0 OLLAMA_FLASH_ATTENTION=1 ollama run majentik/Nemotron-3-Nano-30B-A3B-TurboQuant-GGUF-Q4_K_M
+```
 ## Specifications
 | Property | Value |
 |----------|-------|
+| Base Model | [nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16](https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16) |
+| Architecture | Mamba-2 + Transformer hybrid Sparse MoE |
+| Parameters | 30.7B total, 3.2B active per token |
+| Context Length | 1M |
+| Weight Quantization | GGUF Q4_K_M (popular 4-bit, best quality/size tradeoff) |
+| Original Size (BF16) | ~60 GB |
+| Quantized File Size | ~16.2 GB |
+| KV Cache (TurboQuant) | 3-bit via `--cache-type-k planar3 --cache-type-v planar3` (fork only) |
+| KV Cache (standard) | q8_0, q4_0, f16, etc. (any llama.cpp runtime) |
+| License | other |
+| Modalities | Text only |
+| Compatible Runtimes | llama.cpp, LM Studio, Ollama, koboldcpp |
 ## What is TurboQuant?
+[TurboQuant](https://arxiv.org/abs/2504.19874) (ICLR 2026) is a KV cache compression method that applies a random orthogonal rotation followed by optimal scalar quantization. Bit-identical prefill logits at 4-bit on tested models, with up to 4-8× memory savings for long sequences.
+**Benchmarks from the TurboQuant repository** (Llama 3.1 8B, RTX 5090 — results will vary by model and hardware):
+| Metric | TurboQuant (4-bit) | Standard q4_0 |
+|--------|--------------------|---------------|
+| Quality | Bit-identical prefill | Lossy |
+| KV Compression | ~4× vs FP16 | ~4× vs FP16 |
+| Speedup (Apple Silicon) | 1.4–1.7× | — |
+> **Note:** These benchmarks are from the TurboQuant repository using Llama 3.1 8B on an RTX 5090. Performance on Nemotron-3-Nano-30B-A3B will differ. Independent benchmarks for this specific model are welcome — please open a discussion if you have results to share.
+## Current Status of TurboQuant in the Ecosystem
+| Runtime | TurboQuant Support | Standard KV Quant |
+|---------|---------------------|-------------------|
+| llama.cpp (upstream) | ❌ Not merged | ✅ q8_0, q4_0, q4_1, iq4_nl, q5_0, q5_1 |
+| llama-cpp-turboquant fork | ✅ planar3 | ✅ All standard types |
+| LM Studio | ❌ [Requested](https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1719) | ✅ Via advanced settings |
+| Ollama | ❌ Not supported | ✅ Via OLLAMA_KV_CACHE_TYPE |
+| koboldcpp | ❌ Not supported | ✅ Standard types |
+## Recommended Settings
+For VRAM-constrained setups, standard q8_0 KV cache quantization already halves KV cache memory with negligible quality impact. Flash Attention should always be enabled — it is required for V cache quantization and improves memory efficiency regardless.
+| VRAM | Suggested Configuration |
+|------|------------------------|
+| 24 GB (RTX 4090) | Q4_K_M + q8_0 KV cache + Flash Attention, 8K–16K context |
+| 16 GB | Q4_K_M + q4_0 KV cache + Flash Attention, 4K–8K context |
+| 48+ GB | Q4_K_M + f16 KV cache, full 32K+ context |
 ## See Also
+- [RotorQuant GitHub](https://github.com/scrya-com/rotorquant)
+- [llama-cpp-turboquant fork](https://github.com/johndpope/llama-cpp-turboquant/tree/feature/planarquant-kv-cache)
+- [TurboQuant llama.cpp discussion](https://github.com/ggml-org/llama.cpp/discussions/20969)
+- [TurboQuant paper (arXiv: 2504.19874)](https://arxiv.org/abs/2504.19874)
+- [Base model: nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16](https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16)
+- [Nemotron-3-Nano-30B-A3B announcement](https://huggingface.co/blog/nvidia/nemotron-3-nano-efficient-open-intelligent-models)