Instructions to use unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF",
	filename="BF16/NVIDIA-Nemotron-3-Super-120B-A12B-BF16-00001-of-00005.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 unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-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-Super-120B-A12B-GGUF:UD-Q4_K_M
# Run inference directly in the terminal:
llama cli -hf unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-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-Super-120B-A12B-GGUF:UD-Q4_K_M
# Run inference directly in the terminal:
llama cli -hf unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-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-Super-120B-A12B-GGUF:UD-Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-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-Super-120B-A12B-GGUF:UD-Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF:UD-Q4_K_M

Use Docker

docker model run hf.co/unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF:UD-Q4_K_M

LM Studio
Jan

vLLM

How to use unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-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-Super-120B-A12B-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-Super-120B-A12B-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF:UD-Q4_K_M

Ollama
How to use unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF with Ollama:
```
ollama run hf.co/unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF:UD-Q4_K_M
```

Unsloth Studio

How to use unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF 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 unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF 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 unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF to start chatting

How to use unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-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-Super-120B-A12B-GGUF:UD-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": "unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF:UD-Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-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-Super-120B-A12B-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-Super-120B-A12B-GGUF:UD-Q4_K_M

Run Hermes

hermes

Atomic Chat new

OpenClaw new

How to use unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-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-Super-120B-A12B-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-Super-120B-A12B-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"

Docker Model Runner
How to use unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF with Docker Model Runner:
```
docker model run hf.co/unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF:UD-Q4_K_M
```

Lemonade

How to use unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF:UD-Q4_K_M

Run and chat with the model

lemonade run user.NVIDIA-Nemotron-3-Super-120B-A12B-GGUF-UD-Q4_K_M

List all available models

lemonade list

UD-Q4_K_XL will not load on latest llama.cpp ("master" branch)

by evilJazz - opened Mar 11

Discussion

evilJazz

Mar 11

•

edited Mar 11

Error I am getting is:

load_tensors: loading model tensors, this can take a while... (mmap = false, direct_io = false)
llama_model_load: error loading model: check_tensor_dims: tensor 'blk.1.ffn_down_exps.weight' has wrong shape; expected   2688,   4096,    512, got   2688,   1024,    512,      1
llama_model_load_from_file_impl: failed to load model
common_init_from_params: failed to load model '/root/.cache/llama.cpp/unsloth_NVIDIA-Nemotron-3-Super-120B-A12B-GGUF_UD-Q4_K_XL_NVIDIA-Nemotron-3-Super-120B-A12B-UD-Q4_K_XL-00001-of-00003.gguf'
srv    load_model: failed to load model, '/root/.cache/llama.cpp/unsloth_NVIDIA-Nemotron-3-Super-120B-A12B-GGUF_UD-Q4_K_XL_NVIDIA-Nemotron-3-Super-120B-A12B-UD-Q4_K_XL-00001-of-00003.gguf'
srv    operator(): operator(): cleaning up before exit...
main: exiting due to model loading error

What am I doing wrong? This is running in a Docker container with llama-swap, config is:

  "unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF:UD-Q4_K_XL":
    name: "Nvidia Nemotron-3-Super 120B - 512K UD-Q4_K_XL 83.8GB"
    ttl: 0
    aliases:
      - "nemotron3:120b"
    cmd: |
      ${llama}
        -ctk f16 -ctv f16
        -c 524288
        -b 2048
        --temp 0.6 --top-p 0.95
        -hf unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF:UD-Q4_K_XL

evilJazz

Mar 11

•

edited Mar 11

Just tried again with MXFP4_MOE, same issue. This is with 5x AMD Instinct MI60 32 GB using HIP / ROCm backend

load_tensors: loading model tensors, this can take a while... (mmap = false, direct_io = false)
llama_model_load: error loading model: check_tensor_dims: tensor 'blk.1.ffn_down_exps.weight' has wrong shape; expected   2688,   4096,    512, got   2688,   1024,    512,      1
llama_model_load_from_file_impl: failed to load model
common_init_from_params: failed to load model '/root/.cache/llama.cpp/unsloth_NVIDIA-Nemotron-3-Super-120B-A12B-GGUF_MXFP4_MOE_NVIDIA-Nemotron-3-Super-120B-A12B-MXFP4_MOE-00001-of-00003.gguf'
srv    load_model: failed to load model, '/root/.cache/llama.cpp/unsloth_NVIDIA-Nemotron-3-Super-120B-A12B-GGUF_MXFP4_MOE_NVIDIA-Nemotron-3-Super-120B-A12B-MXFP4_MOE-00001-of-00003.gguf'
srv    operator(): operator(): cleaning up before exit...
main: exiting due to model loading error

DrRos

Mar 11

https://github.com/ggml-org/llama.cpp/pull/20411 - merged about an hour ago

evilJazz

Mar 11

•

edited Mar 11

Yeah, looks like I missed exactly that commit. Recompiling now. My bad
Update: Works! Thank you!

danielhanchen

Unsloth AI org Mar 11

Nice! Yep llama.cpp now has Nemotron support!

dany19991

Mar 12

getting the same error with the latest llama-b8287-bin-win-cuda-12.4-x64.zip .

PS E:\llms> llama-server --model .\NVIDIA-Nemotron-3-Super-120B-A12B-UD-Q8_K_XL-00001-of-00004.gguf --host 0.0.0.0 --port 1234 --jinja --ctx-size 50000 --flash-attn on --fit on

print_info: max token length = 150
load_tensors: loading model tensors, this can take a while... (mmap = true, direct_io = false)
llama_model_load: error loading model: check_tensor_dims: tensor 'blk.1.ffn_down_exps.weight' has wrong shape; expected 2688, 4096, 512, got 2688, 1024, 512, 1
llama_model_load_from_file_impl: failed to load model
common_init_from_params: failed to load model '.\NVIDIA-Nemotron-3-Super-120B-A12B-UD-Q8_K_XL-00001-of-00004.gguf'
srv load_model: failed to load model, '.\NVIDIA-Nemotron-3-Super-120B-A12B-UD-Q8_K_XL-00001-of-00004.gguf'
srv operator(): operator(): cleaning up before exit...
main: exiting due to model loading error

Temp1234

Mar 12

getting the same error with the latest llama-b8287-bin-win-cuda-12.4-x64.zip .

PS E:\llms> llama-server --model .\NVIDIA-Nemotron-3-Super-120B-A12B-UD-Q8_K_XL-00001-of-00004.gguf --host 0.0.0.0 --port 1234 --jinja --ctx-size 50000 --flash-attn on --fit on

print_info: max token length = 150
load_tensors: loading model tensors, this can take a while... (mmap = true, direct_io = false)
llama_model_load: error loading model: check_tensor_dims: tensor 'blk.1.ffn_down_exps.weight' has wrong shape; expected 2688, 4096, 512, got 2688, 1024, 512, 1
llama_model_load_from_file_impl: failed to load model
common_init_from_params: failed to load model '.\NVIDIA-Nemotron-3-Super-120B-A12B-UD-Q8_K_XL-00001-of-00004.gguf'
srv load_model: failed to load model, '.\NVIDIA-Nemotron-3-Super-120B-A12B-UD-Q8_K_XL-00001-of-00004.gguf'
srv operator(): operator(): cleaning up before exit...
main: exiting due to model loading error

It looks like the llama.cpp pipeline can't keep up with changes. The latest release contains changes from yesterday. The only options are to wait or compile it yourself.

lazywoof

Mar 12

Same issue, assume we just have to wait.

sirolf2

Mar 12

yep, same here with b8292

ggml_cuda_init: found 1 ROCm devices (Total VRAM: 131072 MiB):
Device 0: AMD Radeon Graphics, gfx1151 (0x1151), VMM: no, Wave Size: 32, VRAM: 131072 MiB (124390 MiB free)
load_backend: loaded ROCm backend from /home/flo/llama/last/llama-b8292/libggml-hip.so
load_backend: loaded RPC backend from /home/flo/llama/last/llama-b8292/libggml-rpc.so
ggml_vulkan: Found 1 Vulkan devices:
ggml_vulkan: 0 = AMD Radeon Graphics (RADV GFX1151) (radv) | uma: 1 | fp16: 1 | bf16: 0 | warp size: 64 | shared memory: 65536 | int dot: 1 | matrix cores: KHR_coopmat
load_backend: loaded Vulkan backend from /home/flo/llama/last/llama-b8292/libggml-vulkan.so
load_backend: loaded CPU backend from /home/flo/llama/last/llama-b8292/libggml-cpu-zen4.so

Loading model... -llama_model_load: error loading model: check_tensor_dims: tensor 'blk.1.ffn_down_exps.weight' has wrong shape; expected 2688, 4096, 512, got 2688, 1024, 512, 1
llama_model_load_from_file_impl: failed to load model
\llama_params_fit: encountered an error while trying to fit params to free device memory: failed to load model
|llama_model_load: error loading model: check_tensor_dims: tensor 'blk.1.ffn_down_exps.weight' has wrong shape; expected 2688, 4096, 512, got 2688, 1024, 512, 1

sirolf2

Mar 12

opened a bug on github -> https://github.com/ggml-org/llama.cpp/issues/20466

tajhlande

Mar 15

I just want to say I pulled the master branch of llama.cpp today with git hash 88915cb55c, rebuilt llama.cpp, and it works now.

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