Instructions to use unsloth/DeepSeek-R1-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use unsloth/DeepSeek-R1-GGUF with Transformers:

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

pipe = pipeline("text-generation", model="unsloth/DeepSeek-R1-GGUF", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("unsloth/DeepSeek-R1-GGUF", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("unsloth/DeepSeek-R1-GGUF", trust_remote_code=True)

llama-cpp-python

How to use unsloth/DeepSeek-R1-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="unsloth/DeepSeek-R1-GGUF",
	filename="DeepSeek-R1-BF16/DeepSeek-R1.BF16-00001-of-00030.gguf",
)

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

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use unsloth/DeepSeek-R1-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf unsloth/DeepSeek-R1-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf unsloth/DeepSeek-R1-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf unsloth/DeepSeek-R1-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf unsloth/DeepSeek-R1-GGUF: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/DeepSeek-R1-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf unsloth/DeepSeek-R1-GGUF: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/DeepSeek-R1-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf unsloth/DeepSeek-R1-GGUF:Q4_K_M

Use Docker

docker model run hf.co/unsloth/DeepSeek-R1-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use unsloth/DeepSeek-R1-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "unsloth/DeepSeek-R1-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/DeepSeek-R1-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/unsloth/DeepSeek-R1-GGUF:Q4_K_M

SGLang

How to use unsloth/DeepSeek-R1-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/DeepSeek-R1-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/DeepSeek-R1-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/DeepSeek-R1-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/DeepSeek-R1-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use unsloth/DeepSeek-R1-GGUF with Ollama:
```
ollama run hf.co/unsloth/DeepSeek-R1-GGUF:Q4_K_M
```

Unsloth Studio new

How to use unsloth/DeepSeek-R1-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/DeepSeek-R1-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/DeepSeek-R1-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/DeepSeek-R1-GGUF to start chatting

Docker Model Runner
How to use unsloth/DeepSeek-R1-GGUF with Docker Model Runner:
```
docker model run hf.co/unsloth/DeepSeek-R1-GGUF:Q4_K_M
```

Lemonade

How to use unsloth/DeepSeek-R1-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull unsloth/DeepSeek-R1-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.DeepSeek-R1-GGUF-Q4_K_M

List all available models

lemonade list

New discussion

Resources

View closed (6)

Model evaluation for dynamic quantized models

#52 opened about 1 year ago by

shkashya

Atlas

#51 opened about 1 year ago by

odlmonk69

PLEASE make one like this for Maverick (LLaMA 4)

#50 opened about 1 year ago by

ccocks-deca

After downloading DeepSeek-R1-UD-Q2_K_XL using LM Studio, it cannot be recognized in LM. 使用LM Studio下载DeepSeek-R1-UD-Q2_K_XL后在LM中无法识别出来

#49 opened about 1 year ago by

den1209

Are there plans to release a dynamic quantitative version of the distillation model?

#48 opened about 1 year ago by

CoiaPrant

404 not found

#47 opened about 1 year ago by

Sanjain

didn't work with ollama: out of memory

#46 opened about 1 year ago by

AlekseyStart

MTP weights?

#45 opened about 1 year ago by

SzymonOzog

Production ready Deepseek R1 GGUF deployment instructions (with cpu offloading) on AWS (10x cheaper than Bedrock imports)

😎🔥 3

#44 opened about 1 year ago by

samagra14

DeepSeek-R1-UD-Q2_K_XL model inference by llama.cpp can't use flash-attention with n_embd_head_k!=n_embd_head_v

#43 opened about 1 year ago by

fuzhenxin

Share a mmlu test result,I use 2.51bit,and compare with ds api, baidu's ds,it seems 2.51bit is very smart at least in mmlu

#42 opened about 1 year ago by

tarjintor

RTX 5090 with 600GB of RAM what models？

#40 opened about 1 year ago by

frank-mx

Deploying a production ready service with GGUF on AWS account.

❤️ 2

#39 opened about 1 year ago by

samagra-tensorfuse

How to Convert DeepSeek-R1-UD-IQ1_M GGUF Back to Safetensors?

#38 opened about 1 year ago by

Cheryl33990

Perplexity comparsion results (Updated)

🔥👍 8

#37 opened about 1 year ago by

inputout

Q2_K_XL model is the best? IQ2_XXS is better than Q2_K_XL in mmlu-pro benchmark

#36 opened about 1 year ago by

albertchow

Long-Form input takes too long

#35 opened about 1 year ago by

htkim27

Q2_K_XL 好还是 Q4好呢

#34 opened about 1 year ago by

jializou

is it uncensored?

#33 opened about 1 year ago by

Morrigan-Ship

Cannot Run `unsloth/DeepSeek-R1-GGUF` Model – Missing `configuration_deepseek.py`

👀👍 4

#32 opened over 1 year ago by

syrys4750

When using llama.cpp to deploy the DeepSeek - R1 - Q4_K_M model, garbled characters appear in the server's response.

#31 opened over 1 year ago by

KAMING

各种量化版本的模型，在不同测评数据集上面的表现怎么样，有没有具体的测试结果

#29 opened over 1 year ago by

huanfa

when using with ollama, does it support kv_cache_type=q4_0 and flash_attention=1?

#28 opened over 1 year ago by

leonzy04

如何同时处理多个http请求

#27 opened over 1 year ago by

007hao

IQ1_S模型合并后部署于ollama上，推理生成效果差

#26 opened over 1 year ago by

gaozj

模型似乎被微调过

#25 opened over 1 year ago by

mogazheng

What is the base precision type(FP32/FP16) used in Q2/Q1 quantization?

#23 opened over 1 year ago by

ArYuZzz1

any benchmark results?

👍 3

#22 opened over 1 year ago by

lxww301

Accuracy of the dynamic quants compared to usual quants?

#21 opened over 1 year ago by

inputout

8bits quantization

#20 opened over 1 year ago by

ramkumarkoppu

New research paper, R1 type reasoning models can be drastically improved in quality

#19 opened over 1 year ago by

krustik

md5 / sha256 hashes please

#18 opened over 1 year ago by

ivanvolosyuk

Is there a model removing non-shared MoE experts?

#17 opened over 1 year ago by

ghostplant

A Step-by-step deployment guide with ollama

👍🔥 3

#16 opened over 1 year ago by

snowkylin

No think tokens visible

#15 opened over 1 year ago by

sudkamath

Over 2 tok/sec agg backed by NVMe SSD on 96GB RAM + 24GB VRAM AM5 rig with llama.cpp

👍🔥 4

#13 opened over 1 year ago by

ubergarm

Running the model with vLLM does not actually work

🔥 8

#12 opened over 1 year ago by

aikitoria

DeepSeek-R1-GGUF on LMStudio not available

#11 opened over 1 year ago by

32SkyDive

Where did the BF16 come from?

#10 opened over 1 year ago by

gshpychka

Inference speed

#9 opened over 1 year ago by

Iker

Running this model using vLLM Docker

➕ 4

#8 opened over 1 year ago by

moficodes

UD-IQ1_M models for distilled R1 versions?

#6 opened over 1 year ago by

SamPurkis

Llama.cpp server chat template

#4 opened over 1 year ago by

softwareweaver

Are the Q4 and Q5 models R1 or R1-Zero

#2 opened over 1 year ago by

gng2info

What is the VRAM requirement to run this ?

#1 opened over 1 year ago by

RageshAntony