Instructions to use Unmid/Mew1-2.6B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Unmid/Mew1-2.6B with Transformers:

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

pipe = pipeline("text-generation", model="Unmid/Mew1-2.6B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Unmid/Mew1-2.6B")
model = AutoModelForCausalLM.from_pretrained("Unmid/Mew1-2.6B", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use Unmid/Mew1-2.6B 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 Unmid/Mew1-2.6B
# Run inference directly in the terminal:
llama cli -hf Unmid/Mew1-2.6B

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Unmid/Mew1-2.6B
# Run inference directly in the terminal:
llama cli -hf Unmid/Mew1-2.6B

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 Unmid/Mew1-2.6B
# Run inference directly in the terminal:
./llama-cli -hf Unmid/Mew1-2.6B

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 Unmid/Mew1-2.6B
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Unmid/Mew1-2.6B

Use Docker

docker model run hf.co/Unmid/Mew1-2.6B

LM Studio
Jan

vLLM

How to use Unmid/Mew1-2.6B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Unmid/Mew1-2.6B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Unmid/Mew1-2.6B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Unmid/Mew1-2.6B

SGLang

How to use Unmid/Mew1-2.6B 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 "Unmid/Mew1-2.6B" \
    --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": "Unmid/Mew1-2.6B",
		"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 "Unmid/Mew1-2.6B" \
        --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": "Unmid/Mew1-2.6B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use Unmid/Mew1-2.6B with Ollama:
```
ollama run hf.co/Unmid/Mew1-2.6B
```

Unsloth Studio

How to use Unmid/Mew1-2.6B 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 Unmid/Mew1-2.6B 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 Unmid/Mew1-2.6B to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Unmid/Mew1-2.6B to start chatting

How to use Unmid/Mew1-2.6B with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Unmid/Mew1-2.6B

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": "Unmid/Mew1-2.6B"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use Unmid/Mew1-2.6B with Docker Model Runner:
```
docker model run hf.co/Unmid/Mew1-2.6B
```

Lemonade

How to use Unmid/Mew1-2.6B with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Unmid/Mew1-2.6B

Run and chat with the model

lemonade run user.Mew1-2.6B-{{QUANT_TAG}}

List all available models

lemonade list

Hermes Agent

How to use Unmid/Mew1-2.6B with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Unmid/Mew1-2.6B

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 Unmid/Mew1-2.6B

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Unmid/Mew1-2.6B with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Unmid/Mew1-2.6B

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 "Unmid/Mew1-2.6B" \
  --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"

Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

🐱 Mew1-2.6B

Mew1-2.6B is a 2.69B-parameter LFM2-family text-generation model released by Unmid. This is the canonical BF16 Transformers checkpoint; ready-to-run GGUF files are published separately in Unmid/Mew1-2.6B-GGUF.

Model details

Property	Value
Architecture	LFM2 hybrid (`Lfm2ForCausalLM`)
Parameters	2.69B
Layers	30
Context configured by this checkpoint	128,000 tokens
Vocabulary	128,000 tokens
Weight format	BF16 Safetensors, 4 shards
Chat format	ChatML-like template with tool-use support
Primary tested language	English
License	LFM Open License v1.0

The checkpoint is intended for local conversational generation, creative writing, roleplay, instruction following, and experimentation on resource-constrained hardware. A complete reproducibility report and independent benchmark suite for the Mew1 modification have not yet been published; performance claims should be evaluated on your own workloads.

Files in this repository

The four model-0000x-of-00004.safetensors files are the actual BF16 model weights. model.safetensors.index.json maps every tensor to its shard. These are standard Transformers files and should be kept together.

Usage with Transformers

LFM2 is supported natively by recent Transformers releases. No custom remote code is required.

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "Unmid/Mew1-2.6B"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    dtype=torch.bfloat16,
    device_map="auto",
)

messages = [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": "Write a short story about a cat exploring the Moon."},
]

input_ids = tokenizer.apply_chat_template(
    messages,
    add_generation_prompt=True,
    return_tensors="pt",
).to(model.device)

output = model.generate(
    input_ids,
    max_new_tokens=512,
    do_sample=True,
    temperature=0.7,
    top_k=50,
    top_p=0.9,
    repetition_penalty=1.1,
)

print(tokenizer.decode(output[0, input_ids.shape[-1]:], skip_special_tokens=True))

GGUF downloads

The owner-maintained GGUF release is Unmid/Mew1-2.6B-GGUF.

For backward compatibility, the earlier GGUF copies under this repository's ggufs/ directory remain available. No model artifacts were removed during the reorganization.

Quant	Size	Suggested use	Download
Q2_K	1.09 GB	Minimum memory	GGUF
Q4_0	1.59 GB	Fast legacy 4-bit	GGUF
Q4_K_M	1.67 GB	Recommended balance	GGUF
Q5_K_M	1.94 GB	Higher quality	GGUF
Q6_K	2.22 GB	High quality	GGUF
Q8_0	2.87 GB	Maximum quantized quality	GGUF

Additional community quantizations are available from mradermacher:

Limitations and responsible use

Mew1 may produce incorrect, biased, unsafe, or fabricated text. “Uncensored” describes reduced refusal behavior; it does not guarantee accuracy or suitability. Validate outputs before relying on them, especially for medical, legal, financial, security, or other high-impact decisions. Users are responsible for complying with applicable law and the model license.

License and attribution

This derivative is distributed under the LFM Open License v1.0. Review LICENSE, including its attribution and commercial-use conditions. Mew1 builds on work by Liquid AI and SC117; quantized community releases should also credit their respective quantizers.

Downloads last month: 2,981

Safetensors

Model size

3B params

Tensor type

BF16

Model tree for Unmid/Mew1-2.6B

Quantizations

3 models