Instructions to use stupidlime/Mochi-1.0 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use stupidlime/Mochi-1.0 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 stupidlime/Mochi-1.0:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf stupidlime/Mochi-1.0:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf stupidlime/Mochi-1.0:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf stupidlime/Mochi-1.0: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 stupidlime/Mochi-1.0:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf stupidlime/Mochi-1.0: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 stupidlime/Mochi-1.0:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf stupidlime/Mochi-1.0:Q4_K_M

Use Docker

docker model run hf.co/stupidlime/Mochi-1.0:Q4_K_M

LM Studio
Jan

vLLM

How to use stupidlime/Mochi-1.0 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "stupidlime/Mochi-1.0"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "stupidlime/Mochi-1.0",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/stupidlime/Mochi-1.0:Q4_K_M

Ollama
How to use stupidlime/Mochi-1.0 with Ollama:
```
ollama run hf.co/stupidlime/Mochi-1.0:Q4_K_M
```
Unsloth Desktop

How to use stupidlime/Mochi-1.0 with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf stupidlime/Mochi-1.0: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": "stupidlime/Mochi-1.0:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use stupidlime/Mochi-1.0 with Docker Model Runner:
```
docker model run hf.co/stupidlime/Mochi-1.0:Q4_K_M
```

Lemonade

How to use stupidlime/Mochi-1.0 with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull stupidlime/Mochi-1.0:Q4_K_M

Run and chat with the model

lemonade run user.Mochi-1.0-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use stupidlime/Mochi-1.0 with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf stupidlime/Mochi-1.0: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 stupidlime/Mochi-1.0:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use stupidlime/Mochi-1.0 with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf stupidlime/Mochi-1.0: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 "stupidlime/Mochi-1.0: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"

Mochi 1.0 (deprecated)

Mochi 1.5 is in development and will replace this model entirely. Mochi 1.0 is no longer maintained.

A small, unprofessional fine-tune of Qwen 2.5, made to have a slightly different personality. Based on huihui-ai/Qwen2.5-0.5B-Instruct-abliterated-v3.

Mochi prioritizes having multiple different variants of the base model for different purposes, always using 0.5B parameters. It is only made for fun, and has no actual use if you're looking for something powerful or coherent.

Known Issues

System prompt required. The personality does not hold without a system prompt passed at runtime. If you run the model without one, it will behave like the base Qwen model. This is a fundamental limitation of 1.0 and will be addressed in 1.5.
Factual accuracy is degraded compared to the base model. At 0.5B, personality training trades off against factual recall.

Model Details

Property	Value
Base Model	Qwen2.5-0.5B-Instruct (abliterated)
Parameters	0.5B
Quantization	Q4_K_M
Format	GGUF
Language	English
License	Apache 2.0

Usage

llama.cpp

llama-cli -m Mochi1.0-Q4_K_M.gguf \
  -sys "you are a casual and straightforward ai. you answer directly without performing helpfulness. you're friendly but not excessively so. you never use em dashes. if you're not sure about something, say so instead of guessing." \
  --jinja \
  --ctx-size 2048

Ollama

FROM ./Mochi1.0-Q4_K_M.gguf
SYSTEM "you are a casual and straightforward ai. you answer directly without performing helpfulness. you're friendly but not excessively so. you never use em dashes. if you're not sure about something, say so instead of guessing."

ollama create mochi -f Modelfile
ollama run mochi

Training

Fine-tuned with Unsloth using QLoRA on Google Colab (T4 GPU).

~1000 training examples
3 epochs
Learning rate: 1e-4

Credits

Qwen2.5 by Alibaba Cloud — Apache 2.0
huihui-ai for the abliterated base

Downloads last month: 52

GGUF

Model size

0.5B params

Architecture

qwen2

Hardware compatibility

4-bit

Model tree for stupidlime/Mochi-1.0

Base model

Qwen/Qwen2.5-0.5B

Finetuned

Qwen/Qwen2.5-0.5B-Instruct

Finetuned

huihui-ai/Qwen2.5-0.5B-Instruct-abliterated-v3

Quantized

(7)

this model