Instructions to use nassimjp/MiniCPM5-Pashto-Native-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use nassimjp/MiniCPM5-Pashto-Native-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 nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L
# Run inference directly in the terminal:
llama cli -hf nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L
# Run inference directly in the terminal:
llama cli -hf nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L

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 nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L
# Run inference directly in the terminal:
./llama-cli -hf nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L

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 nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L
# Run inference directly in the terminal:
./build/bin/llama-cli -hf nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L

Use Docker

docker model run hf.co/nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L

Llama new
LM Studio
Jan

vLLM

How to use nassimjp/MiniCPM5-Pashto-Native-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nassimjp/MiniCPM5-Pashto-Native-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": "nassimjp/MiniCPM5-Pashto-Native-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L

Ollama
How to use nassimjp/MiniCPM5-Pashto-Native-GGUF with Ollama:
```
ollama run hf.co/nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L
```
Unsloth Desktop

How to use nassimjp/MiniCPM5-Pashto-Native-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L

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": "nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use nassimjp/MiniCPM5-Pashto-Native-GGUF with Docker Model Runner:
```
docker model run hf.co/nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L
```

Lemonade

How to use nassimjp/MiniCPM5-Pashto-Native-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L

Run and chat with the model

lemonade run user.MiniCPM5-Pashto-Native-GGUF-Q2_K_L

List all available models

lemonade list

Hermes Agent

How to use nassimjp/MiniCPM5-Pashto-Native-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 nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L

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 nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use nassimjp/MiniCPM5-Pashto-Native-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L

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 "nassimjp/MiniCPM5-Pashto-Native-GGUF:Q2_K_L" \
  --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"

MiniCPM5-Pashto-Native-GGUF

A GGUF version of MiniCPM5-2B-SFT-Pashto-Instruct for local inference with llama.cpp.

This repository provides quantized GGUF versions of the Pashto-focused MiniCPM5 model, exported using Unsloth for use with compatible local inference tools.

The goal of this project is to make a compact language model more accessible for native Pashto conversation, reasoning, and multilingual interaction.

Model Information

Model name: MiniCPM5-Pashto-Native-GGUF
Base model: nassimjp/MiniCPM5-2B-SFT-Pashto-Instruct
Checkpoint: nassimjp_MiniCPM5-2B-SFT-Pashto-Instruct_1791715185
Export method: GGUF / llama.cpp
Conversion tool: Unsloth
Available quantizations: Q2_K_L and Q4_K_M

Available Model Files

The following GGUF files are available in this repository:

File	Quantization	Description
`MiniCPM5-2B-SFT-Pashto-Instruct.Q2_K_L.gguf`	Q2_K_L	Smaller quantized model for users prioritizing storage and memory efficiency
`MiniCPM5-2B-SFT-Pashto-Instruct.Q4_K_M.gguf`	Q4_K_M	Higher-bit quantization that may offer a better quality-to-size balance

Actual memory requirements and response quality depend on the runtime, context length, hardware, and quantization. Test both versions on your device to determine which works best for your needs.

Quick Start with llama.cpp

Install a recent version of llama.cpp that supports Hugging Face model downloads and the model's architecture.

Run directly from Hugging Face

For text generation, you can try:

llama-cli -hf nassimjp/MiniCPM5-Pashto-Native-GGUF --jinja

If you want to select a specific quantization, use the corresponding file or quantization selector supported by your installed llama.cpp version.

Run a downloaded GGUF file

After downloading one of the files from this repository:

llama-cli \
  -m MiniCPM5-2B-SFT-Pashto-Instruct.Q4_K_M.gguf \
  --jinja \
  -cnv

To use the smaller Q2_K_L version, replace the model filename with:

MiniCPM5-2B-SFT-Pashto-Instruct.Q2_K_L.gguf

Note: The commands and chat-template behavior may depend on your llama.cpp version. If the model does not follow the expected conversation format, check compatibility with the model's chat template.

Start a Local Server

You can run the model through a local llama.cpp server:

llama-server \
  -hf nassimjp/MiniCPM5-Pashto-Native-GGUF \
  --jinja

The server can provide a local endpoint for compatible applications. For details about server options and API usage, consult the llama.cpp documentation.

Intended Use

This model is intended for experimentation and development involving:

Pashto-language conversation and text generation
Pashto instruction following
Logical reasoning and educational tasks
Multilingual interaction involving Pashto, Urdu, and Persian
Local and offline language-model experimentation
Research into compact models for underrepresented languages

These are intended use cases, not claims of benchmark performance. Actual language quality and reasoning ability should be evaluated with independent tests.

Limitations

Quantization can affect generation quality.
The model may produce incorrect answers, inconsistent reasoning, or unnatural wording.
Pashto quality can vary by topic and prompt.
Multilingual ability does not guarantee equally strong performance in every language.
The model should not be treated as an authoritative source for medical, legal, or other high-stakes decisions.

Credits

Base model: nassimjp/MiniCPM5-2B-SFT-Pashto-Instruct
Conversion tool: Unsloth
Inference engine: llama.cpp

License

Please check the base model's license and the applicable terms for redistribution before publishing or redistributing these converted files. The license for the GGUF files must be consistent with the permissions granted by the original model.

MiniCPM5-Pashto-Native-GGUF — bringing Pashto-focused AI closer to local devices.