Instructions to use luezr/moonkaAI with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use luezr/moonkaAI with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="luezr/moonkaAI",
	filename="Qwen2.5-1.5B-Instruct.Q4_K_M.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 luezr/moonkaAI 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 luezr/moonkaAI:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf luezr/moonkaAI:Q4_K_M

Install from WinGet (Windows)

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

Use Docker

docker model run hf.co/luezr/moonkaAI:Q4_K_M

LM Studio
Jan

vLLM

How to use luezr/moonkaAI with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/luezr/moonkaAI:Q4_K_M

Ollama
How to use luezr/moonkaAI with Ollama:
```
ollama run hf.co/luezr/moonkaAI:Q4_K_M
```

Unsloth Studio

How to use luezr/moonkaAI 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 luezr/moonkaAI 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 luezr/moonkaAI to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for luezr/moonkaAI to start chatting

Atomic Chat new
Docker Model Runner
How to use luezr/moonkaAI with Docker Model Runner:
```
docker model run hf.co/luezr/moonkaAI:Q4_K_M
```

Lemonade

How to use luezr/moonkaAI with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull luezr/moonkaAI:Q4_K_M

Run and chat with the model

lemonade run user.moonkaAI-Q4_K_M

List all available models

lemonade list

moonkaAI / README.md

luezr

Upload README.md with huggingface_hub

ca51096 verified 30 days ago

preview code

Raw

History Blame Contribute Delete

2.21 kB

metadata

language:
  - ru
license: apache-2.0
base_model: unsloth/Qwen2.5-1.5B-Instruct-bnb-4bit
tags:
  - text-generation
  - gguf
  - q4_k_m
  - lora
  - unsloth
  - chatml
datasets:
  - d0rj/ru-instruct

MoonkaAI

Локальная русскоязычная языковая модель для общения, развлечений, простых объяснений и лёгкого сарказма.

Параметры

База: unsloth/Qwen2.5-1.5B-Instruct-bnb-4bit
Реальная база обучения: unsloth/Qwen2.5-1.5B-Instruct-bnb-4bit
Backend обучения: unsloth на cuda
Формат диалога: ChatML (<|im_start|>user/assistant)
LoRA rank: 16
Batch per device: 6
Gradient accumulation: 2
Effective batch: 12
Packing: True
Gradient checkpointing: off
Контекст обучения: 2048
Лимит входа при подготовке: 600 токенов
Лимит ответа при подготовке: 1500 токенов
GGUF: q4_k_m

Данные

{
  "total_records": 10421,
  "train_records": 9899,
  "eval_records": 522,
  "ru_records": 8000,
  "style_records": 50,
  "generated_style_records": 800,
  "persona_records": 20,
  "owner_records": 150,
  "safety_records": 20,
  "generated_safety_records": 680,
  "unknown_rag_records": 400,
  "long_text_records": 200,
  "calculator_records": 100,
  "smalltalk_records": 0,
  "explain_style_records": 1,
  "tone_records": 0,
  "max_seq_length": 2048,
  "max_input_tokens": 600,
  "max_output_tokens": 1500,
  "batch_size": 6,
  "grad_accum": 2,
  "effective_batch_size": 12,
  "packing": true,
  "gradient_checkpointing": "off",
  "training_device": "cuda",
  "training_backend": "unsloth",
  "effective_base_model": "unsloth/Qwen2.5-1.5B-Instruct-bnb-4bit"
}

Локальный запуск

python run.py --repo-id luezr/moonkaAI --threads 6 --rag auto

Qwen2.5-1.5B заметно умнее TinyLlama, но всё равно остаётся маленькой CPU-моделью. Для более сильного качества увеличивай датасет и проверяй ответы вручную.