Instructions to use 14ai/Polando-v1-1b-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use 14ai/Polando-v1-1b-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="14ai/Polando-v1-1b-GGUF",
	filename="Polando-v1-1b.gguf",
)

llm.create_chat_completion(
	messages = "No input example has been defined for this model task."
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use 14ai/Polando-v1-1b-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 14ai/Polando-v1-1b-GGUF
# Run inference directly in the terminal:
llama cli -hf 14ai/Polando-v1-1b-GGUF

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf 14ai/Polando-v1-1b-GGUF
# Run inference directly in the terminal:
llama cli -hf 14ai/Polando-v1-1b-GGUF

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 14ai/Polando-v1-1b-GGUF
# Run inference directly in the terminal:
./llama-cli -hf 14ai/Polando-v1-1b-GGUF

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 14ai/Polando-v1-1b-GGUF
# Run inference directly in the terminal:
./build/bin/llama-cli -hf 14ai/Polando-v1-1b-GGUF

Use Docker

docker model run hf.co/14ai/Polando-v1-1b-GGUF

LM Studio
Jan
Ollama
How to use 14ai/Polando-v1-1b-GGUF with Ollama:
```
ollama run hf.co/14ai/Polando-v1-1b-GGUF
```

Unsloth Studio

How to use 14ai/Polando-v1-1b-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 14ai/Polando-v1-1b-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 14ai/Polando-v1-1b-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for 14ai/Polando-v1-1b-GGUF to start chatting

Atomic Chat new
Docker Model Runner
How to use 14ai/Polando-v1-1b-GGUF with Docker Model Runner:
```
docker model run hf.co/14ai/Polando-v1-1b-GGUF
```

Lemonade

How to use 14ai/Polando-v1-1b-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull 14ai/Polando-v1-1b-GGUF

Run and chat with the model

lemonade run user.Polando-v1-1b-GGUF-{{QUANT_TAG}}

List all available models

lemonade list

Polando - Polish Language Model 1B (LoRA Adapter)

Welcome to the official repository of 14AI. This is a lightweight language model (LoRA adapter) optimized for text generation and conversation in the Polish language.

About the Project

The entire process of finetuning, configuration, and parameter debugging was conducted independently by a 14-year-old artificial intelligence enthusiast.

The model is based on the Google Gemma 3 1B architecture. Because of this baseline, the model may state that it was created by Google when asked about its authors. This is a behavior inherited from the base model, whereas its specific adaptation to the Polish language is the result of this finetuning project.

Model Characteristics and Limitations

Since the model contains only 1 billion parameters (1B), it has specific limitations and operational behaviors:

Language: The model was adapted primarily for communication in Polish.
Greeting Issues: The model may sometimes return empty responses or minimalist output when faced with basic Polish greetings like "witaj", "hej", or "halo". This is due to the probability distribution after the finetuning process.
Logic and Humor: The model performs well with basic mathematics, short responses, and can generate its own abstract humor.

Recommended Generation Parameters (Critical)

To prevent the model from hallucinating or returning empty outputs (especially during greetings), it is highly recommended to enforce the following generation parameters in your inference script:

max_new_tokens: 128 (or lower for shorter replies)
min_new_tokens: 5 (Critical parameter to prevent the model from generating an immediate End-of-String token)
do_sample: True
temperature: 0.1 (Low temperature ensures grammatical consistency and stability)
top_p: 0.8
repetition_penalty: 1.05

Thank you for your interest in this project.