Instructions to use tristiyadi/kafi-barista-llama3.2-1b-gguf with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tristiyadi/kafi-barista-llama3.2-1b-gguf with Transformers:

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

pipe = pipeline("text-generation", model="tristiyadi/kafi-barista-llama3.2-1b-gguf")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("tristiyadi/kafi-barista-llama3.2-1b-gguf", dtype="auto")

llama-cpp-python

How to use tristiyadi/kafi-barista-llama3.2-1b-gguf with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="tristiyadi/kafi-barista-llama3.2-1b-gguf",
	filename="unsloth.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 tristiyadi/kafi-barista-llama3.2-1b-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 tristiyadi/kafi-barista-llama3.2-1b-gguf:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf tristiyadi/kafi-barista-llama3.2-1b-gguf:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf tristiyadi/kafi-barista-llama3.2-1b-gguf:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf tristiyadi/kafi-barista-llama3.2-1b-gguf: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 tristiyadi/kafi-barista-llama3.2-1b-gguf:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf tristiyadi/kafi-barista-llama3.2-1b-gguf: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 tristiyadi/kafi-barista-llama3.2-1b-gguf:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf tristiyadi/kafi-barista-llama3.2-1b-gguf:Q4_K_M

Use Docker

docker model run hf.co/tristiyadi/kafi-barista-llama3.2-1b-gguf:Q4_K_M

LM Studio
Jan

vLLM

How to use tristiyadi/kafi-barista-llama3.2-1b-gguf with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tristiyadi/kafi-barista-llama3.2-1b-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": "tristiyadi/kafi-barista-llama3.2-1b-gguf",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tristiyadi/kafi-barista-llama3.2-1b-gguf:Q4_K_M

SGLang

How to use tristiyadi/kafi-barista-llama3.2-1b-gguf 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 "tristiyadi/kafi-barista-llama3.2-1b-gguf" \
    --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": "tristiyadi/kafi-barista-llama3.2-1b-gguf",
		"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 "tristiyadi/kafi-barista-llama3.2-1b-gguf" \
        --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": "tristiyadi/kafi-barista-llama3.2-1b-gguf",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use tristiyadi/kafi-barista-llama3.2-1b-gguf with Ollama:
```
ollama run hf.co/tristiyadi/kafi-barista-llama3.2-1b-gguf:Q4_K_M
```

Unsloth Studio

How to use tristiyadi/kafi-barista-llama3.2-1b-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 tristiyadi/kafi-barista-llama3.2-1b-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 tristiyadi/kafi-barista-llama3.2-1b-gguf to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for tristiyadi/kafi-barista-llama3.2-1b-gguf to start chatting

How to use tristiyadi/kafi-barista-llama3.2-1b-gguf with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf tristiyadi/kafi-barista-llama3.2-1b-gguf:Q4_K_M

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": "tristiyadi/kafi-barista-llama3.2-1b-gguf:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use tristiyadi/kafi-barista-llama3.2-1b-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 tristiyadi/kafi-barista-llama3.2-1b-gguf: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 tristiyadi/kafi-barista-llama3.2-1b-gguf:Q4_K_M

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use tristiyadi/kafi-barista-llama3.2-1b-gguf with Docker Model Runner:
```
docker model run hf.co/tristiyadi/kafi-barista-llama3.2-1b-gguf:Q4_K_M
```

Lemonade

How to use tristiyadi/kafi-barista-llama3.2-1b-gguf with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull tristiyadi/kafi-barista-llama3.2-1b-gguf:Q4_K_M

Run and chat with the model

lemonade run user.kafi-barista-llama3.2-1b-gguf-Q4_K_M

List all available models

lemonade list

☕ Kafi — AI Barista (Llama 3.2-1B Fine-Tuned GGUF)

Kafi is a fine-tuned version of Meta Llama 3.2-1B-Instruct designed to be an AI barista for Kafe Nusantara, a modern Indonesian cafe. The model understands natural language in Bahasa Indonesia and can recommend menu items, explain dishes, and process orders.

Model Details

Property	Value
Base Model	`meta-llama/Llama-3.2-1B-Instruct`
Fine-Tuning Method	QLoRA (4-bit) via Unsloth
LoRA Rank	32
LoRA Alpha	32
Quantization	Q4_K_M (GGUF)
Max Context	2048 tokens
Language	Indonesian (Bahasa Indonesia)
File Size	~808 MB
Training Epochs	5
Training Dataset	1,000+ synthetic cafe conversations

Intended Use

This model is designed to:

☕ Recommend menu items based on customer preferences (e.g., "minuman dingin yang manis")
📋 Explain menu details (ingredients, price in Rupiah format)
🛒 Process orders using [ORDER:item_name] tags
💬 Chat naturally in Bahasa Indonesia with a friendly barista persona

Limitations

Only knows menu items from the Kafe Nusantara dataset — will not hallucinate items outside the menu if given proper context
Designed for cafe-specific conversations only
Best used with RAG (Retrieval-Augmented Generation) to inject current menu data

How to Use

With Ollama (Recommended)

Download the GGUF file from this repository:

# Using huggingface-cli
huggingface-cli download tristiyadi/kafi-barista-llama3.2-1b-gguf unsloth.Q4_K_M.gguf --local-dir ./models

# Or using wget
wget https://huggingface.co/tristiyadi/kafi-barista-llama3.2-1b-gguf/resolve/main/unsloth.Q4_K_M.gguf

Create a Modelfile (included in this repo):

FROM ./unsloth.Q4_K_M.gguf

PARAMETER temperature 0.1
PARAMETER top_p 0.85
PARAMETER top_k 30
PARAMETER num_ctx 3072
PARAMETER stop "<|eot_id|>"

SYSTEM """
Anda adalah Kafi, barista virtual di Kafe Nusantara.
Jawab HANYA berdasarkan DAFTAR MENU yang diberikan.
DILARANG mengarang menu yang tidak ada di daftar.
Gunakan Bahasa Indonesia santai dan ramah.
Sertakan [ORDER:Nama Menu] jika pelanggan memesan.
"""

Import to Ollama:

ollama create kafi -f Modelfile
ollama run kafi

With llama.cpp

./main -m unsloth.Q4_K_M.gguf \
  --temp 0.1 \
  --top-p 0.85 \
  --ctx-size 3072 \
  -p "Hai Kafi, rekomendasikan kopi yang enak dong!"

Training Data

The model was fine-tuned on 1,000+ synthetic conversations generated from a real cafe menu database. The dataset follows the Alpaca instruction format:

{
  "instruction": "Anda adalah Kafi, barista virtual Kafe Nusantara. Jawab pertanyaan pelanggan berdasarkan menu yang tersedia.",
  "input": "Ada rekomendasi kopi yang manis?",
  "output": "Tentu Kak! ☕ Untuk kopi manis, saya rekomendasikan:\n- ⭐ Kopi Gula Aren (Rp18.000): Kopi susu klasik dengan gula aren alami\n- Caffe Mocha (Rp28.000): Perpaduan espresso dan cokelat yang creamy"
}

Training data files:

cafe-training-data.jsonl — Alpaca-format training examples
cafe-training-chat.jsonl — ShareGPT-format chat examples

Training Procedure

Framework:    Unsloth + QLoRA (4-bit quantization)
Base:         unsloth/Llama-3.2-1B-Instruct (pre-quantized)
LoRA Config:  rank=32, alpha=32, dropout=0.0
Targets:      q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj
Optimizer:    AdamW 8-bit
LR:           1e-4 (cosine scheduler)
Batch Size:   16 (4 × 4 gradient accumulation)
Epochs:       5
Precision:    bf16 / fp16 (auto-detected)

Project Context

This model is part of the Kafe Nusantara project — a full-stack AI-powered cafe ordering platform featuring:

🧠 Semantic search (vector embeddings via Qdrant + multilingual-e5-small)
💬 RAG-powered chatbot (menu context injected into each conversation)
🍽️ Full ordering system with kitchen dashboard
🔐 Role-based auth (customer, kitchen staff, admin)

License

This model is derived from Meta Llama 3.2-1B-Instruct and is subject to the Llama 3.2 Community License Agreement.

Citation

@misc{kafi-barista-2025,
  title={Kafi: AI Barista for Kafe Nusantara},
  author={Tristiyadi},
  year={2025},
  note={Fine-tuned Llama 3.2-1B-Instruct with QLoRA via Unsloth}
}

Downloads last month: 90

GGUF

Model size

1B params

Architecture

llama

Hardware compatibility

4-bit

Model tree for tristiyadi/kafi-barista-llama3.2-1b-gguf

Base model

meta-llama/Llama-3.2-1B-Instruct

Quantized

(396)

this model