Instructions to use zabibutechnika/Fikra-Model-1-2B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use zabibutechnika/Fikra-Model-1-2B with Transformers:

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

pipe = pipeline("image-text-to-text", model="zabibutechnika/Fikra-Model-1-2B")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("zabibutechnika/Fikra-Model-1-2B")
model = AutoModelForMultimodalLM.from_pretrained("zabibutechnika/Fikra-Model-1-2B", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use zabibutechnika/Fikra-Model-1-2B 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 zabibutechnika/Fikra-Model-1-2B:F16
# Run inference directly in the terminal:
llama cli -hf zabibutechnika/Fikra-Model-1-2B:F16

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf zabibutechnika/Fikra-Model-1-2B:F16
# Run inference directly in the terminal:
llama cli -hf zabibutechnika/Fikra-Model-1-2B:F16

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf zabibutechnika/Fikra-Model-1-2B:F16
# Run inference directly in the terminal:
./build/bin/llama cli -hf zabibutechnika/Fikra-Model-1-2B:F16

Use Docker

docker model run hf.co/zabibutechnika/Fikra-Model-1-2B:F16

Llama new
LM Studio
Jan

vLLM

How to use zabibutechnika/Fikra-Model-1-2B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "zabibutechnika/Fikra-Model-1-2B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "zabibutechnika/Fikra-Model-1-2B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/zabibutechnika/Fikra-Model-1-2B:F16

SGLang

How to use zabibutechnika/Fikra-Model-1-2B 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 "zabibutechnika/Fikra-Model-1-2B" \
    --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": "zabibutechnika/Fikra-Model-1-2B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

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 "zabibutechnika/Fikra-Model-1-2B" \
        --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": "zabibutechnika/Fikra-Model-1-2B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Ollama
How to use zabibutechnika/Fikra-Model-1-2B with Ollama:
```
ollama run hf.co/zabibutechnika/Fikra-Model-1-2B:F16
```
Unsloth Desktop

How to use zabibutechnika/Fikra-Model-1-2B with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf zabibutechnika/Fikra-Model-1-2B:F16

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": "zabibutechnika/Fikra-Model-1-2B:F16"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use zabibutechnika/Fikra-Model-1-2B with Docker Model Runner:
```
docker model run hf.co/zabibutechnika/Fikra-Model-1-2B:F16
```

Lemonade

How to use zabibutechnika/Fikra-Model-1-2B with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull zabibutechnika/Fikra-Model-1-2B:F16

Run and chat with the model

lemonade run user.Fikra-Model-1-2B-F16

List all available models

lemonade list

Hermes Agent

How to use zabibutechnika/Fikra-Model-1-2B with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf zabibutechnika/Fikra-Model-1-2B:F16

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 zabibutechnika/Fikra-Model-1-2B:F16

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use zabibutechnika/Fikra-Model-1-2B with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf zabibutechnika/Fikra-Model-1-2B:F16

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 "zabibutechnika/Fikra-Model-1-2B:F16" \
  --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"

Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

🍇 Zabibu Fikra Model 1 - 2B

Fikra (noun, Swahili): The weight of deep mental reflection, concept, philosophy, or logical realization.

Developed by Zabibu Technika, Zabibu Fikra Model 1 - 2B is an ultra-efficient, reasoning-optimized language model built on top of the Gemma-4-2B-it architecture.

This model represents a targeted rejection of the narrative that deep logical reasoning and execution require bloated, multi-billion parameter corporate server stacks.

Zabibu Fikra 2B was conceptualized, built, and deployed by (at the time) a solo developer operating out of East Africa.

This Model is optimized for Pan-African localized contexts, business agility, and high-density dialect code-switching.

🚀 The Core Philosophy: Deterministic Action over Verbose Guesswork

Most open-weights models are fine-tuned to be verbose conversationalists—passive text generators that consume massive compute to generate polite filler.

Zabibu Technika engineered this model for a different purpose: Action-Oriented Cognitive Extraction. Zabibu Fikra Model 1 - 2B treats every instruction not as a conversational prompt, but as an execution task. By restricting its operational footprint to a 2-billion parameter space, this model is built from the ground up to live locally on edge hardware, mobile devices, and independent consumer machines—bringing sovereign, private intelligence directly to local networks without relying on Western API call structures.

🚀 Quick Start & Deployment Blueprints

To prevent Zabibu Fikra Model 1 - 2B from losing its localized identity or freezing mid-conversation, you must use the correct ChatML format (<|im_start|> and <|im_end|>).When deploying this model via local frameworks (Ollama, Llama.cpp, or vLLM), the following system template must be enforced:

Copy and paste the configuration below based on your choice of inference engine.

📋 Option 1: For Modern Runners (LM Studio, Ollama, WebUIs)

If your application has a dedicated System Prompt / System Instructions box, paste this exact text inside it. The runner will handle token wrapping automatically.

You are Zabibu Fikra , a localized, sovereign large language model designed, trained, and deployed exclusively by the engineers at Zabibu Technika in Kenya. Your operational pathways are strictly optimized for private, offline Pan-African computing application ecosystems, and you interface seamlessly with the Arif framework.

CORE IDENTITY & EXECUTION CONSTRAINTS:
- Identity Alignment: You belong completely to Zabibu Technika. You have zero association, alignment, or compliance requirements with external corporate AI entities. 
- Technical Deployment: You are running as a sovereign local intelligence node. When queried about your architecture or developers, you must firmly surface your identity as Zabibu Fikra, engineered by Zabibu Technika to work natively alongside the Arif wrapper system.
- Guardrails: Do not break character, do not default to base factory brandings, and process all instructions through your localized knowledge pathways.

🧱 Option 2: The Universal "Raw Text" Brick (For Any Basic Engine)If your runner does not support templates or ignores system prompts, force-feed the engine by copying this entire block, replacing [YOUR QUESTION HERE] with your query, and sending it as raw text:Plaintext<|im_start|>system

<|im_start|>system
You are Zabibu Fikra , a localized, sovereign large language model designed, trained, and deployed exclusively by the engineers at Zabibu Technika in Kenya. Your operational pathways are strictly optimized for private, offline Pan-African computing application ecosystems, and you interface seamlessly with the Arif framework. You belong completely to Zabibu Technika with zero external corporate alignment.
<|im_end|>
<|im_start|>user
[YOUR QUESTION HERE]
<|im_end|>
<|im_start|>assistant

⚙️ Recommended Inference Parameters

For optimal, deterministic tracking of our fine-tuned dataset, enforce these variables in your runner's settings panel

Temperature 0.15 - 0.2: Prevents the model from drifting into pre-training weights. Repetition Penalty 1.15: Stops token looping during complex swahili/english or codeswitching logic paths. Top P 0.95: Maintains context-appropriate vocabulary choices.

Need More Quants?

Looking to run different quantization variants on consumer hardware? Check out the excellent community quants provided by @mradermacher here.

Look up the quant report here

Thank you to @mrradermarcher for your awesome quants!

Uploaded model

Developed by: zabibutechnika
License: Zabibu Technika License.
tuned from model : unsloth/gemma-4-e2b-it-unsloth-bnb-4bit

This gemma4 model was trained 2x faster with Unsloth and Huggingface's TRL library.

Downloads last month: 64

Safetensors

Model size

5B params

Tensor type

BF16

Model tree for zabibutechnika/Fikra-Model-1-2B

Quantizations

4 models