Instructions to use innerloop-dev/gemma3-4b-text with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use innerloop-dev/gemma3-4b-text 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 innerloop-dev/gemma3-4b-text:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf innerloop-dev/gemma3-4b-text:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf innerloop-dev/gemma3-4b-text:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf innerloop-dev/gemma3-4b-text: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 innerloop-dev/gemma3-4b-text:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf innerloop-dev/gemma3-4b-text: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 innerloop-dev/gemma3-4b-text:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf innerloop-dev/gemma3-4b-text:Q4_K_M

Use Docker

docker model run hf.co/innerloop-dev/gemma3-4b-text:Q4_K_M

LM Studio
Jan

vLLM

How to use innerloop-dev/gemma3-4b-text with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "innerloop-dev/gemma3-4b-text"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "innerloop-dev/gemma3-4b-text",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/innerloop-dev/gemma3-4b-text:Q4_K_M

Ollama
How to use innerloop-dev/gemma3-4b-text with Ollama:
```
ollama run hf.co/innerloop-dev/gemma3-4b-text:Q4_K_M
```
Unsloth Desktop
Docker Model Runner
How to use innerloop-dev/gemma3-4b-text with Docker Model Runner:
```
docker model run hf.co/innerloop-dev/gemma3-4b-text:Q4_K_M
```

Lemonade

How to use innerloop-dev/gemma3-4b-text with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull innerloop-dev/gemma3-4b-text:Q4_K_M

Run and chat with the model

lemonade run user.gemma3-4b-text-Q4_K_M

List all available models

lemonade list

Atomic Chat

gemma3-4b-text / NOTICE

innerloop-dev

Upload NOTICE with huggingface_hub

7483964 verified 15 days ago

Raw History Blame Contribute Delete

1.24 kB

	NOTICE

	Gemma is provided under and subject to the Gemma Terms of Use found at ai.google.dev/gemma/terms

	This distribution contains a modified Gemma file. See MODIFICATIONS.
	Everything in this distribution that is not Gemma is © Innerloop, innerloop.works


	MODIFICATIONS

	This file is a modified Gemma file.

	Source: the model layer of Ollama's library build of google/gemma-3-4b-it,
	Q4_K_M, sha256 aeda25e63ebd698fab8638ffb778e68bed908b960d39d0becc650fa981609d25,
	3,338,792,448 bytes.

	Modification: the SigLIP vision tower and the multimodal projection were removed —
	439 tensors (v., mm.) and 9 metadata keys (gemma3.vision., gemma3.mm.).
	The 444 text tensors are byte-for-byte the originals: same name, same ggml
	quantization type, same shape, same bytes. Nothing was requantized, converted
	or retrained. The tokenizer is unchanged.

	Result: 2,498,332,864 bytes, sha256
	199388f8f8cbec06b80bd63b0b1a774103e00c993c107bf1d480e58047420532.
	The file cannot read images. It writes text.

	Modified by Innerloop (innerloop.works) on 2026-09-19 with strip-projector.py.

	Gemma is provided under and subject to the Gemma Terms of Use found at
	ai.google.dev/gemma/terms.