Instructions to use Mediform/gemma4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Mediform/gemma4 with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="Mediform/gemma4",
	filename="gemma4-e4b-v3-Q4_K_M.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 Mediform/gemma4 with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Mediform/gemma4:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf Mediform/gemma4:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Mediform/gemma4:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf Mediform/gemma4: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 Mediform/gemma4:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf Mediform/gemma4: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 Mediform/gemma4:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Mediform/gemma4:Q4_K_M

Use Docker

docker model run hf.co/Mediform/gemma4:Q4_K_M

LM Studio
Jan
Ollama
How to use Mediform/gemma4 with Ollama:
```
ollama run hf.co/Mediform/gemma4:Q4_K_M
```

Unsloth Studio

How to use Mediform/gemma4 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 Mediform/gemma4 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 Mediform/gemma4 to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Mediform/gemma4 to start chatting

How to use Mediform/gemma4 with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf Mediform/gemma4: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": "Mediform/gemma4:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use Mediform/gemma4 with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf Mediform/gemma4: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 Mediform/gemma4:Q4_K_M

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use Mediform/gemma4 with Docker Model Runner:
```
docker model run hf.co/Mediform/gemma4:Q4_K_M
```

Lemonade

How to use Mediform/gemma4 with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Mediform/gemma4:Q4_K_M

Run and chat with the model

lemonade run user.gemma4-Q4_K_M

List all available models

lemonade list

gemma-4-E4B-it — Scribion German Medical Extraction (v3) — GGUF Q4_K_M

On-device (llama.cpp / MLX) build of a LoRA fine-tune of google/gemma-4-E4B-it for German doctor–patient consultation fact extraction in Scribion's exact schema (5 per-type calls → {id,text,section,evidence} over the 24 NOTE_STATE section paths).

Files

gemma4-e4b-v3-Q4_K_M.gguf — language model, Q4_K_M (~4.5 GB).
mmproj-gemma-4-E4B-it-Q8_0.gguf — multimodal projector (audio+vision), unchanged from the base (ggml-org/gemma-4-E4B-it-GGUF); the LoRA only adapted the language tower, so the stock mmproj is correct. Needed for multimodal (audio/vision) inference; omit for text-only.

What it fixes (vs base / earlier v1)

Trained on teacher-distilled (gemma-4-31B-it-AWQ) Scribion-schema targets with per-type section validation + one-section-per-fact + correctly-empty calls. On the held-out Scribion eval (arztbericht, froehlich-krause), vs the base model it:

correctly routes diagnoses (e.g. "Weber-B-Fraktur" → plan.diagnosen_gesichert/findings.befunde, not findings.vitalparameter), imaging (CT → findings.befunde, not laboraufträge);
emits each fact in exactly ONE section (no cross-section duplication of prescriptions);
improves recall + keeps clean clinical one-fact-per-item phrasing (negation/laterality/numbers preserved).

Usage (llama.cpp)

# text-only extraction (per Scribion's per-type system prompt + "Transkript:\n"+transcript):
llama-cli -m gemma4-e4b-v3-Q4_K_M.gguf -sys "<scribion per-type system prompt>" -p "Transkript:\n..."
# multimodal (audio/vision): also pass --mmproj mmproj-gemma-4-E4B-it-Q8_0.gguf

Caveats

Q4_K_M is a 4-bit quant (small quality loss vs bf16). Synthetic-distilled training data; use a real held-out clinical eval as the acceptance gate. Minor known regression: may miss an already-administered med (e.g. Clexane given in the ER). See the training repo for full eval.

Reconverted with llama.cpp @ 039e20a2d (2026-06-10), CPU build.

Downloads last month: 110

GGUF

Model size

7B params

Architecture

gemma4

Hardware compatibility

4-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for Mediform/gemma4

Base model

google/gemma-4-E4B

Finetuned

google/gemma-4-E4B-it

Quantized

(231)

this model