Instructions to use RagnarokReinier/schakel-gemma-router with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use RagnarokReinier/schakel-gemma-router with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("google/gemma-4-E2B-it")
model = PeftModel.from_pretrained(base_model, "RagnarokReinier/schakel-gemma-router")

Transformers

How to use RagnarokReinier/schakel-gemma-router with Transformers:

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

pipe = pipeline("text-generation", model="RagnarokReinier/schakel-gemma-router")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("RagnarokReinier/schakel-gemma-router", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use RagnarokReinier/schakel-gemma-router with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "RagnarokReinier/schakel-gemma-router"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "RagnarokReinier/schakel-gemma-router",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/RagnarokReinier/schakel-gemma-router

SGLang

How to use RagnarokReinier/schakel-gemma-router 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 "RagnarokReinier/schakel-gemma-router" \
    --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": "RagnarokReinier/schakel-gemma-router",
		"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 "RagnarokReinier/schakel-gemma-router" \
        --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": "RagnarokReinier/schakel-gemma-router",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use RagnarokReinier/schakel-gemma-router with Docker Model Runner:
```
docker model run hf.co/RagnarokReinier/schakel-gemma-router
```

Schakel Intent Router

A LoRA adapter fine-tuned on google/gemma-4-E2B-it for Spanish-language intent classification.

Classifies user input into one of three intents and responds with strict JSON:

{"intent": "DOMOTICA"}

Intents

Intent	Description
`DOMOTICA`	Home automation commands (lights, temperature, appliances)
`MUSICA`	Music playback requests
`GENERAL`	Everything else (time, jokes, general questions)

Usage

import torch
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "google/gemma-4-E2B-it"
adapter_path = "RagnarokReinier/schakel-gemma-router"

tokenizer = AutoTokenizer.from_pretrained(adapter_path)
base_model = AutoModelForCausalLM.from_pretrained(model_id, dtype=torch.bfloat16, device_map="auto")
model = PeftModel.from_pretrained(base_model, adapter_path)
model.eval()

messages = [
    {"role": "system", "content": "Clasifica la intención del usuario. Responde solo JSON válido con una clave intent y un valor entre DOMOTICA, MUSICA o GENERAL."},
    {"role": "user", "content": "enciende la luz del salón"},
]

inputs = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_tensors="pt", return_dict=True).to(model.device)

with torch.no_grad():
    output_ids = model.generate(**inputs, max_new_tokens=64, do_sample=False)

new_tokens = output_ids[0, inputs["input_ids"].shape[1]:]
print(tokenizer.decode(new_tokens, skip_special_tokens=True))
# {"intent":"DOMOTICA"}

Training Details

Base model: google/gemma-4-E2B-it
Method: LoRA (rank 16, alpha 32) via SFTTrainer
Precision: bf16
Target modules: language model attention + MLP projections (q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj)
Dataset: 240 train / 60 validation examples (Spanish voice assistant text)
Epochs: 3
Learning rate: 2e-4
Max sequence length: 256
Batch size: 4 (gradient accumulation: 4)

Results

Metric	Value
Train loss	4.07
Eval loss	0.48
Token accuracy	~90%
Training time	~3.5 min (Apple Silicon, MPS)

Framework versions

PEFT 0.18.1
TRL 1.1.0
Transformers 5.5.4
PyTorch 2.11.0

Downloads last month: 1

Model tree for RagnarokReinier/schakel-gemma-router

Base model

google/gemma-4-E2B

Finetuned

google/gemma-4-E2B-it

Adapter

(80)

this model