Instructions to use unsloth/functiongemma-270m-it with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use unsloth/functiongemma-270m-it with Transformers:

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

pipe = pipeline("text-generation", model="unsloth/functiongemma-270m-it")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("unsloth/functiongemma-270m-it")
model = AutoModelForCausalLM.from_pretrained("unsloth/functiongemma-270m-it")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use unsloth/functiongemma-270m-it with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "unsloth/functiongemma-270m-it"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "unsloth/functiongemma-270m-it",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/unsloth/functiongemma-270m-it

SGLang

How to use unsloth/functiongemma-270m-it 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 "unsloth/functiongemma-270m-it" \
    --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": "unsloth/functiongemma-270m-it",
		"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 "unsloth/functiongemma-270m-it" \
        --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": "unsloth/functiongemma-270m-it",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use unsloth/functiongemma-270m-it with Docker Model Runner:
```
docker model run hf.co/unsloth/functiongemma-270m-it
```

Daemontatox commited on Dec 19, 2025

Commit

5c044ab

verified ·

1 Parent(s): 78d4fe7

Update config.json

Browse files

https://huggingface.co/google/functiongemma-270m-it/discussions/5#694457b5ad6f76a0d3c43917

"""
vLLM has started using the Transformers v5 style rope_parameters and patches it into configs loaded with Transformers v4.

vLLM 0.12.0 uses the existence of rope_parameters to decide whether or not to set a default value for it, i.e.:

if rope_theta is not None:
if not hasattr(config, "rope_parameters"):
config.rope_parameters = {"rope_type": "default"}
config.rope_parameters["rope_theta"] = rope_theta

This causes an error with this model because rope_parameters will remain None and the following dict assignment will fail.

https://github.com/vllm-project/vllm/pull/30983 fixes this on the vLLM side for (hopefully) v0.13.0 onwards.

Removing these fields from the checkpoint should fix this model for vLLM v0.12.0 in the meantime.
"""

Files changed (1) hide show

config.json +0 -2

config.json CHANGED Viewed

@@ -44,8 +44,6 @@
   "query_pre_attn_scalar": 256,
   "rms_norm_eps": 1e-06,
   "rope_local_base_freq": 10000.0,
-  "rope_parameters": null,
-  "rope_scaling": null,
   "rope_theta": 1000000.0,
   "sliding_window": 512,
   "transformers_version": "4.57.3",

   "query_pre_attn_scalar": 256,
   "rms_norm_eps": 1e-06,
   "rope_local_base_freq": 10000.0,
   "rope_theta": 1000000.0,
   "sliding_window": 512,
   "transformers_version": "4.57.3",