Instructions to use qikp/gemma-3-110m-english-only-old with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use qikp/gemma-3-110m-english-only-old with Transformers:

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

pipe = pipeline("text-generation", model="qikp/gemma-3-110m-english-only-old")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("qikp/gemma-3-110m-english-only-old")
model = AutoModelForCausalLM.from_pretrained("qikp/gemma-3-110m-english-only-old", device_map="auto")
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 qikp/gemma-3-110m-english-only-old with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "qikp/gemma-3-110m-english-only-old"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "qikp/gemma-3-110m-english-only-old",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/qikp/gemma-3-110m-english-only-old

SGLang

How to use qikp/gemma-3-110m-english-only-old 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 "qikp/gemma-3-110m-english-only-old" \
    --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": "qikp/gemma-3-110m-english-only-old",
		"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 "qikp/gemma-3-110m-english-only-old" \
        --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": "qikp/gemma-3-110m-english-only-old",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use qikp/gemma-3-110m-english-only-old with Docker Model Runner:
```
docker model run hf.co/qikp/gemma-3-110m-english-only-old
```

Gemma 3 110M English Only

This is Gemma 3 270M with the pika 3 tokenizer used and with retrained embedding and output layers, afterwards a gentle retrain.

Chat template

Note: This uses the Zephyr chat template.

Training

It was trained on these using 1 epoch, sparse adam optimizer, and 32 batch size:

HuggingFaceTB/cosmopedia-20k
HuggingFaceTB/smol-smoltalk (first 23,000 rows)
HuggingFaceTB/everyday-conversations-llama3.1-2k

Limitations

The quality of this model needs to be better tested. It will also poorly handle non-English languages.

Downloads last month: 22

Safetensors

Model size

0.1B params

Tensor type

F32

Model tree for qikp/gemma-3-110m-english-only-old

Base model

google/gemma-3-270m

Finetuned

google/gemma-3-270m-it

Finetuned

unsloth/gemma-3-270m-it

Finetuned

(423)

this model

qikp
/

gemma-3-110m-english-only-old

Gemma 3 110M English Only

Chat template

Training

Limitations

Model tree for qikp/gemma-3-110m-english-only-old

Datasets used to train qikp/gemma-3-110m-english-only-old