Instructions to use google/t5gemma-b-b-ul2-it with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use google/t5gemma-b-b-ul2-it with Transformers:

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

pipe = pipeline("text-generation", model="google/t5gemma-b-b-ul2-it")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("google/t5gemma-b-b-ul2-it")
model = AutoModelForSeq2SeqLM.from_pretrained("google/t5gemma-b-b-ul2-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

vLLM

How to use google/t5gemma-b-b-ul2-it with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "google/t5gemma-b-b-ul2-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": "google/t5gemma-b-b-ul2-it",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/google/t5gemma-b-b-ul2-it

SGLang

How to use google/t5gemma-b-b-ul2-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 "google/t5gemma-b-b-ul2-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": "google/t5gemma-b-b-ul2-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 "google/t5gemma-b-b-ul2-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": "google/t5gemma-b-b-ul2-it",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use google/t5gemma-b-b-ul2-it with Docker Model Runner:
```
docker model run hf.co/google/t5gemma-b-b-ul2-it
```

Resize embeddings layer

by Thalesian - opened Dec 3, 2025

Discussion

Thalesian

Dec 3, 2025

I really want to adapt the T5Gemma family to a project I am working on to translate cuneiform tablets - but to do it I have to resize the embeddings layer to introduce token scripts. While this is easy to do for the traditional T5 family of models, I don't see how to do it here. Any guidance would be much appreciated. I can technically do it - but I just end up re-initiating a bunch of layers that erase the models ability to function.

srikanta-221

Google org Jan 29

Hi @Thalesian , apologies for the delayed response!

If you still facing this issue, it's mostly because this particular model uses untied embeddings.
You can try manually resize and copy weights for each matrices.
If not already, you can try to use the newer T5Gemma 2.

Thank you!

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment