google

/

t5gemma-2-4b-4b

like 131

Follow

Google 40.9k

I have cloned the repo and try to run

´´´python
import requests
from PIL import Image
from transformers import AutoProcessor, AutoModelForSeq2SeqLM

processor = AutoProcessor.from_pretrained("t5gemma-2-4b-4b")
model = AutoModelForSeq2SeqLM.from_pretrained("t5gemma-2-4b-4b")

url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg"
image = Image.open(requests.get(url, stream=True).raw)
prompt = " in this image, there is"

model_inputs = processor(text=prompt, images=image, return_tensors="pt")
generation = model.generate(**model_inputs, max_new_tokens=20, do_sample=False)
print(processor.decode(generation[0]))
´´´

and have python 3.12 with

• torch 2.9.1
• torchvision 0.24.1
• transformers 4.57.3
• pillow 12.0.0

But I get

´´´bash
python .\example.py
Using a slow image processor as use_fast is unset and a slow processor was saved with this model. use_fast=True will be the default behavior in v4.52, even if the model was saved with a slow processor. This will result in minor differences in outputs. You'll still be able to use a slow processor with use_fast=False.
The tokenizer you are loading from 't5gemma-2-4b-4b' with an incorrect regex pattern: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503/discussions/84#69121093e8b480e709447d5e. This will lead to incorrect tokenization. You should set the fix_mistral_regex=True flag when loading this tokenizer to fix this issue.
Traceback (most recent call last):
File "C:\Users\test\gemma4t\example.py", line 5, in
processor = AutoProcessor.from_pretrained("t5gemma-2-4b-4b")
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\test\gemma\venv\Lib\site-packages\transformers\models\auto\processing_auto.py", line 396, in from_pretrained
return processor_class.from_pretrained(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\test\gemma\venv\Lib\site-packages\transformers\processing_utils.py", line 1396, in from_pretrained
return cls.from_args_and_dict(args, processor_dict, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\test\gemma\venv\Lib\site-packages\transformers\processing_utils.py", line 1197, in from_args_and_dict
processor = cls(*args, **valid_kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\test\gemma\venv\Lib\site-packages\transformers\models\gemma3\processing_gemma3.py", line 67, in init
self.image_token_id = tokenizer.image_token_id
^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\test\gemma\venv\Lib\site-packages\transformers\tokenization_utils_base.py", line 1128, in getattr
raise AttributeError(f"{self.class.name} has no attribute {key}")
AttributeError: GemmaTokenizerFast has no attribute image_token_id
´´´

What are the versions and settings to run?