AttributeError: 'Config' object has no attribute '_get_non_default_generation_parameters'

[XYZ-916]

System Info

transformers version: 4.47.1
I finetuned the rmbg-2.0 with my own dataset. When I save the model using model.save_pretrained, I got the error as follows:
File "/usr/lib/python3.10/site-packages/transformers/modeling_utils.py", line 2815, in save_pretrained
misplaced_generation_parameters = model_to_save.config._get_non_default_generation_parameters()
AttributeError: 'Config' object has no attribute '_get_non_default_generation_parameters'

Who can help?

No response

Information

• The official example scripts
• My own modified scripts

Tasks

• An officially supported task in the examples folder (such as GLUE/SQuAD, ...)
• My own task or dataset (give details below)

Reproduction

1. finetune rmbg-2;
2. save model.

Expected behavior

save the model sucessfully

[Rocketknight1]

Hi @XYZ-916, can you create some simple reproducer code we can run to diagnose this? For example, just load and save the model or even just the config and check if the bug still occurs

[cycycc]

Hi @Rocketknight1, I encountered the same issue when trying to save a pretrained model.

Here is an example to reproduce the problem:

• transformers==4.48.1
• torch==2.2.0+cu121
• CUDA 12.2

from transformers import AutoModelForImageSegmentation

model = AutoModelForImageSegmentation.from_pretrained(
    'ZhengPeng7/BiRefNet',
    trust_remote_code=True,
)

// This line raised AttributeError: 'Config' object has no attribute '_get_non_default_generation_parameters'
model.save_pretrained("./BiRefNet/")