neulab

/

Pangea-7B-hf

like 13

Follow

NeuLab @ LTI/CMU 113

During the HF inference procedure, due to the patch size being null in processor_config.json, I'm getting this error:
Traceback (most recent call last):
...
model_inputs = processor(
File "/home/user/miniconda3/envs/pangea/lib/python3.10/site-packages/transformers/models/llava_next/processing_llava_next.py", line 166, in call
num_image_tokens = self._get_number_of_features(orig_height, orig_width, height, width)
File "/home/user/miniconda3/envs/pangea/lib/python3.10/site-packages/transformers/models/llava_next/processing_llava_next.py", line 194, in _get_number_of_features
patches_height = height // self.patch_size
TypeError: unsupported operand type(s) for //: 'int' and 'NoneType'

The solution to this is to set patch_size to 14. I've made the change. Kindly accept the PR.