Hugging Face's logo

deepseek-ai
/
deepseek-vl-1.3b-base

Transformers
Safetensors
multi_modality
Model card Files
xet
 Community
4

Finetuning Vision Encoder

#2
by SinanAkkoyun - opened
Discussion
SinanAkkoyun

Hi, thank you very much for the model. I need to finetune the vision encoder, how can I do that?
marcusinthesky
 
for name, parameter in model.named_parameters():
    if 'vision' in name:
        parameter.requires_grad = True
    else:
        parameter.requires_grad = False

Sign up or log in to comment