mlx-vlm finetuning Qwen 3.5 2b
#2
by dcrey7 - opened
i have been trying to finetune this model using the odcumentation in https://github.com/Blaizzy/mlx-vlm/blob/main/mlx_vlm/LORA.MD, i have a mac m4 pro with 48 gb unified memory .
mlx-lm==0.30.7
mlx-metal==0.31.0
mlx-vlm==0.3.12
torch==2.10.0
torchvision==0.25.0
transformers==5.2.0
datasets==4.6.1
huggingface-hub==1.5.0
pillow=12.1.2
tqdm==4.67.3
my mac crashes while finetuning on images with 768x 768 ? can you please let me know if im messing up somewhere or if im doing anything wrong