AsefaH/

architectural-spatial-blindspots-smolvlm

Model Tested: HuggingFaceTB/SmolVLM-BaseParameter Count: 2.2B

I loaded the model in a Google Colab environment using the transformers library with 4-bit quantization to fit within a free-tier GPU. from transformers import AutoProcessor, AutoModelForVision2Seq import torch

model_id = "HuggingFaceTB/SmolVLM-Base" processor = AutoProcessor.from_pretrained(model_id) model =… See the full description on the dataset page: https://huggingface.co/datasets/AsefaH/architectural-spatial-blindspots-smolvlm.