hanslab37/architectural_styles_classifier

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

architecturalstylesclassifier

This model is a fine-tuned version of nvidia/mit-b0 on the Architectural styles dataset, retrieved from https://www.kaggle.com/datasets/dumitrux/architectural-styles-dataset. It achieves the following results on the evaluation set:

• —Loss: 0.9412
• —Accuracy: 0.7223

Model description

Presentation about the model: https://www.canva.com/design/DAGLBMAs1K4/d8qvLN2nchSYVmnrwYzx0w/edit?utmcontent=DAGLBMAs1K4&utmcampaign=designshare&utmmedium=link2&utmsource=sharebutton

You can try the model from Huggingface Space this link: https://huggingface.co/spaces/hanslab37/technospire

Intended uses & limitations

The model were developed as part of experiment to learn about training a model and developing Image Classification model with Gradio in Huggingface. You can use it for experiment only. Not recommended for daily use.

Training and evaluation data

https://www.kaggle.com/datasets/dumitrux/architectural-styles-dataset

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• —learning_rate: 0.0005
• —trainbatchsize: 4
• —evalbatchsize: 4
• —seed: 42
• —gradientaccumulationsteps: 16
• —totaltrainbatch_size: 64
• —optimizer: Adam with betas=(0.9,0.999) and epsilon=0.0003
• —lrschedulertype: linear
• —lrschedulerwarmup_ratio: 0.1
• —num_epochs: 20

Training results

Framework versions

• —Transformers 4.41.1
• —Pytorch 2.3.0
• —Datasets 2.19.1
• —Tokenizers 0.19.1