README.md

metadata

license: other
base_model: nvidia/mit-b0
tags:
  - generated_from_trainer
model-index:
  - name: mit-b0-building-damage-lora
    results: []

mit-b0-building-damage-lora

This model is a fine-tuned version of nvidia/mit-b0 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0719
• Mean Iou: 0.3422
• Mean Accuracy: 0.6845
• Overall Accuracy: 0.6845
• Accuracy Background: nan
• Accuracy Building: 0.6845
• Iou Background: 0.0
• Iou Building: 0.6845

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0005
• train_batch_size: 8
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Background	Accuracy Building	Iou Background	Iou Building
0.0816	1.0	700	0.0954	0.3942	0.7884	0.7884	nan	0.7884	0.0	0.7884
0.0969	2.0	1400	0.0771	0.3662	0.7323	0.7323	nan	0.7323	0.0	0.7323
0.0813	3.0	2100	0.0735	0.3608	0.7216	0.7216	nan	0.7216	0.0	0.7216
0.0847	4.0	2800	0.0732	0.3557	0.7114	0.7114	nan	0.7114	0.0	0.7114
0.0657	5.0	3500	0.0705	0.3352	0.6703	0.6703	nan	0.6703	0.0	0.6703
0.0739	6.0	4200	0.0744	0.3606	0.7211	0.7211	nan	0.7211	0.0	0.7211
0.0642	7.0	4900	0.0737	0.3754	0.7508	0.7508	nan	0.7508	0.0	0.7508
0.0594	8.0	5600	0.0710	0.3128	0.6256	0.6256	nan	0.6256	0.0	0.6256
0.0694	9.0	6300	0.0702	0.3431	0.6863	0.6863	nan	0.6863	0.0	0.6863
0.0658	10.0	7000	0.0730	0.3666	0.7332	0.7332	nan	0.7332	0.0	0.7332
0.0849	11.0	7700	0.0827	0.4007	0.8013	0.8013	nan	0.8013	0.0	0.8013
0.0607	12.0	8400	0.0891	0.3844	0.7689	0.7689	nan	0.7689	0.0	0.7689
0.073	13.0	9100	0.1030	0.4268	0.8536	0.8536	nan	0.8536	0.0	0.8536

Framework versions

• Transformers 4.33.0
• Pytorch 2.0.0
• Datasets 2.1.0
• Tokenizers 0.13.3