README.md

metadata

library_name: transformers
license: other
base_model: nvidia/mit-b0
tags:
  - vision
  - image-segmentation
  - generated_from_trainer
model-index:
  - name: custom-object-test6
    results: []

custom-object-test6

This model is a fine-tuned version of nvidia/mit-b0 on the sungile/custom-object-masking5 dataset. It achieves the following results on the evaluation set:

• Loss: 0.3457
• Mean Iou: 0.3368
• Mean Accuracy: 0.6736
• Overall Accuracy: 0.6736
• Accuracy Unknown: nan
• Accuracy Background: 0.6736
• Accuracy Object: nan
• Iou Unknown: 0.0
• Iou Background: 0.6736
• Iou Object: nan

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: 6e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Unknown	Accuracy Background	Accuracy Object	Iou Unknown	Iou Background	Iou Object
0.7667	0.25	20	1.0272	0.3026	0.9079	0.9079	nan	0.9079	nan	0.0	0.9079	0.0
0.8127	0.5	40	0.8213	0.2803	0.8409	0.8409	nan	0.8409	nan	0.0	0.8409	0.0
0.5588	0.75	60	0.7310	0.4304	0.8608	0.8608	nan	0.8608	nan	0.0	0.8608	nan
0.6156	1.0	80	0.5317	0.3130	0.6261	0.6261	nan	0.6261	nan	0.0	0.6261	nan
0.5077	1.25	100	0.4617	0.3482	0.6964	0.6964	nan	0.6964	nan	0.0	0.6964	nan
0.5612	1.5	120	0.4336	0.3341	0.6683	0.6683	nan	0.6683	nan	0.0	0.6683	nan
0.4468	1.75	140	0.3946	0.3442	0.6883	0.6883	nan	0.6883	nan	0.0	0.6883	nan
0.292	2.0	160	0.3554	0.3041	0.6081	0.6081	nan	0.6081	nan	0.0	0.6081	nan
0.3769	2.25	180	0.3798	0.3402	0.6805	0.6805	nan	0.6805	nan	0.0	0.6805	nan
0.3386	2.5	200	0.3493	0.3147	0.6293	0.6293	nan	0.6293	nan	0.0	0.6293	nan
0.2689	2.75	220	0.3736	0.3492	0.6984	0.6984	nan	0.6984	nan	0.0	0.6984	nan
0.2539	3.0	240	0.3457	0.3368	0.6736	0.6736	nan	0.6736	nan	0.0	0.6736	nan

Framework versions

• Transformers 4.47.1
• Pytorch 2.1.0+cu118
• Datasets 3.2.0
• Tokenizers 0.21.0