segformer-b0-finetuned-segments-sidewalk-oct-22

This model is a fine-tuned version of nvidia/mit-b0 on the mrkprc1/SudokuSegmentation2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.3909
• Mean Iou: 0.3084
• Mean Accuracy: 0.9142
• Overall Accuracy: 0.9142
• Accuracy Empty-cell: nan
• Accuracy Complete-cell: 0.9142
• Iou Empty-cell: 0.0
• Iou Complete-cell: 0.6169

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: 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 Empty-cell	Accuracy Complete-cell	Iou Empty-cell	Iou Complete-cell
0.3784	10.0	20	0.4056	0.3166	0.9695	0.9695	nan	0.9695	0.0	0.6333
0.3037	20.0	40	0.4253	0.2816	0.7997	0.7997	nan	0.7997	0.0	0.5631
0.3326	30.0	60	0.4163	0.2894	0.8280	0.8280	nan	0.8280	0.0	0.5789
0.276	40.0	80	0.4037	0.2990	0.8758	0.8758	nan	0.8758	0.0	0.5979
0.2616	50.0	100	0.3909	0.3084	0.9142	0.9142	nan	0.9142	0.0	0.6169

Framework versions

• Transformers 4.37.1
• Pytorch 2.1.2
• Datasets 2.16.1
• Tokenizers 0.15.1