segformer-b0-mars-testbed-4class

This model is a fine-tuned version of nvidia/mit-b0 on the Tani04/mars_testbed_terrain dataset. It achieves the following results on the evaluation set:

• Loss: 0.6859
• Mean Iou: 0.2495
• Mean Accuracy: 0.9980
• Overall Accuracy: 0.9980
• Accuracy Flat Bedrock: 0.9980
• Accuracy Flat Gravel: nan
• Accuracy Hard Gravel: nan
• Accuracy Obstacle: nan
• Iou Flat Bedrock: 0.9980
• Iou Flat Gravel: 0.0
• Iou Hard Gravel: 0.0
• Iou Obstacle: 0.0

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: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Flat Bedrock	Accuracy Flat Gravel	Accuracy Hard Gravel	Accuracy Obstacle	Iou Flat Bedrock	Iou Flat Gravel	Iou Hard Gravel	Iou Obstacle
1.2007	2.5	20	1.2885	0.2131	0.8524	0.8524	0.8524	nan	nan	nan	0.8524	0.0	0.0	0.0
1.0165	5.0	40	0.9504	0.2466	0.9863	0.9863	0.9863	nan	nan	nan	0.9863	0.0	0.0	0.0
0.8832	7.5	60	0.7721	0.2473	0.9890	0.9890	0.9890	nan	nan	nan	0.9890	0.0	0.0	0.0
0.7765	10.0	80	0.7240	0.2488	0.9950	0.9950	0.9950	nan	nan	nan	0.9950	0.0	0.0	0.0
0.7693	12.5	100	0.6920	0.2489	0.9957	0.9957	0.9957	nan	nan	nan	0.9957	0.0	0.0	0.0
0.7011	15.0	120	0.6859	0.2495	0.9980	0.9980	0.9980	nan	nan	nan	0.9980	0.0	0.0	0.0

Framework versions

• Transformers 4.46.3
• Pytorch 2.4.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3