segformer-b0-finetuned-morphpadver1-no_ckp-3

This model is a fine-tuned version of nvidia/mit-b0 on the NICOPOI-9/morphpad_512_4class dataset. It achieves the following results on the evaluation set:

• Loss: 0.0023
• Mean Iou: 0.9987
• Mean Accuracy: 0.9993
• Overall Accuracy: 0.9994
• Accuracy 0-0: 0.9991
• Accuracy 0-90: 0.9992
• Accuracy 90-0: 0.9993
• Accuracy 90-90: 0.9998
• Iou 0-0: 0.9990
• Iou 0-90: 0.9984
• Iou 90-0: 0.9984
• Iou 90-90: 0.9990

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: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Use OptimizerNames.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: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy 0-0	Accuracy 0-90	Accuracy 90-0	Accuracy 90-90	Iou 0-0	Iou 0-90	Iou 90-0	Iou 90-90
0.0554	0.4449	4000	0.0432	0.9709	0.9852	0.9854	0.9926	0.9779	0.9811	0.9893	0.9855	0.9560	0.9565	0.9854
0.0175	0.8898	8000	0.0254	0.9860	0.9929	0.9930	0.9962	0.9889	0.9897	0.9970	0.9934	0.9785	0.9776	0.9944
0.0213	1.3347	12000	0.0190	0.9921	0.9961	0.9961	0.9977	0.9925	0.9962	0.9979	0.9962	0.9879	0.9879	0.9965
0.0092	1.7796	16000	0.0148	0.9939	0.9970	0.9970	0.9986	0.9941	0.9972	0.9980	0.9973	0.9905	0.9904	0.9974
0.0055	2.2244	20000	0.0054	0.9970	0.9985	0.9985	0.9989	0.9982	0.9976	0.9991	0.9982	0.9957	0.9957	0.9982
0.004	2.6693	24000	0.0040	0.9977	0.9988	0.9988	0.9995	0.9983	0.9984	0.9991	0.9984	0.9970	0.9968	0.9985
0.0029	3.1142	28000	0.0078	0.9973	0.9987	0.9987	0.9993	0.9968	0.9990	0.9995	0.9989	0.9958	0.9958	0.9988
0.0125	3.5591	32000	0.0104	0.9970	0.9985	0.9985	0.9996	0.9958	0.9989	0.9996	0.9990	0.9949	0.9950	0.9989
0.0032	4.0040	36000	0.0030	0.9982	0.9991	0.9991	0.9996	0.9979	0.9992	0.9997	0.9990	0.9974	0.9975	0.9989
0.0051	4.4489	40000	0.0028	0.9981	0.9991	0.9991	0.9997	0.9974	0.9995	0.9997	0.9991	0.9971	0.9973	0.9990
0.0295	4.8938	44000	0.0023	0.9987	0.9993	0.9994	0.9991	0.9992	0.9993	0.9998	0.9990	0.9984	0.9984	0.9990

Framework versions

• Transformers 4.48.3
• Pytorch 2.1.0
• Datasets 3.2.0
• Tokenizers 0.21.0