segformer-b0-finetuned-morphpadver1-hgo-coord-v1

This model is a fine-tuned version of nvidia/mit-b1 on the NICOPOI-9/morphpad_coord_hgo_512_4class dataset. It achieves the following results on the evaluation set:

• Loss: 0.0644
• Mean Iou: 0.9579
• Mean Accuracy: 0.9785
• Overall Accuracy: 0.9785
• Accuracy 0-0: 0.9792
• Accuracy 0-90: 0.9782
• Accuracy 90-0: 0.9762
• Accuracy 90-90: 0.9804
• Iou 0-0: 0.9634
• Iou 0-90: 0.9512
• Iou 90-0: 0.9543
• Iou 90-90: 0.9627

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: 40

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
1.1962	2.5445	4000	1.2063	0.2464	0.3994	0.4010	0.2991	0.2803	0.5096	0.5084	0.2367	0.2111	0.2658	0.2721
1.051	5.0891	8000	1.0734	0.3118	0.4765	0.4765	0.3856	0.5520	0.3937	0.5745	0.3123	0.3148	0.2968	0.3233
0.9309	7.6336	12000	0.9672	0.3612	0.5314	0.5323	0.4806	0.5216	0.7075	0.4158	0.3362	0.3706	0.3778	0.3603
0.8041	10.1781	16000	0.8444	0.4475	0.6180	0.6178	0.6131	0.6672	0.6120	0.5798	0.4403	0.4360	0.4543	0.4593
0.6617	12.7226	20000	0.7405	0.5039	0.6697	0.6700	0.6310	0.6588	0.6714	0.7177	0.5097	0.4912	0.5114	0.5033
0.54	15.2672	24000	0.6090	0.5828	0.7360	0.7362	0.6931	0.7532	0.7427	0.7550	0.5911	0.5709	0.5876	0.5819
0.7378	17.8117	28000	0.3740	0.7401	0.8507	0.8505	0.8789	0.8270	0.8186	0.8783	0.7712	0.7324	0.7203	0.7366
0.58	20.3562	32000	0.1892	0.8644	0.9272	0.9272	0.9329	0.9188	0.9142	0.9430	0.8810	0.8523	0.8539	0.8704
0.1305	22.9008	36000	0.1473	0.8945	0.9443	0.9443	0.9563	0.9245	0.9421	0.9542	0.9021	0.8783	0.8925	0.9049
0.1775	25.4453	40000	0.1133	0.9178	0.9571	0.9571	0.9578	0.9536	0.9583	0.9586	0.9264	0.9068	0.9130	0.9249
0.4792	27.9898	44000	0.0961	0.9306	0.9640	0.9640	0.9662	0.9633	0.9617	0.9650	0.9374	0.9194	0.9268	0.9387
0.1084	30.5344	48000	0.0886	0.9364	0.9671	0.9672	0.9684	0.9600	0.9689	0.9712	0.9429	0.9257	0.9335	0.9437
0.0471	33.0789	52000	0.0721	0.9485	0.9735	0.9735	0.9772	0.9674	0.9729	0.9767	0.9528	0.9402	0.9467	0.9542
0.0722	35.6234	56000	0.0646	0.9554	0.9772	0.9772	0.9809	0.9728	0.9757	0.9794	0.9576	0.9488	0.9522	0.9629
0.0406	38.1679	60000	0.0644	0.9579	0.9785	0.9785	0.9792	0.9782	0.9762	0.9804	0.9634	0.9512	0.9543	0.9627

Framework versions

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