segformer-finetuned-biofilm_MRCNNv1_validation

This model is a fine-tuned version of nvidia/mit-b0 on the heroza/biofilm_MRCNNv1_validation dataset. It achieves the following results on the evaluation set:

• Loss: 0.0667
• Mean Iou: 0.4894
• Mean Accuracy: 0.9788
• Overall Accuracy: 0.9788
• Accuracy Background: 0.9788
• Accuracy Biofilm: nan
• Iou Background: 0.9788
• Iou Biofilm: 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: 8
• eval_batch_size: 8
• seed: 1337
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: polynomial
• training_steps: 10000

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Background	Accuracy Biofilm	Iou Background	Iou Biofilm
0.0896	1.0	351	0.0405	0.4947	0.9894	0.9894	0.9894	nan	0.9894	0.0
0.0556	2.0	702	0.0459	0.4925	0.9849	0.9849	0.9849	nan	0.9849	0.0
0.0532	3.0	1053	0.0352	0.4931	0.9863	0.9863	0.9863	nan	0.9863	0.0
0.0473	4.0	1404	0.0318	0.4936	0.9872	0.9872	0.9872	nan	0.9872	0.0
0.0387	5.0	1755	0.0318	0.4928	0.9857	0.9857	0.9857	nan	0.9857	0.0
0.0388	6.0	2106	0.0394	0.4909	0.9817	0.9817	0.9817	nan	0.9817	0.0
0.0344	7.0	2457	0.0431	0.4906	0.9811	0.9811	0.9811	nan	0.9811	0.0
0.0409	8.0	2808	0.0347	0.4922	0.9844	0.9844	0.9844	nan	0.9844	0.0
0.0322	9.0	3159	0.0415	0.4910	0.9819	0.9819	0.9819	nan	0.9819	0.0
0.0331	10.0	3510	0.0558	0.4884	0.9767	0.9767	0.9767	nan	0.9767	0.0
0.0337	11.0	3861	0.0422	0.4923	0.9847	0.9847	0.9847	nan	0.9847	0.0
0.0357	12.0	4212	0.0421	0.4908	0.9816	0.9816	0.9816	nan	0.9816	0.0
0.0306	13.0	4563	0.0398	0.4913	0.9827	0.9827	0.9827	nan	0.9827	0.0
0.0324	14.0	4914	0.0488	0.4905	0.9810	0.9810	0.9810	nan	0.9810	0.0
0.0293	15.0	5265	0.0401	0.4918	0.9835	0.9835	0.9835	nan	0.9835	0.0
0.0243	16.0	5616	0.0499	0.4894	0.9788	0.9788	0.9788	nan	0.9788	0.0
0.0306	17.0	5967	0.0495	0.4902	0.9805	0.9805	0.9805	nan	0.9805	0.0
0.0267	18.0	6318	0.0498	0.4907	0.9813	0.9813	0.9813	nan	0.9813	0.0
0.0295	19.0	6669	0.0566	0.4903	0.9806	0.9806	0.9806	nan	0.9806	0.0
0.0263	20.0	7020	0.0658	0.4893	0.9786	0.9786	0.9786	nan	0.9786	0.0
0.0319	21.0	7371	0.0646	0.4885	0.9770	0.9770	0.9770	nan	0.9770	0.0
0.0236	22.0	7722	0.0608	0.4897	0.9793	0.9793	0.9793	nan	0.9793	0.0
0.0249	23.0	8073	0.0578	0.4897	0.9795	0.9795	0.9795	nan	0.9795	0.0
0.0242	24.0	8424	0.0558	0.4902	0.9804	0.9804	0.9804	nan	0.9804	0.0
0.0264	25.0	8775	0.0579	0.4899	0.9798	0.9798	0.9798	nan	0.9798	0.0
0.0235	26.0	9126	0.0582	0.4900	0.9801	0.9801	0.9801	nan	0.9801	0.0
0.0235	27.0	9477	0.0609	0.4897	0.9794	0.9794	0.9794	nan	0.9794	0.0
0.0204	28.0	9828	0.0648	0.4896	0.9791	0.9791	0.9791	nan	0.9791	0.0
0.023	28.49	10000	0.0667	0.4894	0.9788	0.9788	0.9788	nan	0.9788	0.0

Framework versions

• Transformers 4.38.0.dev0
• Pytorch 2.0.0+cu117
• Datasets 2.14.4
• Tokenizers 0.15.1