README.md

metadata

license: other
base_model: nvidia/mit-b0
tags:
  - image-segmentation
  - vision
  - generated_from_trainer
model-index:
  - name: segformer-finetuned-biofilm2
    results: []

segformer-finetuned-biofilm2

This model is a fine-tuned version of nvidia/mit-b0 on the heroza/biofilm2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0797
• Mean Iou: 0.4786
• Mean Accuracy: 0.9572
• Overall Accuracy: 0.9572
• Accuracy Background: nan
• Accuracy Biofilm: 0.9572
• Iou Background: 0.0
• Iou Biofilm: 0.9572

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
0.1622	1.0	280	0.1158	0.4714	0.9428	0.9428	nan	0.9428	0.0	0.9428
0.0742	2.0	560	0.0643	0.4545	0.9090	0.9090	nan	0.9090	0.0	0.9090
0.0549	3.0	840	0.0582	0.4797	0.9594	0.9594	nan	0.9594	0.0	0.9594
0.0459	4.0	1120	0.0508	0.4737	0.9475	0.9475	nan	0.9475	0.0	0.9475
0.0506	5.0	1400	0.0405	0.4705	0.9411	0.9411	nan	0.9411	0.0	0.9411
0.0411	6.0	1680	0.0476	0.4865	0.9729	0.9729	nan	0.9729	0.0	0.9729
0.0456	7.0	1960	0.0476	0.4754	0.9509	0.9509	nan	0.9509	0.0	0.9509
0.0381	8.0	2240	0.0554	0.4792	0.9584	0.9584	nan	0.9584	0.0	0.9584
0.0348	9.0	2520	0.0559	0.4889	0.9779	0.9779	nan	0.9779	0.0	0.9779
0.0388	10.0	2800	0.0513	0.4757	0.9514	0.9514	nan	0.9514	0.0	0.9514
0.0385	11.0	3080	0.0660	0.4883	0.9767	0.9767	nan	0.9767	0.0	0.9767
0.0309	12.0	3360	0.0589	0.4808	0.9616	0.9616	nan	0.9616	0.0	0.9616
0.0322	13.0	3640	0.0539	0.4796	0.9592	0.9592	nan	0.9592	0.0	0.9592
0.0361	14.0	3920	0.0621	0.4812	0.9625	0.9625	nan	0.9625	0.0	0.9625
0.0277	15.0	4200	0.0576	0.4836	0.9672	0.9672	nan	0.9672	0.0	0.9672
0.0324	16.0	4480	0.0503	0.4702	0.9404	0.9404	nan	0.9404	0.0	0.9404
0.0355	17.0	4760	0.0583	0.4801	0.9601	0.9601	nan	0.9601	0.0	0.9601
0.032	18.0	5040	0.0528	0.4679	0.9358	0.9358	nan	0.9358	0.0	0.9358
0.0275	19.0	5320	0.0682	0.4828	0.9656	0.9656	nan	0.9656	0.0	0.9656
0.0329	20.0	5600	0.0712	0.4796	0.9591	0.9591	nan	0.9591	0.0	0.9591
0.0284	21.0	5880	0.0769	0.4868	0.9737	0.9737	nan	0.9737	0.0	0.9737
0.028	22.0	6160	0.0615	0.4826	0.9651	0.9651	nan	0.9651	0.0	0.9651
0.0275	23.0	6440	0.0640	0.4797	0.9595	0.9595	nan	0.9595	0.0	0.9595
0.0263	24.0	6720	0.0805	0.4819	0.9639	0.9639	nan	0.9639	0.0	0.9639
0.0252	25.0	7000	0.0700	0.4830	0.9661	0.9661	nan	0.9661	0.0	0.9661
0.0309	26.0	7280	0.0747	0.4854	0.9709	0.9709	nan	0.9709	0.0	0.9709
0.0238	27.0	7560	0.0704	0.4814	0.9628	0.9628	nan	0.9628	0.0	0.9628
0.0277	28.0	7840	0.0757	0.4858	0.9716	0.9716	nan	0.9716	0.0	0.9716
0.0281	29.0	8120	0.0847	0.4830	0.9661	0.9661	nan	0.9661	0.0	0.9661
0.0259	30.0	8400	0.0741	0.4820	0.9640	0.9640	nan	0.9640	0.0	0.9640
0.0231	31.0	8680	0.0726	0.4794	0.9587	0.9587	nan	0.9587	0.0	0.9587
0.0234	32.0	8960	0.0739	0.4779	0.9557	0.9557	nan	0.9557	0.0	0.9557
0.0226	33.0	9240	0.0743	0.4806	0.9613	0.9613	nan	0.9613	0.0	0.9613
0.0242	34.0	9520	0.0776	0.4792	0.9584	0.9584	nan	0.9584	0.0	0.9584
0.0211	35.0	9800	0.0775	0.4765	0.9529	0.9529	nan	0.9529	0.0	0.9529
0.0223	35.71	10000	0.0797	0.4786	0.9572	0.9572	nan	0.9572	0.0	0.9572

Framework versions

• Transformers 4.38.0.dev0
• Pytorch 2.1.0
• Datasets 2.14.4
• Tokenizers 0.15.1