SegFormer_Clean_Set1_240430_V2-Augmented_mit-b5

This model is a fine-tuned version of nvidia/mit-b5 on the Hasano20/Clean_Set1_240430_V2-Augmented dataset. It achieves the following results on the evaluation set:

• Loss: 0.0899
• Mean Iou: 0.8524
• Mean Accuracy: 0.8932
• Overall Accuracy: 0.9653
• Accuracy Background: 0.9900
• Accuracy Melt: 0.7107
• Accuracy Substrate: 0.9788
• Iou Background: 0.9693
• Iou Melt: 0.6451
• Iou Substrate: 0.9429

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: 0.0001
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Background	Accuracy Melt	Accuracy Substrate	Iou Background	Iou Melt	Iou Substrate
0.1394	1.6129	50	0.2486	0.6252	0.6776	0.9171	0.9800	0.0713	0.9816	0.9285	0.0680	0.8792
0.2482	3.2258	100	0.2178	0.6883	0.7470	0.9224	0.9831	0.3037	0.9543	0.9307	0.2490	0.8854
0.1697	4.8387	150	0.2044	0.6993	0.7613	0.9236	0.9847	0.3511	0.9480	0.9313	0.2796	0.8871
0.139	6.4516	200	0.1897	0.7250	0.7835	0.9317	0.9771	0.4086	0.9648	0.9415	0.3395	0.8940
0.0951	8.0645	250	0.1879	0.6863	0.7344	0.9291	0.9851	0.2414	0.9766	0.9372	0.2290	0.8928
0.0812	9.6774	300	0.1875	0.7513	0.8449	0.9285	0.9636	0.6338	0.9372	0.9370	0.4265	0.8903
0.1349	11.2903	350	0.2020	0.6825	0.7357	0.9247	0.9810	0.2577	0.9685	0.9328	0.2265	0.8882
0.1312	12.9032	400	0.1401	0.7627	0.8053	0.9465	0.9864	0.4477	0.9816	0.9624	0.4169	0.9090
0.1061	14.5161	450	0.1051	0.8297	0.8811	0.9586	0.9890	0.6853	0.9691	0.9657	0.5932	0.9302
0.0287	16.1290	500	0.1045	0.8349	0.8835	0.9598	0.9850	0.6905	0.9749	0.9640	0.6073	0.9335
0.2051	17.7419	550	0.0928	0.8466	0.8868	0.9644	0.9875	0.6906	0.9824	0.9687	0.6290	0.9420
0.0898	19.3548	600	0.0899	0.8524	0.8932	0.9653	0.9900	0.7107	0.9788	0.9693	0.6451	0.9429

Framework versions

• Transformers 4.41.2
• Pytorch 2.0.1+cu117
• Datasets 2.19.2
• Tokenizers 0.19.1