dropoff-utcustom-train-SF-RGB-b5_7

This model is a fine-tuned version of nvidia/mit-b5 on the sam1120/dropoff-utcustom-TRAIN dataset. It achieves the following results on the evaluation set:

• Loss: 0.1841
• Mean Iou: 0.7025
• Mean Accuracy: 0.7532
• Overall Accuracy: 0.9721
• Accuracy Unlabeled: nan
• Accuracy Dropoff: 0.5145
• Accuracy Undropoff: 0.9919
• Iou Unlabeled: nan
• Iou Dropoff: 0.4336
• Iou Undropoff: 0.9715

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: 5e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 120

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Unlabeled	Accuracy Dropoff	Accuracy Undropoff	Iou Unlabeled	Iou Dropoff	Iou Undropoff
0.8255	5.0	10	0.7949	0.4128	0.7856	0.9393	nan	0.6179	0.9533	0.0	0.3007	0.9377
0.4434	10.0	20	0.4247	0.4471	0.7066	0.9705	nan	0.4187	0.9944	0.0	0.3714	0.9700
0.2107	15.0	30	0.2726	0.6711	0.7003	0.9715	nan	0.4046	0.9961	nan	0.3713	0.9710
0.1678	20.0	40	0.2388	0.6801	0.7343	0.9691	nan	0.4782	0.9904	nan	0.3917	0.9685
0.0972	25.0	50	0.1849	0.6764	0.7096	0.9715	nan	0.4241	0.9952	nan	0.3818	0.9709
0.0604	30.0	60	0.2019	0.4644	0.7568	0.9704	nan	0.5239	0.9897	0.0	0.4236	0.9697
0.0497	35.0	70	0.1793	0.6838	0.7345	0.9700	nan	0.4775	0.9914	nan	0.3983	0.9694
0.0492	40.0	80	0.2000	0.4639	0.7567	0.9702	nan	0.5239	0.9896	0.0	0.4223	0.9695
0.0409	45.0	90	0.1893	0.7030	0.7778	0.9696	nan	0.5687	0.9869	nan	0.4372	0.9688
0.0328	50.0	100	0.1842	0.7040	0.7715	0.9704	nan	0.5545	0.9885	nan	0.4382	0.9697
0.0332	55.0	110	0.1781	0.7015	0.7563	0.9715	nan	0.5216	0.9910	nan	0.4322	0.9709
0.0314	60.0	120	0.1732	0.6890	0.7305	0.9717	nan	0.4675	0.9935	nan	0.4068	0.9711
0.0318	65.0	130	0.1786	0.6971	0.7477	0.9715	nan	0.5037	0.9918	nan	0.4233	0.9709
0.0291	70.0	140	0.1814	0.7119	0.7687	0.9725	nan	0.5466	0.9909	nan	0.4521	0.9718
0.0273	75.0	150	0.1755	0.7101	0.7677	0.9722	nan	0.5446	0.9907	nan	0.4487	0.9715
0.0274	80.0	160	0.1786	0.7006	0.7494	0.9720	nan	0.5066	0.9922	nan	0.4297	0.9714
0.0248	85.0	170	0.1741	0.7029	0.7526	0.9722	nan	0.5131	0.9921	nan	0.4341	0.9716
0.0248	90.0	180	0.1832	0.7050	0.7595	0.9719	nan	0.5278	0.9912	nan	0.4387	0.9713
0.0242	95.0	190	0.1808	0.7028	0.7539	0.9720	nan	0.5160	0.9918	nan	0.4341	0.9714
0.024	100.0	200	0.1796	0.7022	0.7501	0.9723	nan	0.5077	0.9925	nan	0.4327	0.9717
0.0231	105.0	210	0.1835	0.7137	0.7731	0.9724	nan	0.5557	0.9905	nan	0.4556	0.9717
0.0238	110.0	220	0.1823	0.7046	0.7565	0.9721	nan	0.5214	0.9917	nan	0.4376	0.9715
0.0228	115.0	230	0.1833	0.7009	0.7504	0.9720	nan	0.5088	0.9921	nan	0.4305	0.9714
0.0255	120.0	240	0.1841	0.7025	0.7532	0.9721	nan	0.5145	0.9919	nan	0.4336	0.9715

Framework versions

• Transformers 4.30.2
• Pytorch 2.0.1+cu117
• Datasets 2.13.1
• Tokenizers 0.13.3