dropoff-utcustom-train-SF-RGBD-b5_1

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: 0.3428
• Mean Iou: 0.4792
• Mean Accuracy: 0.5000
• Overall Accuracy: 0.9583
• Accuracy Unlabeled: nan
• Accuracy Dropoff: 0.0001
• Accuracy Undropoff: 0.9999
• Iou Unlabeled: nan
• Iou Dropoff: 0.0001
• Iou Undropoff: 0.9583

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: 3e-06
• 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
0.8047	5.0	10	0.9867	0.2744	0.6315	0.7475	nan	0.5049	0.7581	0.0	0.0812	0.7422
0.7528	10.0	20	0.8526	0.3461	0.5957	0.9213	nan	0.2406	0.9508	0.0	0.1178	0.9205
0.7087	15.0	30	0.7023	0.3450	0.5533	0.9467	nan	0.1243	0.9824	0.0	0.0887	0.9464
0.6601	20.0	40	0.6251	0.3381	0.5390	0.9462	nan	0.0948	0.9832	0.0	0.0684	0.9460
0.6274	25.0	50	0.5828	0.3286	0.5178	0.9486	nan	0.0479	0.9876	0.0	0.0374	0.9485
0.5929	30.0	60	0.5478	0.3257	0.5122	0.9488	nan	0.0359	0.9884	0.0	0.0284	0.9487
0.5672	35.0	70	0.5237	0.3240	0.5088	0.9494	nan	0.0283	0.9893	0.0	0.0227	0.9493
0.5454	40.0	80	0.4966	0.4856	0.5072	0.9529	nan	0.0212	0.9933	nan	0.0183	0.9528
0.5261	45.0	90	0.4700	0.3234	0.5062	0.9553	nan	0.0163	0.9960	0.0	0.0149	0.9552
0.5012	50.0	100	0.4576	0.4832	0.5041	0.9563	nan	0.0107	0.9974	nan	0.0101	0.9563
0.4875	55.0	110	0.4430	0.4811	0.5018	0.9566	nan	0.0058	0.9978	nan	0.0056	0.9565
0.4622	60.0	120	0.4328	0.4800	0.5007	0.9570	nan	0.0031	0.9983	nan	0.0030	0.9570
0.4394	65.0	130	0.4179	0.4796	0.5004	0.9572	nan	0.0021	0.9986	nan	0.0021	0.9572
0.4352	70.0	140	0.4048	0.4795	0.5002	0.9573	nan	0.0016	0.9988	nan	0.0016	0.9573
0.426	75.0	150	0.3881	0.4796	0.5003	0.9577	nan	0.0015	0.9992	nan	0.0014	0.9577
0.4175	80.0	160	0.3794	0.4797	0.5004	0.9579	nan	0.0014	0.9994	nan	0.0014	0.9579
0.4087	85.0	170	0.3742	0.3196	0.5002	0.9577	nan	0.0012	0.9992	0.0	0.0012	0.9577
0.3887	90.0	180	0.3645	0.4792	0.4999	0.9581	nan	0.0003	0.9996	nan	0.0003	0.9581
0.3799	95.0	190	0.3540	0.4791	0.4999	0.9581	nan	0.0001	0.9997	nan	0.0001	0.9581
0.376	100.0	200	0.3511	0.4792	0.4999	0.9582	nan	0.0001	0.9998	nan	0.0001	0.9582
0.3677	105.0	210	0.3452	0.4792	0.4999	0.9582	nan	0.0001	0.9998	nan	0.0001	0.9582
0.358	110.0	220	0.3437	0.4792	0.4999	0.9582	nan	0.0001	0.9998	nan	0.0001	0.9582
0.3997	115.0	230	0.3434	0.4792	0.5000	0.9583	nan	0.0001	0.9999	nan	0.0001	0.9583
0.3769	120.0	240	0.3428	0.4792	0.5000	0.9583	nan	0.0001	0.9999	nan	0.0001	0.9583

Framework versions

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