dropoff-utcustom-train-SF-RGBD-b5_4

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: 0.2351
• Mean Iou: 0.4792
• Mean Accuracy: 0.5
• Overall Accuracy: 0.9584
• Accuracy Unlabeled: nan
• Accuracy Dropoff: 0.0
• Accuracy Undropoff: 1.0
• Iou Unlabeled: nan
• Iou Dropoff: 0.0
• Iou Undropoff: 0.9584

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: 7e-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
1.0114	5.0	10	1.0037	0.2459	0.4345	0.7074	nan	0.1368	0.7322	0.0	0.0286	0.7089
0.9088	10.0	20	0.8245	0.3119	0.5046	0.8887	nan	0.0857	0.9235	0.0	0.0460	0.8897
0.8029	15.0	30	0.6620	0.3157	0.4998	0.9214	nan	0.0399	0.9596	0.0	0.0253	0.9219
0.6935	20.0	40	0.5662	0.3154	0.4959	0.9309	nan	0.0214	0.9704	0.0	0.0151	0.9311
0.635	25.0	50	0.5018	0.3175	0.4978	0.9401	nan	0.0153	0.9803	0.0	0.0121	0.9404
0.5579	30.0	60	0.4701	0.3178	0.4978	0.9422	nan	0.0131	0.9825	0.0	0.0111	0.9423
0.5086	35.0	70	0.4403	0.3181	0.4977	0.9459	nan	0.0088	0.9866	0.0	0.0080	0.9461
0.472	40.0	80	0.4328	0.3177	0.4971	0.9471	nan	0.0063	0.9879	0.0	0.0059	0.9473
0.4484	45.0	90	0.4136	0.3184	0.4981	0.9506	nan	0.0046	0.9916	0.0	0.0044	0.9508
0.4026	50.0	100	0.4013	0.3186	0.4985	0.9516	nan	0.0043	0.9926	0.0	0.0042	0.9517
0.3873	55.0	110	0.3621	0.3189	0.4991	0.9557	nan	0.0010	0.9971	0.0	0.0009	0.9557
0.3549	60.0	120	0.3479	0.3189	0.4992	0.9564	nan	0.0004	0.9979	0.0	0.0004	0.9564
0.3358	65.0	130	0.3282	0.3191	0.4994	0.9571	nan	0.0001	0.9986	0.0	0.0001	0.9571
0.3146	70.0	140	0.3141	0.3193	0.4996	0.9577	nan	0.0000	0.9993	0.0	0.0000	0.9577
0.3116	75.0	150	0.2941	0.3194	0.4999	0.9582	nan	0.0	0.9998	0.0	0.0	0.9582
0.3151	80.0	160	0.2809	0.3195	0.5000	0.9584	nan	0.0	0.9999	0.0	0.0	0.9584
0.2778	85.0	170	0.2750	0.3195	0.5000	0.9584	nan	0.0	1.0000	0.0	0.0	0.9584
0.2753	90.0	180	0.2615	0.3195	0.5000	0.9584	nan	0.0	1.0000	0.0	0.0	0.9584
0.2809	95.0	190	0.2547	0.4792	0.5	0.9584	nan	0.0	1.0	nan	0.0	0.9584
0.2606	100.0	200	0.2464	0.4792	0.5	0.9584	nan	0.0	1.0	nan	0.0	0.9584
0.2563	105.0	210	0.2459	0.4792	0.5	0.9584	nan	0.0	1.0	nan	0.0	0.9584
0.2454	110.0	220	0.2393	0.4792	0.5	0.9584	nan	0.0	1.0	nan	0.0	0.9584
0.2707	115.0	230	0.2368	0.4792	0.5	0.9584	nan	0.0	1.0	nan	0.0	0.9584
0.2433	120.0	240	0.2351	0.4792	0.5	0.9584	nan	0.0	1.0	nan	0.0	0.9584

Framework versions

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