dropoff-utcustom-train-SF-RGBD-b0_7

This model is a fine-tuned version of nvidia/mit-b0 on the sam1120/dropoff-utcustom-TRAIN dataset. It achieves the following results on the evaluation set:

• Loss: 0.2075
• Mean Iou: 0.6372
• Mean Accuracy: 0.6861
• Overall Accuracy: 0.9647
• Accuracy Unlabeled: nan
• Accuracy Dropoff: 0.3822
• Accuracy Undropoff: 0.9900
• Iou Unlabeled: nan
• Iou Dropoff: 0.3104
• Iou Undropoff: 0.9641

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: 8e-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
0.9508	5.0	10	1.0263	0.3104	0.5474	0.8717	nan	0.1937	0.9011	0.0	0.0605	0.8706
0.7814	10.0	20	0.7568	0.4971	0.5339	0.9361	nan	0.0952	0.9726	nan	0.0584	0.9359
0.642	15.0	30	0.5907	0.5134	0.5443	0.9494	nan	0.1026	0.9861	nan	0.0777	0.9492
0.5118	20.0	40	0.4804	0.3658	0.5923	0.9513	nan	0.2006	0.9839	0.0	0.1464	0.9509
0.4581	25.0	50	0.4405	0.3715	0.5915	0.9569	nan	0.1930	0.9900	0.0	0.1578	0.9565
0.4213	30.0	60	0.4146	0.3828	0.6136	0.9580	nan	0.2379	0.9892	0.0	0.1910	0.9575
0.3571	35.0	70	0.3750	0.3846	0.6180	0.9578	nan	0.2474	0.9887	0.0	0.1963	0.9574
0.3205	40.0	80	0.3478	0.5777	0.6202	0.9576	nan	0.2522	0.9882	nan	0.1982	0.9571
0.3114	45.0	90	0.3461	0.3895	0.6423	0.9541	nan	0.3022	0.9824	0.0	0.2150	0.9535
0.2747	50.0	100	0.3253	0.5875	0.6357	0.9575	nan	0.2847	0.9867	nan	0.2180	0.9570
0.2593	55.0	110	0.3083	0.5967	0.6599	0.9552	nan	0.3377	0.9820	nan	0.2387	0.9546
0.2293	60.0	120	0.2762	0.5966	0.6389	0.9606	nan	0.2880	0.9898	nan	0.2331	0.9601
0.2306	65.0	130	0.2655	0.6016	0.6587	0.9577	nan	0.3326	0.9848	nan	0.2462	0.9571
0.2118	70.0	140	0.2446	0.6039	0.6509	0.9605	nan	0.3133	0.9886	nan	0.2479	0.9600
0.2038	75.0	150	0.2395	0.6164	0.6708	0.9607	nan	0.3547	0.9870	nan	0.2727	0.9601
0.1895	80.0	160	0.2196	0.6254	0.6721	0.9636	nan	0.3542	0.9900	nan	0.2878	0.9630
0.1681	85.0	170	0.2176	0.6302	0.6829	0.9630	nan	0.3773	0.9884	nan	0.2979	0.9624
0.1612	90.0	180	0.2175	0.6334	0.6870	0.9633	nan	0.3857	0.9884	nan	0.3042	0.9627
0.1545	95.0	190	0.2140	0.6337	0.6816	0.9644	nan	0.3732	0.9900	nan	0.3035	0.9638
0.1551	100.0	200	0.2134	0.6357	0.6891	0.9637	nan	0.3896	0.9886	nan	0.3083	0.9631
0.1508	105.0	210	0.2090	0.6359	0.6865	0.9642	nan	0.3837	0.9894	nan	0.3083	0.9636
0.1536	110.0	220	0.2057	0.6346	0.6801	0.9650	nan	0.3694	0.9908	nan	0.3048	0.9644
0.1392	115.0	230	0.2083	0.6387	0.6890	0.9646	nan	0.3883	0.9896	nan	0.3133	0.9640
0.1446	120.0	240	0.2075	0.6372	0.6861	0.9647	nan	0.3822	0.9900	nan	0.3104	0.9641

Framework versions

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