dropoff-utcustom-train-SF-RGBD-b0_1

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: 0.4979
• Mean Iou: 0.4170
• Mean Accuracy: 0.6846
• Overall Accuracy: 0.9603
• Accuracy Unlabeled: nan
• Accuracy Dropoff: 0.3839
• Accuracy Undropoff: 0.9853
• Iou Unlabeled: 0.0
• Iou Dropoff: 0.2914
• Iou Undropoff: 0.9597

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: 2e-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
1.0495	5.0	10	1.0890	0.1852	0.3572	0.4990	nan	0.2026	0.5119	0.0	0.0474	0.5081
0.9941	10.0	20	1.0479	0.3452	0.8357	0.8479	nan	0.8225	0.8490	0.0	0.1931	0.8425
0.9448	15.0	30	0.9839	0.3790	0.8217	0.9010	nan	0.7351	0.9082	0.0	0.2390	0.8980
0.8912	20.0	40	0.9041	0.3845	0.7150	0.9247	nan	0.4863	0.9437	0.0	0.2303	0.9233
0.8458	25.0	50	0.7997	0.3835	0.6687	0.9326	nan	0.3808	0.9565	0.0	0.2188	0.9316
0.8299	30.0	60	0.7387	0.3751	0.6333	0.9326	nan	0.3068	0.9597	0.0	0.1934	0.9318
0.7518	35.0	70	0.6810	0.3791	0.6322	0.9404	nan	0.2961	0.9683	0.0	0.1975	0.9397
0.6943	40.0	80	0.6322	0.3703	0.6069	0.9422	nan	0.2411	0.9726	0.0	0.1691	0.9417
0.6617	45.0	90	0.6071	0.3780	0.6240	0.9454	nan	0.2734	0.9746	0.0	0.1892	0.9449
0.634	50.0	100	0.5932	0.3765	0.6106	0.9497	nan	0.2407	0.9805	0.0	0.1802	0.9494
0.6157	55.0	110	0.5829	0.3982	0.6538	0.9524	nan	0.3281	0.9795	0.0	0.2425	0.9520
0.5814	60.0	120	0.5708	0.4038	0.6699	0.9533	nan	0.3608	0.9790	0.0	0.2586	0.9528
0.5988	65.0	130	0.5575	0.3974	0.6456	0.9569	nan	0.3061	0.9851	0.0	0.2357	0.9564
0.5583	70.0	140	0.5530	0.4224	0.7075	0.9576	nan	0.4346	0.9803	0.0	0.3103	0.9570
0.5596	75.0	150	0.5264	0.4034	0.6522	0.9598	nan	0.3167	0.9877	0.0	0.2510	0.9593
0.5524	80.0	160	0.5392	0.4208	0.7109	0.9567	nan	0.4429	0.9790	0.0	0.3065	0.9560
0.5294	85.0	170	0.5257	0.4161	0.6913	0.9582	nan	0.4002	0.9824	0.0	0.2909	0.9576
0.5477	90.0	180	0.5178	0.4207	0.6962	0.9591	nan	0.4095	0.9829	0.0	0.3035	0.9584
0.528	95.0	190	0.5185	0.4183	0.6939	0.9590	nan	0.4047	0.9831	0.0	0.2965	0.9584
0.5144	100.0	200	0.5004	0.4153	0.6788	0.9604	nan	0.3716	0.9860	0.0	0.2859	0.9599
0.5313	105.0	210	0.5032	0.4199	0.7005	0.9585	nan	0.4191	0.9819	0.0	0.3020	0.9578
0.5172	110.0	220	0.4993	0.4188	0.6931	0.9591	nan	0.4030	0.9832	0.0	0.2978	0.9585
0.5124	115.0	230	0.4999	0.4167	0.6828	0.9606	nan	0.3799	0.9858	0.0	0.2901	0.9600
0.5025	120.0	240	0.4979	0.4170	0.6846	0.9603	nan	0.3839	0.9853	0.0	0.2914	0.9597

Framework versions

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