dropoff-utcustom-train-SF-RGB-b0_6

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: 0.1833
• Mean Iou: 0.6595
• Mean Accuracy: 0.7018
• Overall Accuracy: 0.9666
• Accuracy Unlabeled: nan
• Accuracy Dropoff: 0.4113
• Accuracy Undropoff: 0.9924
• Iou Unlabeled: nan
• Iou Dropoff: 0.3531
• Iou Undropoff: 0.9660

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-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
1.1234	3.33	10	1.0973	0.1779	0.5629	0.3723	nan	0.7720	0.3538	0.0	0.1801	0.3536
0.975	6.67	20	1.0260	0.3499	0.8180	0.8109	nan	0.8259	0.8102	0.0	0.2428	0.8069
0.9464	10.0	30	0.8130	0.4297	0.7456	0.9502	nan	0.5212	0.9700	0.0	0.3385	0.9507
0.6167	13.33	40	0.6001	0.4451	0.7438	0.9617	nan	0.5048	0.9829	0.0	0.3743	0.9610
0.4818	16.67	50	0.4629	0.4491	0.7237	0.9666	nan	0.4573	0.9902	0.0	0.3815	0.9659
0.4733	20.0	60	0.4379	0.4335	0.7067	0.9630	nan	0.4256	0.9879	0.0	0.3383	0.9623
0.3843	23.33	70	0.4073	0.4310	0.6872	0.9652	nan	0.3821	0.9922	0.0	0.3283	0.9646
0.3579	26.67	80	0.3731	0.4354	0.6999	0.9651	nan	0.4090	0.9908	0.0	0.3418	0.9644
0.3212	30.0	90	0.3655	0.6589	0.7129	0.9647	nan	0.4366	0.9892	nan	0.3538	0.9640
0.3088	33.33	100	0.3306	0.6310	0.6689	0.9641	nan	0.3451	0.9928	nan	0.2985	0.9635
0.2825	36.67	110	0.3253	0.6633	0.7103	0.9663	nan	0.4293	0.9912	nan	0.3609	0.9657
0.3029	40.0	120	0.3130	0.6556	0.7079	0.9645	nan	0.4264	0.9895	nan	0.3474	0.9638
0.252	43.33	130	0.2898	0.6703	0.7310	0.9652	nan	0.4740	0.9880	nan	0.3762	0.9645
0.2395	46.67	140	0.2843	0.6587	0.7088	0.9653	nan	0.4275	0.9902	nan	0.3527	0.9646
0.2308	50.0	150	0.2744	0.6481	0.6870	0.9659	nan	0.3811	0.9930	nan	0.3309	0.9653
0.2125	53.33	160	0.2579	0.6555	0.7028	0.9653	nan	0.4147	0.9909	nan	0.3464	0.9647
0.1953	56.67	170	0.2551	0.6549	0.7054	0.9647	nan	0.4209	0.9899	nan	0.3458	0.9641
0.1743	60.0	180	0.2377	0.6393	0.6768	0.9651	nan	0.3605	0.9931	nan	0.3140	0.9646
0.17	63.33	190	0.2342	0.6564	0.7002	0.9660	nan	0.4086	0.9918	nan	0.3474	0.9654
0.173	66.67	200	0.2296	0.6629	0.7095	0.9664	nan	0.4277	0.9913	nan	0.3602	0.9657
0.1487	70.0	210	0.2152	0.6525	0.6861	0.9673	nan	0.3777	0.9946	nan	0.3383	0.9667
0.1501	73.33	220	0.2179	0.6593	0.7019	0.9665	nan	0.4116	0.9923	nan	0.3527	0.9659
0.1419	76.67	230	0.2055	0.6605	0.7057	0.9663	nan	0.4199	0.9916	nan	0.3553	0.9656
0.2049	80.0	240	0.2060	0.6563	0.7004	0.9659	nan	0.4091	0.9917	nan	0.3472	0.9653
0.1339	83.33	250	0.2006	0.6514	0.6921	0.9660	nan	0.3916	0.9926	nan	0.3375	0.9654
0.1262	86.67	260	0.1963	0.6559	0.7033	0.9654	nan	0.4158	0.9908	nan	0.3470	0.9647
0.179	90.0	270	0.1907	0.6549	0.6976	0.9660	nan	0.4032	0.9921	nan	0.3445	0.9654
0.1216	93.33	280	0.1901	0.6561	0.6994	0.9661	nan	0.4068	0.9920	nan	0.3468	0.9655
0.1144	96.67	290	0.1917	0.6565	0.7017	0.9658	nan	0.4119	0.9915	nan	0.3478	0.9652
0.1095	100.0	300	0.1900	0.6621	0.7108	0.9659	nan	0.4309	0.9907	nan	0.3590	0.9653
0.1144	103.33	310	0.1848	0.6595	0.6994	0.9670	nan	0.4058	0.9930	nan	0.3526	0.9664
0.1144	106.67	320	0.1849	0.6585	0.7011	0.9665	nan	0.4100	0.9922	nan	0.3512	0.9658
0.1574	110.0	330	0.1852	0.6592	0.7025	0.9664	nan	0.4128	0.9921	nan	0.3526	0.9658
0.1085	113.33	340	0.1819	0.6595	0.7016	0.9667	nan	0.4108	0.9924	nan	0.3530	0.9660
0.1099	116.67	350	0.1856	0.6602	0.7057	0.9662	nan	0.4198	0.9915	nan	0.3548	0.9656
0.1048	120.0	360	0.1833	0.6595	0.7018	0.9666	nan	0.4113	0.9924	nan	0.3531	0.9660

Framework versions

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