dropoff-utcustom-train-SF-RGBD-b5_7

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: 0.1296
• Mean Iou: 0.6242
• Mean Accuracy: 0.6623
• Overall Accuracy: 0.9652
• Accuracy Unlabeled: nan
• Accuracy Dropoff: 0.3319
• Accuracy Undropoff: 0.9926
• Iou Unlabeled: nan
• Iou Dropoff: 0.2838
• Iou Undropoff: 0.9647

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: 5e-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
0.9278	5.0	10	0.8454	0.3197	0.5545	0.8788	nan	0.2009	0.9082	0.0	0.0807	0.8785
0.5551	10.0	20	0.4668	0.3221	0.5042	0.9540	nan	0.0135	0.9948	0.0	0.0122	0.9540
0.3667	15.0	30	0.3354	0.3218	0.5035	0.9570	nan	0.0088	0.9982	0.0	0.0085	0.9570
0.2402	20.0	40	0.2678	0.5985	0.6492	0.9587	nan	0.3116	0.9868	nan	0.2388	0.9582
0.1562	25.0	50	0.2101	0.6240	0.6719	0.9631	nan	0.3544	0.9895	nan	0.2854	0.9625
0.1159	30.0	60	0.1704	0.6262	0.6641	0.9654	nan	0.3353	0.9928	nan	0.2875	0.9650
0.0869	35.0	70	0.1443	0.6380	0.6817	0.9657	nan	0.3720	0.9915	nan	0.3108	0.9652
0.079	40.0	80	0.1350	0.6072	0.6360	0.9654	nan	0.2766	0.9953	nan	0.2494	0.9650
0.0647	45.0	90	0.1370	0.5800	0.6031	0.9643	nan	0.2090	0.9971	nan	0.1959	0.9640
0.0587	50.0	100	0.1336	0.6276	0.6796	0.9628	nan	0.3707	0.9885	nan	0.2929	0.9622
0.0575	55.0	110	0.1313	0.6189	0.6531	0.9654	nan	0.3126	0.9937	nan	0.2729	0.9649
0.0527	60.0	120	0.1298	0.6252	0.6655	0.9648	nan	0.3391	0.9920	nan	0.2860	0.9643
0.0491	65.0	130	0.1313	0.6110	0.6492	0.9635	nan	0.3063	0.9920	nan	0.2589	0.9631
0.0441	70.0	140	0.1295	0.6103	0.6429	0.9648	nan	0.2919	0.9939	nan	0.2562	0.9643
0.0426	75.0	150	0.1233	0.6271	0.6633	0.9659	nan	0.3333	0.9933	nan	0.2887	0.9654
0.0477	80.0	160	0.1286	0.6255	0.6629	0.9655	nan	0.3328	0.9929	nan	0.2861	0.9650
0.039	85.0	170	0.1265	0.6380	0.6824	0.9656	nan	0.3735	0.9913	nan	0.3109	0.9650
0.0378	90.0	180	0.1309	0.6185	0.6543	0.9650	nan	0.3154	0.9932	nan	0.2725	0.9645
0.0362	95.0	190	0.1266	0.6311	0.6715	0.9655	nan	0.3508	0.9922	nan	0.2973	0.9650
0.0394	100.0	200	0.1307	0.6274	0.6635	0.9659	nan	0.3337	0.9934	nan	0.2894	0.9655
0.0362	105.0	210	0.1271	0.6366	0.6789	0.9658	nan	0.3661	0.9918	nan	0.3080	0.9653
0.0361	110.0	220	0.1274	0.6317	0.6736	0.9653	nan	0.3554	0.9918	nan	0.2987	0.9648
0.0353	115.0	230	0.1290	0.6216	0.6579	0.9652	nan	0.3228	0.9931	nan	0.2784	0.9647
0.0344	120.0	240	0.1296	0.6242	0.6623	0.9652	nan	0.3319	0.9926	nan	0.2838	0.9647

Framework versions

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