dropoff-utcustom-train-SF-RGBD-b0_3

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: 0.3666
• Mean Iou: 0.6400
• Mean Accuracy: 0.7120
• Overall Accuracy: 0.9610
• Accuracy Unlabeled: nan
• Accuracy Dropoff: 0.4404
• Accuracy Undropoff: 0.9836
• Iou Unlabeled: nan
• Iou Dropoff: 0.3196
• Iou Undropoff: 0.9603

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: 4e-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
1.0352	5.0	10	1.0676	0.2560	0.5776	0.7142	nan	0.4286	0.7266	0.0	0.0589	0.7090
0.9564	10.0	20	0.9743	0.3355	0.5576	0.9248	nan	0.1571	0.9581	0.0	0.0822	0.9243
0.8577	15.0	30	0.8504	0.3318	0.5283	0.9409	nan	0.0782	0.9784	0.0	0.0545	0.9407
0.7512	20.0	40	0.6972	0.3270	0.5122	0.9527	nan	0.0318	0.9926	0.0	0.0283	0.9526
0.6955	25.0	50	0.5761	0.3259	0.5099	0.9545	nan	0.0250	0.9948	0.0	0.0234	0.9544
0.6691	30.0	60	0.5209	0.3360	0.5271	0.9525	nan	0.0632	0.9911	0.0	0.0557	0.9524
0.626	35.0	70	0.5297	0.3408	0.5362	0.9505	nan	0.0844	0.9881	0.0	0.0719	0.9503
0.5544	40.0	80	0.5263	0.3616	0.5757	0.9521	nan	0.1652	0.9862	0.0	0.1330	0.9518
0.5316	45.0	90	0.4825	0.3836	0.6353	0.9506	nan	0.2915	0.9792	0.0	0.2009	0.9500
0.4929	50.0	100	0.4763	0.3958	0.6588	0.9530	nan	0.3378	0.9797	0.0	0.2352	0.9524
0.468	55.0	110	0.4583	0.4077	0.6974	0.9528	nan	0.4188	0.9759	0.0	0.2713	0.9519
0.429	60.0	120	0.4268	0.3985	0.6526	0.9575	nan	0.3199	0.9852	0.0	0.2386	0.9569
0.4211	65.0	130	0.3988	0.3951	0.6406	0.9584	nan	0.2939	0.9872	0.0	0.2275	0.9578
0.3926	70.0	140	0.4085	0.4102	0.6780	0.9587	nan	0.3718	0.9842	0.0	0.2726	0.9581
0.4006	75.0	150	0.3944	0.6077	0.6574	0.9604	nan	0.3269	0.9879	nan	0.2555	0.9599
0.3978	80.0	160	0.3881	0.6216	0.6875	0.9591	nan	0.3912	0.9838	nan	0.2848	0.9585
0.3553	85.0	170	0.3877	0.6333	0.7077	0.9595	nan	0.4329	0.9824	nan	0.3079	0.9588
0.3637	90.0	180	0.4004	0.6428	0.7273	0.9594	nan	0.4741	0.9805	nan	0.3270	0.9586
0.3416	95.0	190	0.3835	0.6403	0.7166	0.9604	nan	0.4507	0.9825	nan	0.3210	0.9596
0.342	100.0	200	0.3634	0.6371	0.7061	0.9611	nan	0.4279	0.9842	nan	0.3137	0.9604
0.3393	105.0	210	0.3740	0.6429	0.7217	0.9604	nan	0.4614	0.9820	nan	0.3262	0.9596
0.3535	110.0	220	0.3771	0.6423	0.7199	0.9605	nan	0.4575	0.9823	nan	0.3249	0.9597
0.3159	115.0	230	0.3710	0.6423	0.7167	0.9610	nan	0.4502	0.9832	nan	0.3243	0.9603
0.3278	120.0	240	0.3666	0.6400	0.7120	0.9610	nan	0.4404	0.9836	nan	0.3196	0.9603

Framework versions

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