dropoff-utcustom-train-SF-RGBD-b0_2

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: 0.4274
• Mean Iou: 0.6102
• Mean Accuracy: 0.6603
• Overall Accuracy: 0.9607
• Accuracy Unlabeled: nan
• Accuracy Dropoff: 0.3326
• Accuracy Undropoff: 0.9879
• Iou Unlabeled: nan
• Iou Dropoff: 0.2602
• Iou Undropoff: 0.9601

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: 3e-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
1.0555	5.0	10	1.0734	0.2254	0.4211	0.6018	nan	0.2240	0.6182	0.0	0.0622	0.6140
0.9825	10.0	20	1.0261	0.2992	0.6380	0.7780	nan	0.4852	0.7907	0.0	0.1170	0.7807
0.8991	15.0	30	0.8985	0.3231	0.5517	0.8892	nan	0.1836	0.9198	0.0	0.0776	0.8917
0.8191	20.0	40	0.7413	0.3270	0.5262	0.9299	nan	0.0858	0.9665	0.0	0.0513	0.9296
0.7562	25.0	50	0.6268	0.3259	0.5130	0.9436	nan	0.0433	0.9826	0.0	0.0343	0.9435
0.7395	30.0	60	0.5872	0.3235	0.5073	0.9498	nan	0.0246	0.9900	0.0	0.0206	0.9498
0.7272	35.0	70	0.5820	0.3379	0.5415	0.9411	nan	0.1055	0.9774	0.0	0.0729	0.9409
0.6525	40.0	80	0.5571	0.3445	0.5451	0.9498	nan	0.1036	0.9865	0.0	0.0839	0.9496
0.6161	45.0	90	0.5465	0.3480	0.5480	0.9528	nan	0.1064	0.9895	0.0	0.0914	0.9526
0.6131	50.0	100	0.5379	0.3712	0.5917	0.9555	nan	0.1949	0.9885	0.0	0.1584	0.9551
0.579	55.0	110	0.5229	0.3892	0.6411	0.9536	nan	0.3002	0.9819	0.0	0.2146	0.9530
0.5133	60.0	120	0.5113	0.3962	0.6596	0.9541	nan	0.3384	0.9808	0.0	0.2352	0.9535
0.535	65.0	130	0.4925	0.3981	0.6566	0.9561	nan	0.3299	0.9833	0.0	0.2386	0.9555
0.4866	70.0	140	0.4717	0.5993	0.6516	0.9584	nan	0.3169	0.9863	nan	0.2407	0.9579
0.5119	75.0	150	0.4712	0.5976	0.6513	0.9578	nan	0.3171	0.9856	nan	0.2380	0.9572
0.5034	80.0	160	0.4737	0.6120	0.6840	0.9562	nan	0.3872	0.9808	nan	0.2686	0.9554
0.4503	85.0	170	0.4496	0.6103	0.6618	0.9604	nan	0.3361	0.9875	nan	0.2607	0.9598
0.4653	90.0	180	0.4617	0.6201	0.6907	0.9580	nan	0.3992	0.9822	nan	0.2830	0.9572
0.4375	95.0	190	0.4412	0.6090	0.6592	0.9605	nan	0.3305	0.9878	nan	0.2580	0.9599
0.4306	100.0	200	0.4355	0.6120	0.6653	0.9602	nan	0.3436	0.9870	nan	0.2643	0.9597
0.4456	105.0	210	0.4414	0.6178	0.6756	0.9601	nan	0.3653	0.9860	nan	0.2760	0.9595
0.4435	110.0	220	0.4387	0.6150	0.6681	0.9608	nan	0.3489	0.9873	nan	0.2699	0.9602
0.4263	115.0	230	0.4348	0.6156	0.6692	0.9607	nan	0.3512	0.9872	nan	0.2711	0.9602
0.4123	120.0	240	0.4274	0.6102	0.6603	0.9607	nan	0.3326	0.9879	nan	0.2602	0.9601

Framework versions

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