Modelpublic

sam1120/dropoff-utcustom-train-SF-RGBD-b5_2

sourceHugging Faceotherupdated 3y agoView on Hugging Face

0likes8downloads

Model Card

dropoff-utcustom-train-SF-RGBD-b5_2

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: 0.4198
—Mean Iou: 0.3194
—Mean Accuracy: 0.4998
—Overall Accuracy: 0.9558
—Accuracy Unlabeled: nan
—Accuracy Dropoff: 0.0023
—Accuracy Undropoff: 0.9972
—Iou Unlabeled: 0.0
—Iou Dropoff: 0.0022
—Iou Undropoff: 0.9558

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-06
—trainbatchsize: 16
—evalbatchsize: 16
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—lrschedulerwarmup_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 Dropoff	Iou Undropoff
0.989	5.0	10	1.0190	0.2162	0.5831	0.5879	nan	0.5779	0.5883	0.0657	0.5829
0.9092	10.0	20	0.8686	0.3164	0.5199	0.8922	nan	0.1137	0.9260	0.0539	0.8953
0.8483	15.0	30	0.7438	0.3256	0.5234	0.9219	nan	0.0888	0.9581	0.0545	0.9224
0.7856	20.0	40	0.6571	0.3182	0.5013	0.9336	nan	0.0297	0.9728	0.0210	0.9335
0.7459	25.0	50	0.6144	0.3164	0.4980	0.9324	nan	0.0242	0.9718	0.0168	0.9324
0.7027	30.0	60	0.5861	0.3168	0.4975	0.9351	nan	0.0202	0.9748	0.0151	0.9353
0.6827	35.0	70	0.5568	0.3171	0.4975	0.9391	nan	0.0159	0.9791	0.0122	0.9391
0.6362	40.0	80	0.5405	0.3179	0.4982	0.9424	nan	0.0138	0.9827	0.0112	0.9425
0.6098	45.0	90	0.5192	0.3174	0.4971	0.9449	nan	0.0087	0.9855	0.0073	0.9449
0.5946	50.0	100	0.5025	0.3179	0.4978	0.9475	nan	0.0072	0.9883	0.0062	0.9477
0.5868	55.0	110	0.4943	0.3179	0.4976	0.9490	nan	0.0052	0.9900	0.0046	0.9491
0.5557	60.0	120	0.4798	0.3184	0.4983	0.9505	nan	0.0051	0.9915	0.0045	0.9506
0.5327	65.0	130	0.4736	0.3184	0.4983	0.9514	nan	0.0041	0.9925	0.0038	0.9514
0.525	70.0	140	0.4657	0.3187	0.4987	0.9526	nan	0.0038	0.9937	0.0035	0.9526
0.5266	75.0	150	0.4528	0.3190	0.4992	0.9534	nan	0.0037	0.9946	0.0034	0.9535
0.5139	80.0	160	0.4538	0.3189	0.4991	0.9533	nan	0.0037	0.9945	0.0035	0.9534
0.5128	85.0	170	0.4460	0.3192	0.4995	0.9543	nan	0.0033	0.9956	0.0031	0.9543
0.4901	90.0	180	0.4371	0.3192	0.4995	0.9548	nan	0.0029	0.9961	0.0027	0.9548
0.4767	95.0	190	0.4325	0.3193	0.4997	0.9552	nan	0.0029	0.9965	0.0027	0.9552
0.4692	100.0	200	0.4272	0.3193	0.4997	0.9556	nan	0.0024	0.9970	0.0023	0.9556
0.4632	105.0	210	0.4251	0.3193	0.4996	0.9556	nan	0.0023	0.9969	0.0023	0.9556
0.4626	110.0	220	0.4236	0.3193	0.4997	0.9556	nan	0.0024	0.9970	0.0024	0.9556
0.4837	115.0	230	0.4216	0.3194	0.4998	0.9558	nan	0.0023	0.9972	0.0023	0.9558
0.4809	120.0	240	0.4198	0.3194	0.4998	0.9558	nan	0.0023	0.9972	0.0022	0.9558

Framework versions

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