Modelpublic

sam1120/dropoff-utcustom-train-SF-RGB-b0_4

sourceHugging Faceotherupdated 3y agoView on Hugging Face

0likes18downloads

Model Card

dropoff-utcustom-train-SF-RGB-b0_4

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: 0.3032
—Mean Iou: 0.6301
—Mean Accuracy: 0.6710
—Overall Accuracy: 0.9634
—Accuracy Unlabeled: nan
—Accuracy Dropoff: 0.3502
—Accuracy Undropoff: 0.9918
—Iou Unlabeled: nan
—Iou Dropoff: 0.2973
—Iou Undropoff: 0.9628

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
—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 Unlabeled	Iou Dropoff	Iou Undropoff
1.0311	3.33	10	1.0742	0.2063	0.6373	0.5492	nan	0.7339	0.5406	0.0	0.0848	0.5342
0.9741	6.67	20	1.0151	0.3072	0.8067	0.7686	nan	0.8485	0.7649	0.0	0.1619	0.7596
0.9441	10.0	30	0.9345	0.3432	0.8327	0.8408	nan	0.8239	0.8416	0.0	0.1947	0.8348
0.8222	13.33	40	0.8358	0.3643	0.8236	0.8773	nan	0.7646	0.8825	0.0	0.2199	0.8731
0.7243	16.67	50	0.7135	0.3924	0.7838	0.9194	nan	0.6350	0.9325	0.0	0.2603	0.9170
0.7213	20.0	60	0.6358	0.4054	0.7528	0.9374	nan	0.5502	0.9554	0.0	0.2805	0.9359
0.5836	23.33	70	0.5604	0.4211	0.7412	0.9505	nan	0.5115	0.9708	0.0	0.3139	0.9493
0.5285	26.67	80	0.5227	0.4281	0.7570	0.9519	nan	0.5432	0.9708	0.0	0.3335	0.9507
0.4955	30.0	90	0.4478	0.4191	0.6945	0.9581	nan	0.4052	0.9837	0.0	0.2999	0.9573
0.4646	33.33	100	0.4537	0.4215	0.6998	0.9584	nan	0.4161	0.9835	0.0	0.3069	0.9576
0.4356	36.67	110	0.4454	0.4224	0.7105	0.9569	nan	0.4402	0.9808	0.0	0.3112	0.9560
0.4829	40.0	120	0.4099	0.4196	0.6901	0.9593	nan	0.3947	0.9854	0.0	0.3002	0.9585
0.4051	43.33	130	0.3911	0.6267	0.6784	0.9607	nan	0.3687	0.9881	nan	0.2933	0.9600
0.3916	46.67	140	0.3841	0.4183	0.6897	0.9586	nan	0.3946	0.9847	0.0	0.2969	0.9579
0.3713	50.0	150	0.3788	0.4248	0.7001	0.9600	nan	0.4149	0.9853	0.0	0.3150	0.9593
0.359	53.33	160	0.3719	0.6254	0.6761	0.9607	nan	0.3639	0.9883	nan	0.2908	0.9601
0.3459	56.67	170	0.3610	0.6245	0.6774	0.9601	nan	0.3673	0.9876	nan	0.2895	0.9594
0.3099	60.0	180	0.3455	0.6246	0.6687	0.9620	nan	0.3468	0.9905	nan	0.2879	0.9614
0.3124	63.33	190	0.3436	0.6277	0.6763	0.9615	nan	0.3634	0.9892	nan	0.2946	0.9608
0.3283	66.67	200	0.3344	0.6237	0.6607	0.9634	nan	0.3286	0.9928	nan	0.2845	0.9629
0.2974	70.0	210	0.3412	0.6312	0.6817	0.9616	nan	0.3746	0.9888	nan	0.3014	0.9609
0.3003	73.33	220	0.3322	0.6320	0.6877	0.9607	nan	0.3881	0.9872	nan	0.3041	0.9600
0.2968	76.67	230	0.3289	0.6344	0.6807	0.9628	nan	0.3712	0.9902	nan	0.3066	0.9622
0.4415	80.0	240	0.3333	0.6320	0.6800	0.9622	nan	0.3705	0.9896	nan	0.3024	0.9615
0.2836	83.33	250	0.3271	0.6287	0.6757	0.9619	nan	0.3617	0.9897	nan	0.2960	0.9613
0.2762	86.67	260	0.3203	0.6263	0.6673	0.9629	nan	0.3429	0.9916	nan	0.2903	0.9623
0.3901	90.0	270	0.3186	0.6290	0.6787	0.9614	nan	0.3685	0.9889	nan	0.2971	0.9608
0.2755	93.33	280	0.3086	0.6283	0.6693	0.9631	nan	0.3468	0.9917	nan	0.2940	0.9625
0.2652	96.67	290	0.3099	0.6302	0.6779	0.9620	nan	0.3661	0.9896	nan	0.2991	0.9614
0.2627	100.0	300	0.3056	0.6294	0.6728	0.9627	nan	0.3548	0.9909	nan	0.2966	0.9622
0.2647	103.33	310	0.3036	0.6292	0.6689	0.9635	nan	0.3458	0.9921	nan	0.2954	0.9629
0.2697	106.67	320	0.3043	0.6298	0.6713	0.9632	nan	0.3510	0.9916	nan	0.2970	0.9626
0.3878	110.0	330	0.3037	0.6297	0.6740	0.9626	nan	0.3573	0.9907	nan	0.2973	0.9620
0.2521	113.33	340	0.3013	0.6300	0.6714	0.9633	nan	0.3513	0.9916	nan	0.2974	0.9627
0.2663	116.67	350	0.3060	0.6298	0.6766	0.9621	nan	0.3634	0.9899	nan	0.2981	0.9615
0.2507	120.0	360	0.3032	0.6301	0.6710	0.9634	nan	0.3502	0.9918	nan	0.2973	0.9628

Framework versions

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