safety-utcustom-train-SF-RGBD-b0

This model is a fine-tuned version of nvidia/mit-b0 on the sam1120/safety-utcustom-TRAIN dataset. It achieves the following results on the evaluation set:

• Loss: 0.1043
• Mean Iou: 0.7188
• Mean Accuracy: 0.7607
• Overall Accuracy: 0.9815
• Accuracy Unlabeled: nan
• Accuracy Safe: 0.5261
• Accuracy Unsafe: 0.9953
• Iou Unlabeled: nan
• Iou Safe: 0.4564
• Iou Unsafe: 0.9812

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: 130

Training results

Training Loss	Epoch	Step	Accuracy Safe	Accuracy Unlabeled	Accuracy Unsafe	Iou Safe	Iou Unlabeled	Iou Unsafe	Validation Loss	Mean Accuracy	Mean Iou	Overall Accuracy
1.0084	1.0	10	0.0368	nan	0.7845	0.0163	0.0	0.7666	1.0688	0.4107	0.2610	0.7625
0.8483	2.0	20	0.0002	nan	0.9980	0.0002	0.0	0.9687	0.8740	0.4991	0.3230	0.9686
0.7058	3.0	30	0.0009	nan	0.9930	0.0009	0.0	0.9641	0.7416	0.4969	0.3217	0.9637
0.578	4.0	40	0.0007	nan	0.9953	0.0007	0.0	0.9662	0.5969	0.4980	0.3223	0.9659
0.5531	5.0	50	0.0061	nan	0.9974	0.0059	0.0	0.9682	0.5068	0.5018	0.3247	0.9681
0.4786	6.0	60	0.0097	nan	0.9961	0.0092	0.0	0.9671	0.4575	0.5029	0.3254	0.9670
0.4681	7.0	70	0.0067	nan	0.9983	0.0064	0.0	0.9690	0.4382	0.5025	0.3251	0.9690
0.4139	8.0	80	0.0017	nan	0.9980	0.0016	0.0	0.9686	0.3973	0.4998	0.3234	0.9686
0.4275	9.0	90	0.0077	nan	0.9994	0.0076	nan	0.9701	0.3983	0.5036	0.4888	0.9701
0.3975	10.0	100	0.0008	nan	0.9998	0.0008	0.0	0.9702	0.3398	0.5003	0.3237	0.9702
0.4325	11.0	110	0.0941	nan	0.9993	0.0919	0.0	0.9725	0.3785	0.5467	0.3548	0.9725
0.3239	12.0	120	0.0772	nan	0.9995	0.0759	0.0	0.9722	0.3338	0.5383	0.3493	0.9722
0.3733	13.0	130	0.0763	nan	0.9995	0.0751	nan	0.9722	0.3013	0.5379	0.5236	0.9722
0.3165	14.0	140	0.0800	nan	0.9994	0.0786	nan	0.9722	0.2849	0.5397	0.5254	0.9723
0.3329	15.0	150	0.1118	nan	0.9990	0.1083	nan	0.9727	0.3002	0.5554	0.5405	0.9728
0.3214	16.0	160	0.0908	nan	0.9995	0.0892	nan	0.9726	0.2725	0.5451	0.5309	0.9726
0.2744	17.0	170	0.1573	nan	0.9986	0.1503	nan	0.9736	0.2896	0.5780	0.5620	0.9737
0.2948	18.0	180	0.1330	nan	0.9989	0.1282	nan	0.9732	0.2564	0.5659	0.5507	0.9733
0.2653	19.0	190	0.1732	nan	0.9987	0.1660	nan	0.9742	0.2518	0.5860	0.5701	0.9743
0.3026	20.0	200	0.1408	nan	0.9990	0.1364	nan	0.9735	0.2531	0.5699	0.5550	0.9737
0.2649	21.0	210	0.1802	nan	0.9986	0.1722	nan	0.9743	0.2384	0.5894	0.5732	0.9744
0.2431	22.0	220	0.1993	nan	0.9983	0.1890	nan	0.9746	0.2390	0.5988	0.5818	0.9747
0.2608	23.0	230	0.2317	nan	0.9981	0.2181	nan	0.9753	0.2355	0.6149	0.5967	0.9755
0.223	24.0	240	0.1697	nan	0.9989	0.1637	nan	0.9743	0.2290	0.5843	0.5690	0.9744
0.2448	25.0	250	0.2141	nan	0.9985	0.2037	nan	0.9751	0.2262	0.6063	0.5894	0.9753
0.2547	26.0	260	0.2737	nan	0.9978	0.2555	nan	0.9763	0.2281	0.6357	0.6159	0.9764
0.2266	27.0	270	0.2391	nan	0.9981	0.2252	nan	0.9755	0.2191	0.6186	0.6004	0.9757
0.2357	28.0	280	0.2227	nan	0.9985	0.2122	nan	0.9754	0.2218	0.6106	0.5938	0.9756
0.2563	29.0	290	0.1852	nan	0.9988	0.1782	nan	0.9746	0.2096	0.5920	0.5764	0.9748
0.226	30.0	300	0.2844	nan	0.9977	0.2643	nan	0.9764	0.2121	0.6410	0.6203	0.9766
0.2221	31.0	310	0.2718	nan	0.9978	0.2533	nan	0.9761	0.2016	0.6348	0.6147	0.9763
0.2317	32.0	320	0.2649	nan	0.9982	0.2499	nan	0.9763	0.2008	0.6315	0.6131	0.9765
0.2643	33.0	330	0.3254	nan	0.9976	0.3014	nan	0.9775	0.1989	0.6615	0.6394	0.9777
0.2118	34.0	340	0.3347	nan	0.9977	0.3117	nan	0.9779	0.1901	0.6662	0.6448	0.9782
0.2133	35.0	350	0.3619	nan	0.9976	0.3350	nan	0.9785	0.1917	0.6797	0.6568	0.9788
0.2064	36.0	360	0.3401	nan	0.9978	0.3174	nan	0.9782	0.1860	0.6690	0.6478	0.9784
0.2341	37.0	370	0.2704	nan	0.9983	0.2557	nan	0.9766	0.1775	0.6343	0.6162	0.9768
0.2093	38.0	380	0.3552	nan	0.9928	0.2874	nan	0.9737	0.1934	0.6740	0.6306	0.9740
0.1958	39.0	390	0.3001	nan	0.9980	0.2818	nan	0.9772	0.1755	0.6491	0.6295	0.9774
0.1886	40.0	400	0.3881	nan	0.9969	0.3522	nan	0.9787	0.1768	0.6925	0.6654	0.9789
0.1734	41.0	410	0.3948	nan	0.9973	0.3626	nan	0.9793	0.1745	0.6960	0.6709	0.9795
0.1795	42.0	420	0.4168	nan	0.9970	0.3789	nan	0.9796	0.1710	0.7069	0.6793	0.9798
0.222	43.0	430	0.4041	nan	0.9972	0.3700	nan	0.9794	0.1706	0.7007	0.6747	0.9797
0.1831	44.0	440	0.4044	nan	0.9972	0.3708	nan	0.9795	0.1687	0.7008	0.6752	0.9797
0.1935	45.0	450	0.4347	nan	0.9964	0.3889	nan	0.9796	0.1711	0.7155	0.6842	0.9798
0.1728	46.0	460	0.4208	nan	0.9969	0.3819	nan	0.9796	0.1714	0.7088	0.6808	0.9799
0.1742	47.0	470	0.3898	nan	0.9974	0.3590	nan	0.9792	0.1670	0.6936	0.6691	0.9794
0.2064	48.0	480	0.4209	nan	0.9970	0.3827	nan	0.9797	0.1683	0.7089	0.6812	0.9799
0.1946	49.0	490	0.3746	nan	0.9976	0.3471	nan	0.9790	0.1659	0.6861	0.6630	0.9792
0.1836	50.0	500	0.4487	nan	0.9965	0.4020	nan	0.9800	0.1618	0.7226	0.6910	0.9803
0.1786	51.0	510	0.4327	nan	0.9966	0.3896	nan	0.9797	0.1595	0.7147	0.6846	0.9800
0.1867	52.0	520	0.4540	nan	0.9966	0.4083	nan	0.9803	0.1555	0.7253	0.6943	0.9806
0.1824	53.0	530	0.4386	nan	0.9966	0.3942	nan	0.9798	0.1564	0.7176	0.6870	0.9801
0.1494	54.0	540	0.4920	nan	0.9956	0.4299	nan	0.9804	0.1540	0.7438	0.7052	0.9807
0.1583	55.0	550	0.4558	nan	0.9964	0.4075	nan	0.9802	0.1502	0.7261	0.6939	0.9804
0.1648	56.0	560	0.4791	nan	0.9958	0.4208	nan	0.9802	0.1523	0.7374	0.7005	0.9805
0.1993	57.0	570	0.4586	nan	0.9964	0.4103	nan	0.9803	0.1502	0.7275	0.6953	0.9805
0.2243	58.0	580	0.3920	nan	0.9973	0.3599	nan	0.9792	0.1474	0.6946	0.6695	0.9794
0.1551	59.0	590	0.4687	nan	0.9961	0.4157	nan	0.9803	0.1445	0.7324	0.6980	0.9805
0.1666	60.0	600	0.4460	nan	0.9964	0.3986	nan	0.9799	0.1444	0.7212	0.6892	0.9801
0.1632	61.0	610	0.5120	nan	0.9951	0.4411	nan	0.9805	0.1504	0.7535	0.7108	0.9808
0.1589	62.0	620	0.4059	nan	0.9971	0.3704	nan	0.9794	0.1430	0.7015	0.6749	0.9796
0.1454	63.0	630	0.4835	nan	0.9959	0.4260	nan	0.9805	0.1423	0.7397	0.7032	0.9808
0.1635	64.0	640	0.4902	nan	0.9957	0.4299	nan	0.9805	0.1424	0.7430	0.7052	0.9808
0.1515	65.0	650	0.4775	nan	0.9962	0.4239	nan	0.9806	0.1422	0.7368	0.7022	0.9808
0.151	66.0	660	0.4718	nan	0.9962	0.4195	nan	0.9804	0.1423	0.7340	0.7000	0.9807
0.166	67.0	670	0.4721	nan	0.9963	0.4208	nan	0.9805	0.1427	0.7342	0.7007	0.9808
0.1561	68.0	680	0.4916	nan	0.9959	0.4332	nan	0.9807	0.1420	0.7437	0.7070	0.9810
0.1501	69.0	690	0.4906	nan	0.9958	0.4311	nan	0.9806	0.1437	0.7432	0.7058	0.9809
0.1598	70.0	700	0.3445	nan	0.9977	0.3204	nan	0.9782	0.1379	0.6711	0.6493	0.9784
0.1431	71.0	710	0.4898	nan	0.9960	0.4325	nan	0.9807	0.1400	0.7429	0.7066	0.9810
0.164	72.0	720	0.4698	nan	0.9964	0.4196	nan	0.9805	0.1347	0.7331	0.7001	0.9808
0.1555	73.0	730	0.5271	nan	0.9937	0.4364	nan	0.9796	0.1368	0.7604	0.7080	0.9799
0.1924	74.0	740	0.4638	nan	0.9965	0.4159	nan	0.9805	0.1312	0.7301	0.6982	0.9808
0.1612	75.0	750	0.5052	nan	0.9956	0.4409	nan	0.9808	0.1340	0.7504	0.7108	0.9811
0.1234	76.0	760	0.5301	nan	0.9946	0.4501	nan	0.9806	0.1354	0.7624	0.7153	0.9809
0.1679	77.0	770	0.4644	nan	0.9964	0.4156	nan	0.9804	0.1323	0.7304	0.6980	0.9807
0.1375	78.0	780	0.4804	nan	0.9961	0.4263	nan	0.9806	0.1355	0.7383	0.7035	0.9809
0.1839	79.0	790	0.5070	nan	0.9955	0.4422	nan	0.9808	0.1319	0.7512	0.7115	0.9811
0.155	80.0	800	0.4846	nan	0.9961	0.4295	nan	0.9807	0.1298	0.7403	0.7051	0.9810
0.1219	81.0	810	0.4671	nan	0.9963	0.4167	nan	0.9804	0.1302	0.7317	0.6986	0.9807
0.1218	82.0	820	0.4864	nan	0.9960	0.4300	nan	0.9807	0.1313	0.7412	0.7054	0.9810
0.138	83.0	830	0.5097	nan	0.9955	0.4445	nan	0.9809	0.1318	0.7526	0.7127	0.9812
0.1399	84.0	840	0.5067	nan	0.9957	0.4441	nan	0.9810	0.1290	0.7512	0.7126	0.9813
0.1455	85.0	850	0.5024	nan	0.9957	0.4404	nan	0.9809	0.1277	0.7491	0.7106	0.9811
0.1466	86.0	860	0.4920	nan	0.9959	0.4341	nan	0.9808	0.1243	0.7440	0.7074	0.9811
0.1769	87.0	870	0.5737	nan	0.9924	0.4592	nan	0.9797	0.1317	0.7831	0.7194	0.9800
0.1453	88.0	880	0.3341	nan	0.9978	0.3115	nan	0.9780	0.1254	0.6659	0.6447	0.9782
0.133	89.0	890	0.5257	nan	0.9950	0.4518	nan	0.9809	0.1283	0.7603	0.7163	0.9812
0.1288	90.0	900	0.5049	nan	0.9957	0.4420	nan	0.9809	0.1221	0.7503	0.7115	0.9812
0.1318	91.0	910	0.4838	nan	0.9961	0.4290	nan	0.9807	0.1219	0.7400	0.7049	0.9810
0.1211	92.0	920	0.5355	nan	0.9950	0.4596	nan	0.9811	0.1242	0.7652	0.7203	0.9814
0.1137	93.0	930	0.5135	nan	0.9958	0.4517	nan	0.9813	0.1181	0.7547	0.7165	0.9816
0.1312	94.0	940	0.4775	nan	0.9963	0.4262	nan	0.9807	0.1199	0.7369	0.7035	0.9810
0.1591	95.0	950	0.5115	nan	0.9956	0.4473	nan	0.9810	0.1182	0.7536	0.7142	0.9813
0.1207	96.0	960	0.5206	nan	0.9956	0.4544	nan	0.9812	0.1156	0.7581	0.7178	0.9815
0.1203	97.0	970	0.5054	nan	0.9958	0.4439	nan	0.9810	0.1165	0.7506	0.7124	0.9813
0.1196	98.0	980	0.5296	nan	0.9953	0.4585	nan	0.9812	0.1131	0.7624	0.7199	0.9815
0.1304	99.0	990	0.5269	nan	0.9953	0.4568	nan	0.9812	0.1155	0.7611	0.7190	0.9815
0.1058	100.0	1000	0.5163	nan	0.9955	0.4496	nan	0.9810	0.1144	0.7559	0.7153	0.9813
0.1135	101.0	1010	0.4934	nan	0.9961	0.4368	nan	0.9809	0.1113	0.7447	0.7089	0.9812
0.1116	102.0	1020	0.5878	nan	0.9932	0.4799	nan	0.9808	0.1128	0.7905	0.7304	0.9812
0.1036	103.0	1030	0.4826	nan	0.9963	0.4304	nan	0.9809	0.1078	0.7394	0.7056	0.9811
0.1195	104.0	1040	0.4364	nan	0.9966	0.3930	nan	0.9798	0.1110	0.7165	0.6864	0.9801
0.1205	105.0	1050	0.5793	nan	0.9934	0.4762	nan	0.9808	0.1120	0.7864	0.7285	0.9812
0.1453	106.0	1060	0.4707	nan	0.9964	0.4205	nan	0.9806	0.1110	0.7336	0.7005	0.9808
0.0965	107.0	1070	0.5638	nan	0.9941	0.4723	nan	0.9811	0.1091	0.7789	0.7267	0.9814
0.1058	108.0	1080	0.4881	nan	0.9962	0.4337	nan	0.9809	0.1085	0.7422	0.7073	0.9812
0.1163	109.0	1090	0.5128	nan	0.9957	0.4493	nan	0.9811	0.1077	0.7542	0.7152	0.9814
0.1145	110.0	1100	0.5228	nan	0.9954	0.4547	nan	0.9812	0.1081	0.7591	0.7179	0.9815
0.1031	111.0	1110	0.5522	nan	0.9945	0.4673	nan	0.9811	0.1073	0.7733	0.7242	0.9814
0.1042	112.0	1120	0.5490	nan	0.9947	0.4669	nan	0.9812	0.1064	0.7718	0.7241	0.9815
0.1119	113.0	1130	0.5064	nan	0.9958	0.4449	nan	0.9811	0.1063	0.7511	0.7130	0.9813
0.1116	114.0	1140	0.5172	nan	0.9956	0.4520	nan	0.9812	0.1074	0.7564	0.7166	0.9815
0.1063	115.0	1150	0.5163	nan	0.9956	0.4511	nan	0.9812	0.1072	0.7560	0.7161	0.9814
0.1054	116.0	1160	0.4994	nan	0.9960	0.4408	nan	0.9810	0.1065	0.7477	0.7109	0.9813
0.1613	117.0	1170	0.5251	nan	0.9955	0.4576	nan	0.9813	0.1060	0.7603	0.7195	0.9816
0.1542	118.0	1180	0.5454	nan	0.9947	0.4649	nan	0.9812	0.1058	0.7701	0.7230	0.9815
0.1226	119.0	1190	0.5469	nan	0.9947	0.4658	nan	0.9812	0.1064	0.7708	0.7235	0.9815
0.1295	120.0	1200	0.5437	nan	0.9948	0.4646	nan	0.9812	0.1060	0.7693	0.7229	0.9815
0.1438	121.0	1210	0.1076	0.7084	0.7435	0.9812	nan	0.4909	0.9962	nan	0.4358	0.9809
0.1391	122.0	1220	0.1081	0.7221	0.7683	0.9814	nan	0.5417	0.9948	nan	0.4630	0.9811
0.1756	123.0	1230	0.1041	0.7233	0.7710	0.9814	nan	0.5473	0.9947	nan	0.4655	0.9811
0.1174	124.0	1240	0.1029	0.7189	0.7614	0.9815	nan	0.5275	0.9953	nan	0.4566	0.9812
0.1025	125.0	1250	0.1043	0.7100	0.7470	0.9812	nan	0.4980	0.9959	nan	0.4391	0.9809
0.0997	126.0	1260	0.1038	0.7211	0.7638	0.9816	nan	0.5323	0.9953	nan	0.4609	0.9813
0.1768	127.0	1270	0.1037	0.7204	0.7617	0.9817	nan	0.5279	0.9955	nan	0.4594	0.9814
0.1527	128.0	1280	0.1027	0.7167	0.7564	0.9815	nan	0.5171	0.9956	nan	0.4522	0.9812
0.1269	129.0	1290	0.1041	0.7178	0.7583	0.9815	nan	0.5211	0.9955	nan	0.4543	0.9812
0.0968	130.0	1300	0.1043	0.7188	0.7607	0.9815	nan	0.5261	0.9953	nan	0.4564	0.9812

Framework versions

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