safety-utcustom-train-SF-RGB-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: 0.3221
• Mean Iou: 0.7557
• Mean Accuracy: 0.8092
• Overall Accuracy: 0.9835
• Accuracy Unlabeled: nan
• Accuracy Safe: 0.6240
• Accuracy Unsafe: 0.9945
• Iou Unlabeled: nan
• Iou Safe: 0.5281
• Iou Unsafe: 0.9832

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: 9e-06
• 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 Safe	Accuracy Unsafe	Iou Unlabeled	Iou Safe	Iou Unsafe
1.2069	1.0	10	1.1287	0.0406	0.3613	0.1117	nan	0.6267	0.0960	0.0	0.0261	0.0958
1.196	2.0	20	1.1408	0.0465	0.3971	0.1274	nan	0.6837	0.1105	0.0	0.0290	0.1104
1.1866	3.0	30	1.1441	0.0662	0.4586	0.1826	nan	0.7519	0.1653	0.0	0.0335	0.1652
1.1701	4.0	40	1.1350	0.1016	0.5469	0.2805	nan	0.8301	0.2638	0.0	0.0410	0.2637
1.1467	5.0	50	1.1285	0.1325	0.6266	0.3646	nan	0.9052	0.3481	0.0	0.0496	0.3481
1.1126	6.0	60	1.0914	0.1933	0.7318	0.5257	nan	0.9508	0.5128	0.0	0.0673	0.5127
1.0735	7.0	70	1.0392	0.2462	0.8075	0.6582	nan	0.9662	0.6489	0.0	0.0900	0.6487
1.0335	8.0	80	1.0015	0.2783	0.8456	0.7301	nan	0.9683	0.7228	0.0	0.1122	0.7226
1.0088	9.0	90	0.9502	0.3061	0.8736	0.7884	nan	0.9643	0.7830	0.0	0.1359	0.7825
0.9993	10.0	100	0.9158	0.3246	0.8886	0.8232	nan	0.9581	0.8191	0.0	0.1556	0.8183
0.9114	11.0	110	0.8472	0.3562	0.9061	0.8732	nan	0.9411	0.8711	0.0	0.1990	0.8697
0.9027	12.0	120	0.8073	0.3687	0.9085	0.8909	nan	0.9271	0.8898	0.0	0.2182	0.8881
0.8775	13.0	130	0.7756	0.3819	0.9011	0.9086	nan	0.8931	0.9090	0.0	0.2394	0.9062
0.8532	14.0	140	0.7544	0.3883	0.9005	0.9156	nan	0.8844	0.9166	0.0	0.2513	0.9135
0.7509	15.0	150	0.7137	0.4039	0.8965	0.9311	nan	0.8597	0.9333	0.0	0.2824	0.9294
0.7711	16.0	160	0.6837	0.4131	0.8959	0.9394	nan	0.8497	0.9422	0.0	0.3014	0.9379
0.7163	17.0	170	0.6573	0.4230	0.8859	0.9467	nan	0.8212	0.9505	0.0	0.3234	0.9454
0.6609	18.0	180	0.6698	0.4200	0.8889	0.9449	nan	0.8294	0.9484	0.0	0.3163	0.9436
0.7237	19.0	190	0.6465	0.4236	0.8821	0.9479	nan	0.8121	0.9520	0.0	0.3241	0.9467
0.6264	20.0	200	0.6300	0.4293	0.8776	0.9520	nan	0.7985	0.9566	0.0	0.3372	0.9508
0.6711	21.0	210	0.6050	0.4391	0.8731	0.9576	nan	0.7833	0.9630	0.0	0.3605	0.9567
0.626	22.0	220	0.5855	0.4409	0.8742	0.9585	nan	0.7846	0.9637	0.0	0.3653	0.9575
0.6103	23.0	230	0.5651	0.4474	0.8671	0.9623	nan	0.7658	0.9683	0.0	0.3807	0.9615
0.6462	24.0	240	0.5621	0.4489	0.8643	0.9631	nan	0.7592	0.9693	0.0	0.3844	0.9623
0.5442	25.0	250	0.5460	0.4563	0.8592	0.9668	nan	0.7450	0.9735	0.0	0.4028	0.9660
0.6764	26.0	260	0.5673	0.4544	0.8646	0.9657	nan	0.7571	0.9721	0.0	0.3983	0.9650
0.6471	27.0	270	0.5412	0.4586	0.8561	0.9679	nan	0.7374	0.9749	0.0	0.4087	0.9672
0.5589	28.0	280	0.5427	0.4573	0.8601	0.9671	nan	0.7465	0.9738	0.0	0.4057	0.9663
0.6512	29.0	290	0.5264	0.4600	0.8567	0.9681	nan	0.7384	0.9751	0.0	0.4126	0.9674
0.6146	30.0	300	0.5321	0.4616	0.8619	0.9688	nan	0.7482	0.9755	0.0	0.4167	0.9681
0.4938	31.0	310	0.5025	0.4751	0.8475	0.9744	nan	0.7127	0.9823	0.0	0.4515	0.9738
0.4868	32.0	320	0.4836	0.4781	0.8342	0.9761	nan	0.6833	0.9851	0.0	0.4586	0.9756
0.6315	33.0	330	0.4918	0.4739	0.8479	0.9740	nan	0.7139	0.9819	0.0	0.4483	0.9735
0.5529	34.0	340	0.4879	0.4753	0.8414	0.9749	nan	0.6995	0.9832	0.0	0.4516	0.9743
0.4592	35.0	350	0.4826	0.4764	0.8364	0.9754	nan	0.6887	0.9842	0.0	0.4542	0.9749
0.5904	36.0	360	0.4611	0.4859	0.8159	0.9793	nan	0.6423	0.9896	0.0	0.4789	0.9789
0.4804	37.0	370	0.4654	0.4796	0.8359	0.9764	nan	0.6865	0.9853	0.0	0.4627	0.9760
0.4701	38.0	380	0.4625	0.4846	0.8251	0.9784	nan	0.6623	0.9880	0.0	0.4758	0.9779
0.4729	39.0	390	0.4536	0.4838	0.8231	0.9783	nan	0.6582	0.9881	0.0	0.4736	0.9779
0.4219	40.0	400	0.4514	0.4838	0.8305	0.9779	nan	0.6738	0.9872	0.0	0.4739	0.9775
0.6494	41.0	410	0.4425	0.4892	0.8162	0.9801	nan	0.6420	0.9904	0.0	0.4878	0.9797
0.4616	42.0	420	0.4390	0.7316	0.8225	0.9794	nan	0.6558	0.9892	nan	0.4842	0.9790
0.4408	43.0	430	0.4419	0.7358	0.8272	0.9797	nan	0.6652	0.9893	nan	0.4923	0.9793
0.4532	44.0	440	0.4371	0.7375	0.8274	0.9800	nan	0.6651	0.9896	nan	0.4954	0.9796
0.5015	45.0	450	0.4376	0.7364	0.8276	0.9798	nan	0.6659	0.9894	nan	0.4933	0.9794
0.4965	46.0	460	0.4201	0.7405	0.8137	0.9812	nan	0.6357	0.9918	nan	0.5002	0.9809
0.4837	47.0	470	0.4281	0.7378	0.8279	0.9800	nan	0.6662	0.9896	nan	0.4961	0.9796
0.4221	48.0	480	0.4288	0.7371	0.8227	0.9802	nan	0.6553	0.9901	nan	0.4944	0.9798
0.4491	49.0	490	0.4152	0.7371	0.8074	0.9811	nan	0.6228	0.9920	nan	0.4935	0.9808
0.4121	50.0	500	0.4159	0.7367	0.8063	0.9811	nan	0.6205	0.9921	nan	0.4927	0.9808
0.4727	51.0	510	0.4199	0.7354	0.8095	0.9807	nan	0.6274	0.9915	nan	0.4905	0.9804
0.5323	52.0	520	0.4079	0.7383	0.8074	0.9813	nan	0.6227	0.9922	nan	0.4957	0.9809
0.409	53.0	530	0.4103	0.7392	0.8161	0.9809	nan	0.6409	0.9913	nan	0.4978	0.9805
0.6391	54.0	540	0.4063	0.7406	0.8133	0.9813	nan	0.6349	0.9918	nan	0.5003	0.9809
0.3905	55.0	550	0.4000	0.7409	0.8122	0.9814	nan	0.6325	0.9920	nan	0.5007	0.9810
0.4138	56.0	560	0.4028	0.7398	0.8183	0.9809	nan	0.6455	0.9911	nan	0.4990	0.9805
0.3977	57.0	570	0.3865	0.7372	0.7912	0.9821	nan	0.5884	0.9941	nan	0.4926	0.9818
0.4186	58.0	580	0.3845	0.7416	0.7994	0.9822	nan	0.6050	0.9937	nan	0.5014	0.9819
0.4921	59.0	590	0.3881	0.7427	0.8102	0.9817	nan	0.6278	0.9925	nan	0.5039	0.9814
0.3953	60.0	600	0.3823	0.7429	0.8027	0.9822	nan	0.6119	0.9935	nan	0.5039	0.9819
0.4263	61.0	610	0.3841	0.7420	0.8075	0.9818	nan	0.6222	0.9928	nan	0.5026	0.9815
0.3798	62.0	620	0.3763	0.7446	0.8054	0.9823	nan	0.6174	0.9934	nan	0.5072	0.9820
0.4208	63.0	630	0.3724	0.7437	0.7919	0.9829	nan	0.5888	0.9949	nan	0.5047	0.9826
0.3627	64.0	640	0.3760	0.7466	0.8111	0.9822	nan	0.6292	0.9930	nan	0.5112	0.9819
0.4156	65.0	650	0.3669	0.7478	0.8018	0.9829	nan	0.6092	0.9943	nan	0.5130	0.9826
0.468	66.0	660	0.3706	0.7508	0.8145	0.9826	nan	0.6359	0.9932	nan	0.5193	0.9823
0.4547	67.0	670	0.3692	0.7512	0.8189	0.9824	nan	0.6451	0.9927	nan	0.5204	0.9821
0.3604	68.0	680	0.3691	0.7520	0.8152	0.9827	nan	0.6371	0.9933	nan	0.5215	0.9824
0.4476	69.0	690	0.3679	0.7516	0.8195	0.9825	nan	0.6463	0.9927	nan	0.5210	0.9821
0.3535	70.0	700	0.3589	0.7522	0.8097	0.9831	nan	0.6255	0.9939	nan	0.5217	0.9827
0.3539	71.0	710	0.3572	0.7526	0.8091	0.9831	nan	0.6242	0.9941	nan	0.5224	0.9828
0.3675	72.0	720	0.3589	0.7518	0.8100	0.9830	nan	0.6261	0.9939	nan	0.5209	0.9827
0.4148	73.0	730	0.3536	0.7504	0.8093	0.9828	nan	0.6249	0.9937	nan	0.5182	0.9825
0.3941	74.0	740	0.3538	0.7497	0.8099	0.9827	nan	0.6263	0.9936	nan	0.5169	0.9824
0.4264	75.0	750	0.3595	0.7469	0.8197	0.9818	nan	0.6473	0.9920	nan	0.5123	0.9814
0.3815	76.0	760	0.3525	0.7492	0.8097	0.9827	nan	0.6258	0.9935	nan	0.5162	0.9823
0.3459	77.0	770	0.3443	0.7452	0.7926	0.9831	nan	0.5901	0.9951	nan	0.5076	0.9828
0.3794	78.0	780	0.3538	0.7501	0.8154	0.9825	nan	0.6377	0.9930	nan	0.5180	0.9821
0.3761	79.0	790	0.3525	0.7483	0.8169	0.9821	nan	0.6412	0.9925	nan	0.5147	0.9818
0.3612	80.0	800	0.3495	0.7513	0.8128	0.9828	nan	0.6321	0.9934	nan	0.5201	0.9824
0.405	81.0	810	0.3466	0.7502	0.8148	0.9825	nan	0.6365	0.9931	nan	0.5182	0.9822
0.4289	82.0	820	0.3458	0.7498	0.8092	0.9828	nan	0.6247	0.9937	nan	0.5171	0.9824
0.3523	83.0	830	0.3435	0.7503	0.8112	0.9827	nan	0.6288	0.9935	nan	0.5183	0.9824
0.4254	84.0	840	0.3403	0.7495	0.8000	0.9832	nan	0.6052	0.9947	nan	0.5160	0.9829
0.3399	85.0	850	0.3355	0.7492	0.8003	0.9832	nan	0.6059	0.9947	nan	0.5155	0.9829
0.3251	86.0	860	0.3395	0.7503	0.8028	0.9832	nan	0.6111	0.9945	nan	0.5178	0.9829
0.3748	87.0	870	0.3400	0.7502	0.8117	0.9827	nan	0.6299	0.9934	nan	0.5181	0.9824
0.4398	88.0	880	0.3450	0.7527	0.8197	0.9826	nan	0.6466	0.9928	nan	0.5231	0.9822
0.3782	89.0	890	0.3454	0.7547	0.8180	0.9829	nan	0.6426	0.9933	nan	0.5268	0.9826
0.4318	90.0	900	0.3424	0.7541	0.8162	0.9830	nan	0.6390	0.9934	nan	0.5255	0.9826
0.3428	91.0	910	0.3327	0.7541	0.8124	0.9832	nan	0.6309	0.9939	nan	0.5253	0.9828
0.4303	92.0	920	0.3364	0.7525	0.8108	0.9830	nan	0.6277	0.9939	nan	0.5223	0.9827
0.3624	93.0	930	0.3277	0.7531	0.8063	0.9834	nan	0.6182	0.9945	nan	0.5231	0.9830
0.3418	94.0	940	0.3315	0.7548	0.8125	0.9833	nan	0.6311	0.9940	nan	0.5267	0.9829
0.321	95.0	950	0.3266	0.7541	0.8070	0.9835	nan	0.6195	0.9945	nan	0.5251	0.9831
0.3152	96.0	960	0.3265	0.7531	0.8025	0.9836	nan	0.6101	0.9949	nan	0.5230	0.9833
0.3153	97.0	970	0.3263	0.7537	0.8048	0.9835	nan	0.6149	0.9947	nan	0.5243	0.9832
0.3158	98.0	980	0.3299	0.7553	0.8139	0.9832	nan	0.6340	0.9939	nan	0.5278	0.9829
0.3162	99.0	990	0.3248	0.7546	0.8076	0.9835	nan	0.6207	0.9945	nan	0.5260	0.9832
0.3748	100.0	1000	0.3238	0.7553	0.8077	0.9836	nan	0.6208	0.9946	nan	0.5274	0.9833
0.3598	101.0	1010	0.3221	0.7544	0.8096	0.9833	nan	0.6250	0.9943	nan	0.5257	0.9830
0.3245	102.0	1020	0.3247	0.7527	0.8156	0.9828	nan	0.6380	0.9933	nan	0.5228	0.9825
0.3527	103.0	1030	0.3275	0.7537	0.8193	0.9827	nan	0.6456	0.9930	nan	0.5250	0.9824
0.5087	104.0	1040	0.3221	0.7559	0.8105	0.9835	nan	0.6266	0.9944	nan	0.5287	0.9832
0.3331	105.0	1050	0.3183	0.7560	0.8064	0.9837	nan	0.6180	0.9948	nan	0.5285	0.9834
0.324	106.0	1060	0.3198	0.7561	0.8090	0.9836	nan	0.6235	0.9946	nan	0.5289	0.9833
0.3512	107.0	1070	0.3194	0.7549	0.8052	0.9836	nan	0.6155	0.9949	nan	0.5265	0.9833
0.3274	108.0	1080	0.3185	0.7569	0.8122	0.9835	nan	0.6301	0.9943	nan	0.5306	0.9832
0.335	109.0	1090	0.3177	0.7554	0.8081	0.9836	nan	0.6217	0.9946	nan	0.5276	0.9832
0.3581	110.0	1100	0.3204	0.7568	0.8146	0.9834	nan	0.6352	0.9940	nan	0.5306	0.9831
0.3307	111.0	1110	0.3216	0.7571	0.8138	0.9835	nan	0.6335	0.9941	nan	0.5310	0.9832
0.3162	112.0	1120	0.3227	0.7575	0.8181	0.9833	nan	0.6425	0.9937	nan	0.5320	0.9830
0.3687	113.0	1130	0.3188	0.7567	0.8124	0.9835	nan	0.6306	0.9942	nan	0.5302	0.9832
0.4099	114.0	1140	0.3151	0.7550	0.8063	0.9836	nan	0.6178	0.9947	nan	0.5266	0.9833
0.3283	115.0	1150	0.3152	0.7557	0.8088	0.9836	nan	0.6232	0.9945	nan	0.5281	0.9832
0.3118	116.0	1160	0.3180	0.7556	0.8097	0.9835	nan	0.6249	0.9944	nan	0.5280	0.9832
0.3233	117.0	1170	0.3164	0.7551	0.8070	0.9836	nan	0.6192	0.9947	nan	0.5269	0.9833
0.3401	118.0	1180	0.3192	0.7562	0.8122	0.9834	nan	0.6303	0.9942	nan	0.5292	0.9831
0.3867	119.0	1190	0.3199	0.7566	0.8160	0.9833	nan	0.6382	0.9938	nan	0.5302	0.9830
0.3217	120.0	1200	0.3221	0.7557	0.8092	0.9835	nan	0.6240	0.9945	nan	0.5281	0.9832

Framework versions

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