dungeon-maps-seg-v0.0.1

This model is a fine-tuned version of nvidia/mit-b0 on the cephelos/dungeon-maps-seg dataset. It achieves the following results on the evaluation set:

• Loss: 0.0361
• Mean Iou: 0.9518
• Mean Accuracy: 0.9783
• Overall Accuracy: 0.9893
• Accuracy Unlabeled: nan
• Accuracy Room: 0.9923
• Accuracy Wall: 0.9490
• Accuracy Outside: 0.9935
• Iou Unlabeled: nan
• Iou Room: 0.9857
• Iou Wall: 0.8788
• Iou Outside: 0.9911

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: 6e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Unlabeled	Accuracy Room	Accuracy Wall	Accuracy Outside	Iou Unlabeled	Iou Room	Iou Wall	Iou Outside
0.2922	0.7692	20	0.2745	0.8581	0.9561	0.9598	nan	0.9526	0.9466	0.9690	nan	0.9466	0.6646	0.9632
0.2099	1.5385	40	0.2072	0.8639	0.9584	0.9625	nan	0.9680	0.9472	0.9599	nan	0.9600	0.6732	0.9584
0.2009	2.3077	60	0.1688	0.8968	0.9623	0.9741	nan	0.9718	0.9316	0.9835	nan	0.9649	0.7477	0.9778
0.1258	3.0769	80	0.1482	0.8991	0.9676	0.9745	nan	0.9773	0.9492	0.9762	nan	0.9708	0.7529	0.9736
0.1624	3.8462	100	0.1333	0.9115	0.9682	0.9785	nan	0.9807	0.9410	0.9829	nan	0.9734	0.7817	0.9795
0.1098	4.6154	120	0.1079	0.9173	0.9624	0.9805	nan	0.9859	0.9145	0.9868	nan	0.9753	0.7950	0.9817
0.1629	5.3846	140	0.1041	0.9195	0.9711	0.9806	nan	0.9790	0.9462	0.9881	nan	0.9738	0.8013	0.9833
0.1243	6.1538	160	0.0872	0.9243	0.9675	0.9821	nan	0.9852	0.9288	0.9884	nan	0.9766	0.8125	0.9836
0.0974	6.9231	180	0.0996	0.9217	0.9731	0.9811	nan	0.9754	0.9525	0.9915	nan	0.9717	0.8073	0.9861
0.0861	7.6923	200	0.0798	0.9248	0.9706	0.9821	nan	0.9829	0.9403	0.9886	nan	0.9764	0.8142	0.9836
0.0928	8.4615	220	0.0718	0.9276	0.9740	0.9828	nan	0.9830	0.9507	0.9882	nan	0.9773	0.8209	0.9847
0.0583	9.2308	240	0.0726	0.9240	0.9686	0.9822	nan	0.9870	0.9326	0.9862	nan	0.9789	0.8111	0.9821
0.0886	10.0	260	0.0700	0.9296	0.9740	0.9835	nan	0.9845	0.9491	0.9885	nan	0.9786	0.8250	0.9852
0.1133	10.7692	280	0.0651	0.9322	0.9633	0.9848	nan	0.9912	0.9064	0.9922	nan	0.9794	0.8301	0.9872
0.0821	11.5385	300	0.0616	0.9302	0.9721	0.9836	nan	0.9833	0.9417	0.9912	nan	0.9779	0.8270	0.9857
0.07	12.3077	320	0.0586	0.9394	0.9690	0.9864	nan	0.9896	0.9232	0.9942	nan	0.9810	0.8485	0.9887
0.076	13.0769	340	0.0566	0.9349	0.9651	0.9854	nan	0.9919	0.9113	0.9920	nan	0.9803	0.8365	0.9878
0.0577	13.8462	360	0.0570	0.9378	0.9755	0.9857	nan	0.9850	0.9488	0.9926	nan	0.9797	0.8452	0.9886
0.1261	14.6154	380	0.0548	0.9403	0.9739	0.9864	nan	0.9867	0.9410	0.9939	nan	0.9808	0.8511	0.9891
0.0583	15.3846	400	0.0523	0.9428	0.9736	0.9871	nan	0.9895	0.9379	0.9934	nan	0.9820	0.8566	0.9896
0.0602	16.1538	420	0.0488	0.9409	0.9737	0.9866	nan	0.9899	0.9394	0.9917	nan	0.9820	0.8519	0.9887
0.0728	16.9231	440	0.0504	0.9380	0.9716	0.9860	nan	0.9907	0.9335	0.9905	nan	0.9819	0.8448	0.9873
0.0507	17.6923	460	0.0503	0.9378	0.9739	0.9858	nan	0.9892	0.9424	0.9901	nan	0.9820	0.8445	0.9869
0.077	18.4615	480	0.0474	0.9429	0.9740	0.9871	nan	0.9876	0.9396	0.9949	nan	0.9819	0.8570	0.9897
0.2137	19.2308	500	0.0500	0.9413	0.9763	0.9866	nan	0.9892	0.9489	0.9907	nan	0.9823	0.8532	0.9882
0.0991	20.0	520	0.0459	0.9440	0.9719	0.9875	nan	0.9899	0.9309	0.9950	nan	0.9827	0.8595	0.9898
0.0691	20.7692	540	0.0447	0.9451	0.9743	0.9877	nan	0.9906	0.9390	0.9933	nan	0.9831	0.8623	0.9897
0.0602	21.5385	560	0.0447	0.9462	0.9754	0.9879	nan	0.9885	0.9424	0.9952	nan	0.9828	0.8654	0.9904
0.0469	22.3077	580	0.0429	0.9466	0.9767	0.9879	nan	0.9889	0.9471	0.9940	nan	0.9830	0.8664	0.9903
0.0553	23.0769	600	0.0445	0.9468	0.9722	0.9882	nan	0.9913	0.9301	0.9952	nan	0.9832	0.8666	0.9906
0.0671	23.8462	620	0.0424	0.9455	0.9748	0.9878	nan	0.9900	0.9407	0.9938	nan	0.9833	0.8635	0.9898
0.0431	24.6154	640	0.0417	0.9475	0.9732	0.9883	nan	0.9921	0.9331	0.9943	nan	0.9836	0.8681	0.9907
0.0381	25.3846	660	0.0429	0.9449	0.9763	0.9876	nan	0.9881	0.9467	0.9942	nan	0.9827	0.8620	0.9901
0.0503	26.1538	680	0.0403	0.9471	0.9746	0.9882	nan	0.9924	0.9384	0.9929	nan	0.9841	0.8669	0.9902
0.0685	26.9231	700	0.0410	0.9496	0.9743	0.9888	nan	0.9913	0.9361	0.9957	nan	0.9842	0.8732	0.9912
0.0381	27.6923	720	0.0398	0.9494	0.9771	0.9887	nan	0.9906	0.9466	0.9942	nan	0.9843	0.8729	0.9909
0.0587	28.4615	740	0.0397	0.9500	0.9760	0.9889	nan	0.9913	0.9421	0.9947	nan	0.9843	0.8743	0.9913
0.0573	29.2308	760	0.0402	0.9489	0.9756	0.9887	nan	0.9913	0.9411	0.9945	nan	0.9845	0.8715	0.9908
0.0686	30.0	780	0.0386	0.9499	0.9763	0.9889	nan	0.9914	0.9433	0.9944	nan	0.9844	0.8740	0.9912
0.037	30.7692	800	0.0386	0.9503	0.9752	0.9890	nan	0.9925	0.9387	0.9944	nan	0.9849	0.8748	0.9911
0.0565	31.5385	820	0.0389	0.9497	0.9773	0.9888	nan	0.9898	0.9471	0.9950	nan	0.9840	0.8738	0.9913
0.0405	32.3077	840	0.0383	0.9483	0.9743	0.9886	nan	0.9933	0.9366	0.9930	nan	0.9848	0.8698	0.9903
0.0618	33.0769	860	0.0383	0.9497	0.9757	0.9889	nan	0.9920	0.9408	0.9942	nan	0.9847	0.8734	0.9910
0.0398	33.8462	880	0.0379	0.9494	0.9766	0.9888	nan	0.9917	0.9446	0.9936	nan	0.9846	0.8729	0.9908
0.0488	34.6154	900	0.0376	0.9501	0.9769	0.9889	nan	0.9915	0.9450	0.9941	nan	0.9851	0.8745	0.9907
0.0574	35.3846	920	0.0379	0.9512	0.9762	0.9892	nan	0.9914	0.9419	0.9953	nan	0.9849	0.8773	0.9914
0.0331	36.1538	940	0.0368	0.9514	0.9764	0.9893	nan	0.9921	0.9424	0.9947	nan	0.9852	0.8777	0.9913
0.0578	36.9231	960	0.0368	0.9520	0.9770	0.9894	nan	0.9916	0.9443	0.9951	nan	0.9852	0.8790	0.9917
0.0471	37.6923	980	0.0369	0.9517	0.9779	0.9893	nan	0.9912	0.9480	0.9947	nan	0.9852	0.8786	0.9915
0.0388	38.4615	1000	0.0369	0.9511	0.9776	0.9892	nan	0.9904	0.9473	0.9952	nan	0.9846	0.8770	0.9916
0.0455	39.2308	1020	0.0367	0.9517	0.9753	0.9894	nan	0.9928	0.9379	0.9950	nan	0.9853	0.8784	0.9915
0.0359	40.0	1040	0.0360	0.9516	0.9773	0.9893	nan	0.9917	0.9457	0.9945	nan	0.9853	0.8783	0.9913
0.0281	40.7692	1060	0.0363	0.9519	0.9775	0.9894	nan	0.9917	0.9462	0.9946	nan	0.9854	0.8790	0.9913
0.0394	41.5385	1080	0.0367	0.9508	0.9769	0.9891	nan	0.9922	0.9446	0.9939	nan	0.9854	0.8761	0.9909
0.0286	42.3077	1100	0.0360	0.9525	0.9761	0.9896	nan	0.9924	0.9405	0.9953	nan	0.9855	0.8804	0.9917
0.028	43.0769	1120	0.0363	0.9509	0.9791	0.9891	nan	0.9909	0.9530	0.9936	nan	0.9850	0.8767	0.9911
0.0523	43.8462	1140	0.0366	0.9526	0.9777	0.9895	nan	0.9919	0.9466	0.9947	nan	0.9856	0.8806	0.9915
0.0492	44.6154	1160	0.0364	0.9523	0.9764	0.9895	nan	0.9926	0.9419	0.9948	nan	0.9856	0.8799	0.9915
0.0331	45.3846	1180	0.0356	0.9523	0.9781	0.9894	nan	0.9906	0.9484	0.9954	nan	0.9852	0.8799	0.9917
0.0443	46.1538	1200	0.0358	0.9533	0.9772	0.9897	nan	0.9921	0.9443	0.9953	nan	0.9857	0.8824	0.9918
0.0331	46.9231	1220	0.0356	0.9527	0.9771	0.9896	nan	0.9929	0.9441	0.9943	nan	0.9858	0.8808	0.9915
0.0546	47.6923	1240	0.0357	0.9532	0.9774	0.9897	nan	0.9916	0.9450	0.9956	nan	0.9856	0.8821	0.9919
0.0297	48.4615	1260	0.0351	0.9526	0.9776	0.9896	nan	0.9925	0.9461	0.9942	nan	0.9857	0.8807	0.9915
0.053	49.2308	1280	0.0349	0.9527	0.9779	0.9896	nan	0.9921	0.9471	0.9945	nan	0.9856	0.8809	0.9916
0.0474	50.0	1300	0.0361	0.9518	0.9783	0.9893	nan	0.9923	0.9490	0.9935	nan	0.9857	0.8788	0.9911

Framework versions

• Transformers 4.40.2
• Pytorch 2.2.0+cpu
• Datasets 2.19.1
• Tokenizers 0.19.1