HDCSHseg_models

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

• Loss: 0.5558
• Mean Iou: 0.3834
• Mean Accuracy: 0.7668
• Overall Accuracy: 0.7668
• Accuracy 背景: nan
• Accuracy 未水化水泥颗粒与高密度c-s-h混合: 0.7668
• Iou 背景: 0.0
• Iou 未水化水泥颗粒与高密度c-s-h混合: 0.7668

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: 8
• eval_batch_size: 8
• seed: 1337
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: polynomial
• training_steps: 10000

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy 背景	Accuracy 未水化水泥颗粒与高密度c-s-h混合	Iou 背景	Iou 未水化水泥颗粒与高密度c-s-h混合
0.4717	1.0	100	0.5748	0.3844	0.7688	0.7688	nan	0.7688	0.0	0.7688
0.4353	2.0	200	0.5744	0.3665	0.7330	0.7330	nan	0.7330	0.0	0.7330
0.4298	3.0	300	0.5576	0.4389	0.8778	0.8778	nan	0.8778	0.0	0.8778
0.4229	4.0	400	0.5609	0.3817	0.7635	0.7635	nan	0.7635	0.0	0.7635
0.4175	5.0	500	0.5680	0.4079	0.8158	0.8158	nan	0.8158	0.0	0.8158
0.4081	6.0	600	0.5563	0.3917	0.7835	0.7835	nan	0.7835	0.0	0.7835
0.4002	7.0	700	0.6005	0.3144	0.6287	0.6287	nan	0.6287	0.0	0.6287
0.3904	8.0	800	0.5915	0.3378	0.6757	0.6757	nan	0.6757	0.0	0.6757
0.3944	9.0	900	0.5991	0.3534	0.7069	0.7069	nan	0.7069	0.0	0.7069
0.3805	10.0	1000	0.6093	0.3250	0.6500	0.6500	nan	0.6500	0.0	0.6500
0.3813	11.0	1100	0.5965	0.3201	0.6401	0.6401	nan	0.6401	0.0	0.6401
0.3685	12.0	1200	0.6125	0.3076	0.6151	0.6151	nan	0.6151	0.0	0.6151
0.37	13.0	1300	0.5745	0.4050	0.8099	0.8099	nan	0.8099	0.0	0.8099
0.3645	14.0	1400	0.5719	0.3685	0.7369	0.7369	nan	0.7369	0.0	0.7369
0.3643	15.0	1500	0.6316	0.2919	0.5839	0.5839	nan	0.5839	0.0	0.5839
0.3591	16.0	1600	0.6657	0.2730	0.5460	0.5460	nan	0.5460	0.0	0.5460
0.3572	17.0	1700	0.5844	0.3909	0.7817	0.7817	nan	0.7817	0.0	0.7817
0.361	18.0	1800	0.6129	0.3164	0.6328	0.6328	nan	0.6328	0.0	0.6328
0.3571	19.0	1900	0.5721	0.3744	0.7488	0.7488	nan	0.7488	0.0	0.7488
0.3571	20.0	2000	0.5961	0.3331	0.6662	0.6662	nan	0.6662	0.0	0.6662
0.356	21.0	2100	0.6015	0.3264	0.6529	0.6529	nan	0.6529	0.0	0.6529
0.3535	22.0	2200	0.5709	0.3636	0.7272	0.7272	nan	0.7272	0.0	0.7272
0.3511	23.0	2300	0.5912	0.3393	0.6786	0.6786	nan	0.6786	0.0	0.6786
0.3512	24.0	2400	0.5624	0.3725	0.7451	0.7451	nan	0.7451	0.0	0.7451
0.352	25.0	2500	0.5981	0.3490	0.6981	0.6981	nan	0.6981	0.0	0.6981
0.3523	26.0	2600	0.6001	0.3504	0.7008	0.7008	nan	0.7008	0.0	0.7008
0.3485	27.0	2700	0.5707	0.3596	0.7192	0.7192	nan	0.7192	0.0	0.7192
0.3499	28.0	2800	0.5805	0.3538	0.7076	0.7076	nan	0.7076	0.0	0.7076
0.3486	29.0	2900	0.5713	0.3630	0.7261	0.7261	nan	0.7261	0.0	0.7261
0.3494	30.0	3000	0.5824	0.3648	0.7295	0.7295	nan	0.7295	0.0	0.7295
0.348	31.0	3100	0.5707	0.3538	0.7076	0.7076	nan	0.7076	0.0	0.7076
0.3465	32.0	3200	0.5624	0.3765	0.7530	0.7530	nan	0.7530	0.0	0.7530
0.3473	33.0	3300	0.5723	0.3702	0.7405	0.7405	nan	0.7405	0.0	0.7405
0.3454	34.0	3400	0.5645	0.3953	0.7907	0.7907	nan	0.7907	0.0	0.7907
0.3466	35.0	3500	0.5618	0.3832	0.7663	0.7663	nan	0.7663	0.0	0.7663
0.3451	36.0	3600	0.5704	0.3535	0.7070	0.7070	nan	0.7070	0.0	0.7070
0.3459	37.0	3700	0.5625	0.3714	0.7427	0.7427	nan	0.7427	0.0	0.7427
0.345	38.0	3800	0.5720	0.3567	0.7135	0.7135	nan	0.7135	0.0	0.7135
0.3448	39.0	3900	0.5719	0.3688	0.7376	0.7376	nan	0.7376	0.0	0.7376
0.3444	40.0	4000	0.5646	0.3809	0.7618	0.7618	nan	0.7618	0.0	0.7618
0.343	41.0	4100	0.5525	0.3833	0.7665	0.7665	nan	0.7665	0.0	0.7665
0.3438	42.0	4200	0.5547	0.3888	0.7777	0.7777	nan	0.7777	0.0	0.7777
0.3425	43.0	4300	0.5579	0.3811	0.7622	0.7622	nan	0.7622	0.0	0.7622
0.3439	44.0	4400	0.5756	0.3577	0.7153	0.7153	nan	0.7153	0.0	0.7153
0.3422	45.0	4500	0.5612	0.3775	0.7550	0.7550	nan	0.7550	0.0	0.7550
0.3418	46.0	4600	0.5654	0.3783	0.7567	0.7567	nan	0.7567	0.0	0.7567
0.3408	47.0	4700	0.5787	0.3764	0.7529	0.7529	nan	0.7529	0.0	0.7529
0.3408	48.0	4800	0.5709	0.3717	0.7435	0.7435	nan	0.7435	0.0	0.7435
0.343	49.0	4900	0.5771	0.3472	0.6944	0.6944	nan	0.6944	0.0	0.6944
0.3395	50.0	5000	0.5552	0.3786	0.7572	0.7572	nan	0.7572	0.0	0.7572
0.3394	51.0	5100	0.5626	0.3632	0.7264	0.7264	nan	0.7264	0.0	0.7264
0.3396	52.0	5200	0.5580	0.3849	0.7697	0.7697	nan	0.7697	0.0	0.7697
0.3396	53.0	5300	0.5599	0.3669	0.7338	0.7338	nan	0.7338	0.0	0.7338
0.3397	54.0	5400	0.5610	0.3740	0.7480	0.7480	nan	0.7480	0.0	0.7480
0.3385	55.0	5500	0.5594	0.3836	0.7671	0.7671	nan	0.7671	0.0	0.7671
0.338	56.0	5600	0.5567	0.3940	0.7881	0.7881	nan	0.7881	0.0	0.7881
0.3378	57.0	5700	0.5648	0.3753	0.7506	0.7506	nan	0.7506	0.0	0.7506
0.3375	58.0	5800	0.5605	0.3795	0.7589	0.7589	nan	0.7589	0.0	0.7589
0.3364	59.0	5900	0.5653	0.3839	0.7678	0.7678	nan	0.7678	0.0	0.7678
0.3373	60.0	6000	0.5649	0.3989	0.7978	0.7978	nan	0.7978	0.0	0.7978
0.3367	61.0	6100	0.5661	0.3809	0.7617	0.7617	nan	0.7617	0.0	0.7617
0.3368	62.0	6200	0.5739	0.3818	0.7637	0.7637	nan	0.7637	0.0	0.7637
0.3352	63.0	6300	0.5631	0.3965	0.7930	0.7930	nan	0.7930	0.0	0.7930
0.336	64.0	6400	0.5722	0.3745	0.7490	0.7490	nan	0.7490	0.0	0.7490
0.3352	65.0	6500	0.5622	0.3864	0.7728	0.7728	nan	0.7728	0.0	0.7728
0.3356	66.0	6600	0.5627	0.3816	0.7631	0.7631	nan	0.7631	0.0	0.7631
0.3338	67.0	6700	0.5616	0.3741	0.7483	0.7483	nan	0.7483	0.0	0.7483
0.3343	68.0	6800	0.5657	0.3706	0.7412	0.7412	nan	0.7412	0.0	0.7412
0.3343	69.0	6900	0.5603	0.3805	0.7610	0.7610	nan	0.7610	0.0	0.7610
0.3345	70.0	7000	0.5608	0.3872	0.7744	0.7744	nan	0.7744	0.0	0.7744
0.3339	71.0	7100	0.5668	0.3855	0.7710	0.7710	nan	0.7710	0.0	0.7710
0.3342	72.0	7200	0.5625	0.3954	0.7909	0.7909	nan	0.7909	0.0	0.7909
0.3334	73.0	7300	0.5556	0.3790	0.7579	0.7579	nan	0.7579	0.0	0.7579
0.3336	74.0	7400	0.5555	0.3819	0.7639	0.7639	nan	0.7639	0.0	0.7639
0.3341	75.0	7500	0.5574	0.3782	0.7563	0.7563	nan	0.7563	0.0	0.7563
0.3326	76.0	7600	0.5628	0.3701	0.7401	0.7401	nan	0.7401	0.0	0.7401
0.3325	77.0	7700	0.5575	0.3818	0.7635	0.7635	nan	0.7635	0.0	0.7635
0.3333	78.0	7800	0.5515	0.3835	0.7670	0.7670	nan	0.7670	0.0	0.7670
0.3327	79.0	7900	0.5540	0.3776	0.7552	0.7552	nan	0.7552	0.0	0.7552
0.332	80.0	8000	0.5584	0.3852	0.7705	0.7705	nan	0.7705	0.0	0.7705
0.3322	81.0	8100	0.5616	0.3770	0.7540	0.7540	nan	0.7540	0.0	0.7540
0.3316	82.0	8200	0.5555	0.3793	0.7585	0.7585	nan	0.7585	0.0	0.7585
0.3325	83.0	8300	0.5567	0.3809	0.7618	0.7618	nan	0.7618	0.0	0.7618
0.3321	84.0	8400	0.5557	0.3803	0.7606	0.7606	nan	0.7606	0.0	0.7606
0.3314	85.0	8500	0.5545	0.3776	0.7553	0.7553	nan	0.7553	0.0	0.7553
0.3315	86.0	8600	0.5552	0.3805	0.7610	0.7610	nan	0.7610	0.0	0.7610
0.3313	87.0	8700	0.5550	0.3751	0.7501	0.7501	nan	0.7501	0.0	0.7501
0.3308	88.0	8800	0.5552	0.3839	0.7678	0.7678	nan	0.7678	0.0	0.7678
0.3315	89.0	8900	0.5547	0.3799	0.7598	0.7598	nan	0.7598	0.0	0.7598
0.3312	90.0	9000	0.5567	0.3771	0.7543	0.7543	nan	0.7543	0.0	0.7543
0.3314	91.0	9100	0.5536	0.3798	0.7597	0.7597	nan	0.7597	0.0	0.7597
0.3312	92.0	9200	0.5550	0.3789	0.7578	0.7578	nan	0.7578	0.0	0.7578
0.3308	93.0	9300	0.5555	0.3798	0.7596	0.7596	nan	0.7596	0.0	0.7596
0.3301	94.0	9400	0.5590	0.3784	0.7568	0.7568	nan	0.7568	0.0	0.7568
0.3306	95.0	9500	0.5563	0.3831	0.7662	0.7662	nan	0.7662	0.0	0.7662
0.3312	96.0	9600	0.5599	0.3758	0.7517	0.7517	nan	0.7517	0.0	0.7517
0.3309	97.0	9700	0.5552	0.3832	0.7663	0.7663	nan	0.7663	0.0	0.7663
0.331	98.0	9800	0.5558	0.3867	0.7734	0.7734	nan	0.7734	0.0	0.7734
0.3306	99.0	9900	0.5550	0.3849	0.7698	0.7698	nan	0.7698	0.0	0.7698
0.3307	100.0	10000	0.5558	0.3834	0.7668	0.7668	nan	0.7668	0.0	0.7668

Framework versions

• Transformers 4.52.0.dev0
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1