segformer-b1-improved

This model is a fine-tuned version of nvidia/mit-b1 on the samitizerxu/kelp_data dataset. It achieves the following results on the evaluation set:

• Iou Kelp: 0.0
• Loss: 0.0024

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: 2e-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: 0.2
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Iou Kelp	Validation Loss
0.0769	0.11	30	0.0	0.0650
0.075	0.21	60	0.0	0.0710
0.0701	0.32	90	0.0	0.0738
0.0704	0.43	120	0.0	0.0710
0.0669	0.53	150	0.0	0.0699
0.0694	0.64	180	0.0	0.0679
0.0635	0.74	210	0.0	0.0676
0.0584	0.85	240	0.0	0.0647
0.0552	0.96	270	0.0	0.0631
0.0525	1.06	300	0.0	0.0592
0.052	1.17	330	0.0	0.0540
0.0449	1.28	360	0.0	0.0517
0.0443	1.38	390	0.0	0.0459
0.0364	1.49	420	0.0	0.0422
0.0361	1.6	450	0.0	0.0374
0.0339	1.7	480	0.0	0.0352
0.0307	1.81	510	0.0	0.0319
0.0283	1.91	540	0.0	0.0283
0.0302	2.02	570	0.0	0.0275
0.0277	2.13	600	0.0	0.0252
0.0231	2.23	630	0.0	0.0223
0.0261	2.34	660	0.0	0.0219
0.025	2.45	690	0.0	0.0183
0.0262	2.55	720	0.0	0.0163
0.0182	2.66	750	0.0	0.0175
0.0186	2.77	780	0.0	0.0147
0.0156	2.87	810	0.0	0.0151
0.0171	2.98	840	0.0	0.0148
0.0164	3.09	870	0.0	0.0161
0.0155	3.19	900	0.0	0.0139
0.0127	3.3	930	0.0	0.0126
0.0111	3.4	960	0.0	0.0105
0.0139	3.51	990	0.0	0.0103
0.0113	3.62	1020	0.0	0.0096
0.0093	3.72	1050	0.0	0.0088
0.0099	3.83	1080	0.0	0.0096
0.0092	3.94	1110	0.0	0.0096
0.0095	4.04	1140	0.0	0.0083
0.0113	4.15	1170	0.0	0.0086
0.0084	4.26	1200	0.0	0.0073
0.0066	4.36	1230	0.0	0.0081
0.01	4.47	1260	0.0	0.0074
0.0062	4.57	1290	0.0	0.0067
0.0053	4.68	1320	0.0	0.0065
0.0073	4.79	1350	0.0	0.0058
0.0069	4.89	1380	0.0	0.0065
0.0063	5.0	1410	0.0	0.0060
0.0062	5.11	1440	0.0	0.0055
0.0069	5.21	1470	0.0	0.0058
0.0049	5.32	1500	0.0	0.0055
0.0047	5.43	1530	0.0	0.0050
0.0092	5.53	1560	0.0	0.0054
0.0041	5.64	1590	0.0	0.0049
0.0042	5.74	1620	0.0	0.0049
0.005	5.85	1650	0.0	0.0044
0.0036	5.96	1680	0.0	0.0047
0.0059	6.06	1710	0.0	0.0048
0.0078	6.17	1740	0.0	0.0045
0.0036	6.28	1770	0.0	0.0042
0.004	6.38	1800	0.0	0.0039
0.0038	6.49	1830	0.0	0.0041
0.0038	6.6	1860	0.0	0.0042
0.0028	6.7	1890	0.0	0.0042
0.0036	6.81	1920	0.0	0.0036
0.0032	6.91	1950	0.0	0.0037
0.0029	7.02	1980	0.0	0.0035
0.0022	7.13	2010	0.0	0.0036
0.0025	7.23	2040	0.0	0.0036
0.0024	7.34	2070	0.0	0.0035
0.0022	7.45	2100	0.0	0.0032
0.0031	7.55	2130	0.0	0.0033
0.0028	7.66	2160	0.0	0.0031
0.0022	7.77	2190	0.0	0.0032
0.0023	7.87	2220	0.0	0.0034
0.0021	7.98	2250	0.0	0.0033
0.0018	8.09	2280	0.0	0.0031
0.0021	8.19	2310	0.0	0.0032
0.0038	8.3	2340	0.0	0.0030
0.0016	8.4	2370	0.0	0.0031
0.0066	8.51	2400	0.0	0.0029
0.0015	8.62	2430	0.0	0.0030
0.0022	8.72	2460	0.0	0.0028
0.0022	8.83	2490	0.0	0.0028
0.0012	8.94	2520	0.0	0.0027
0.0017	9.04	2550	0.0	0.0028
0.0019	9.15	2580	0.0	0.0027
0.0014	9.26	2610	0.0	0.0028
0.0014	9.36	2640	0.0	0.0028
0.0013	9.47	2670	0.0	0.0027
0.0014	9.57	2700	0.0	0.0027
0.0015	9.68	2730	0.0	0.0028
0.0014	9.79	2760	0.0	0.0027
0.001	9.89	2790	0.0	0.0026
0.0013	10.0	2820	0.0	0.0026
0.0014	10.11	2850	0.0	0.0026
0.0014	10.21	2880	0.0	0.0026
0.0031	10.32	2910	0.0	0.0025
0.0013	10.43	2940	0.0	0.0026
0.0012	10.53	2970	0.0	0.0026
0.0011	10.64	3000	0.0	0.0025
0.0009	10.74	3030	0.0	0.0025
0.0016	10.85	3060	0.0	0.0025
0.0014	10.96	3090	0.0	0.0025
0.0008	11.06	3120	0.0	0.0025
0.0008	11.17	3150	0.0	0.0025
0.0019	11.28	3180	0.0	0.0026
0.001	11.38	3210	0.0	0.0025
0.0009	11.49	3240	0.0	0.0025
0.0014	11.6	3270	0.0	0.0026
0.0014	11.7	3300	0.0	0.0024
0.0008	11.81	3330	0.0	0.0025
0.001	11.91	3360	0.0	0.0024
0.0012	12.02	3390	0.0	0.0024
0.0009	12.13	3420	0.0	0.0024
0.0007	12.23	3450	0.0	0.0025
0.0011	12.34	3480	0.0	0.0024
0.0012	12.45	3510	0.0	0.0023
0.0008	12.55	3540	0.0	0.0023
0.0005	12.66	3570	0.0	0.0024
0.0007	12.77	3600	0.0	0.0023
0.0006	12.87	3630	0.0	0.0024
0.0006	12.98	3660	0.0	0.0024
0.0014	13.09	3690	0.0	0.0025
0.0008	13.19	3720	0.0	0.0023
0.0007	13.3	3750	0.0	0.0022
0.0005	13.4	3780	0.0	0.0022
0.0006	13.51	3810	0.0	0.0022
0.0005	13.62	3840	0.0	0.0023
0.0007	13.72	3870	0.0	0.0022
0.0009	13.83	3900	0.0	0.0023
0.0005	13.94	3930	0.0	0.0022
0.0006	14.04	3960	0.0	0.0023
0.0005	14.15	3990	0.0	0.0023
0.0005	14.26	4020	0.0	0.0023
0.0008	14.36	4050	0.0	0.0022
0.0006	14.47	4080	0.0	0.0022
0.001	14.57	4110	0.0	0.0022
0.0004	14.68	4140	0.0	0.0021
0.0005	14.79	4170	0.0	0.0022
0.0006	14.89	4200	0.0	0.0023
0.0006	15.0	4230	0.0	0.0023
0.0007	15.11	4260	0.0	0.0022
0.0006	15.21	4290	0.0	0.0022
0.0005	15.32	4320	0.0	0.0021
0.0004	15.43	4350	0.0	0.0021
0.0008	15.53	4380	0.0	0.0021
0.0005	15.64	4410	0.0	0.0021
0.0005	15.74	4440	0.0	0.0022
0.0003	15.85	4470	0.0	0.0021
0.0005	15.96	4500	0.0	0.0021
0.0004	16.06	4530	0.0	0.0023
0.0004	16.17	4560	0.0	0.0022
0.0004	16.28	4590	0.0	0.0021
0.0005	16.38	4620	0.0	0.0020
0.0007	16.49	4650	0.0	0.0022
0.0005	16.6	4680	0.0	0.0021
0.0011	16.7	4710	0.0	0.0020
0.0005	16.81	4740	0.0	0.0020
0.0005	16.91	4770	0.0	0.0022
0.0004	17.02	4800	0.0	0.0021
0.0004	17.13	4830	0.0	0.0022
0.0004	17.23	4860	0.0	0.0021
0.0007	17.34	4890	0.0	0.0021
0.0006	17.45	4920	0.0	0.0021
0.0006	17.55	4950	0.0	0.0021
0.001	17.66	4980	0.0	0.0021
0.0003	17.77	5010	0.0	0.0020
0.0002	17.87	5040	0.0	0.0021
0.0005	17.98	5070	0.0	0.0021
0.0005	18.09	5100	0.0	0.0021
0.0004	18.19	5130	0.0	0.0021
0.0008	18.3	5160	0.0	0.0023
0.0004	18.4	5190	0.0	0.0020
0.0002	18.51	5220	0.0	0.0021
0.0003	18.62	5250	0.0	0.0020
0.0006	18.72	5280	0.0	0.0021
0.0004	18.83	5310	0.0	0.0021
0.0005	18.94	5340	0.0	0.0020
0.0004	19.04	5370	0.0	0.0020
0.0007	19.15	5400	0.0	0.0020
0.0005	19.26	5430	0.0	0.0021
0.0006	19.36	5460	0.0	0.0020
0.0004	19.47	5490	0.0	0.0021
0.0004	19.57	5520	0.0	0.0021
0.0002	19.68	5550	0.0	0.0021
0.0008	19.79	5580	0.0	0.0020
0.0006	19.89	5610	0.0	0.0024
0.0005	20.0	5640	0.0	0.0022
0.0002	20.11	5670	0.0	0.0021
0.0002	20.21	5700	0.0	0.0022
0.0002	20.32	5730	0.0	0.0022
0.0004	20.43	5760	0.0	0.0022
0.0002	20.53	5790	0.0	0.0023
0.0008	20.64	5820	0.0	0.0022
0.0002	20.74	5850	0.0	0.0023
0.0002	20.85	5880	0.0	0.0023
0.0021	20.96	5910	0.0	0.0023
0.0002	21.06	5940	0.0	0.0023
0.0002	21.17	5970	0.0	0.0023
0.0003	21.28	6000	0.0	0.0022
0.0002	21.38	6030	0.0	0.0023
0.0004	21.49	6060	0.0	0.0022
0.0002	21.6	6090	0.0	0.0023
0.0002	21.7	6120	0.0	0.0021
0.0007	21.81	6150	0.0	0.0022
0.0004	21.91	6180	0.0	0.0021
0.0002	22.02	6210	0.0	0.0021
0.0004	22.13	6240	0.0	0.0023
0.0004	22.23	6270	0.0	0.0023
0.0002	22.34	6300	0.0	0.0023
0.0002	22.45	6330	0.0	0.0024
0.0006	22.55	6360	0.0	0.0023
0.0002	22.66	6390	0.0	0.0021
0.0002	22.77	6420	0.0	0.0022
0.0002	22.87	6450	0.0	0.0023
0.0003	22.98	6480	0.0	0.0022
0.0004	23.09	6510	0.0	0.0021
0.0001	23.19	6540	0.0	0.0021
0.0004	23.3	6570	0.0	0.0023
0.0002	23.4	6600	0.0	0.0023
0.0002	23.51	6630	0.0	0.0021
0.0007	23.62	6660	0.0	0.0023
0.0001	23.72	6690	0.0	0.0022
0.0005	23.83	6720	0.0	0.0022
0.0003	23.94	6750	0.0	0.0022
0.0003	24.04	6780	0.0	0.0023
0.0013	24.15	6810	0.0	0.0022
0.0001	24.26	6840	0.0	0.0023
0.0003	24.36	6870	0.0	0.0022
0.0003	24.47	6900	0.0	0.0019
0.0001	24.57	6930	0.0	0.0023
0.0002	24.68	6960	0.0	0.0023
0.0002	24.79	6990	0.0	0.0021
0.0003	24.89	7020	0.0	0.0022
0.0003	25.0	7050	0.0	0.0022
0.0001	25.11	7080	0.0	0.0023
0.0002	25.21	7110	0.0	0.0022
0.0002	25.32	7140	0.0	0.0022
0.0002	25.43	7170	0.0	0.0022
0.0002	25.53	7200	0.0	0.0022
0.0003	25.64	7230	0.0	0.0023
0.0001	25.74	7260	0.0	0.0024
0.0001	25.85	7290	0.0	0.0023
0.0002	25.96	7320	0.0	0.0022
0.0002	26.06	7350	0.0	0.0021
0.0004	26.17	7380	0.0	0.0023
0.0002	26.28	7410	0.0	0.0023
0.0001	26.38	7440	0.0	0.0023
0.0002	26.49	7470	0.0	0.0020
0.0006	26.6	7500	0.0	0.0027
0.0002	26.7	7530	0.0	0.0022
0.0002	26.81	7560	0.0	0.0022
0.0006	26.91	7590	0.0	0.0020
0.0002	27.02	7620	0.0	0.0023
0.0004	27.13	7650	0.0	0.0022
0.0002	27.23	7680	0.0	0.0020
0.0004	27.34	7710	0.0	0.0021
0.0002	27.45	7740	0.0	0.0023
0.0001	27.55	7770	0.0	0.0023
0.0005	27.66	7800	0.0	0.0022
0.0002	27.77	7830	0.0	0.0023
0.0001	27.87	7860	0.0	0.0021
0.0001	27.98	7890	0.0	0.0022
0.0001	28.09	7920	0.0	0.0022
0.0002	28.19	7950	0.0	0.0021
0.0002	28.3	7980	0.0	0.0020
0.0001	28.4	8010	0.0	0.0023
0.0006	28.51	8040	0.0	0.0022
0.0001	28.62	8070	0.0	0.0023
0.0001	28.72	8100	0.0	0.0022
0.0001	28.83	8130	0.0	0.0021
0.0001	28.94	8160	0.0	0.0021
0.0002	29.04	8190	0.0	0.0023
0.0002	29.15	8220	0.0	0.0022
0.0001	29.26	8250	0.0	0.0020
0.0003	29.36	8280	0.0	0.0023
0.0001	29.47	8310	0.0	0.0022
0.0001	29.57	8340	0.0	0.0023
0.0002	29.68	8370	0.0	0.0023
0.0002	29.79	8400	0.0	0.0021
0.0002	29.89	8430	0.0	0.0023
0.0001	30.0	8460	0.0	0.0023
0.0001	30.11	8490	0.0	0.0025
0.0001	30.21	8520	0.0	0.0024
0.0001	30.32	8550	0.0	0.0024
0.0002	30.43	8580	0.0	0.0024
0.0003	30.53	8610	0.0	0.0024
0.0004	30.64	8640	0.0	0.0023
0.0003	30.74	8670	0.0	0.0021
0.0001	30.85	8700	0.0	0.0019
0.0001	30.96	8730	0.0	0.0021
0.0001	31.06	8760	0.0	0.0023
0.0004	31.17	8790	0.0	0.0022
0.0002	31.28	8820	0.0	0.0019
0.0001	31.38	8850	0.0	0.0023
0.0005	31.49	8880	0.0	0.0024
0.0001	31.6	8910	0.0	0.0024
0.0003	31.7	8940	0.0	0.0024
0.0005	31.81	8970	0.0	0.0024
0.0002	31.91	9000	0.0	0.0022
0.0003	32.02	9030	0.0	0.0023
0.0001	32.13	9060	0.0	0.0023
0.0003	32.23	9090	0.0	0.0023
0.0002	32.34	9120	0.0	0.0021
0.0002	32.45	9150	0.0	0.0023
0.0006	32.55	9180	0.0	0.0022
0.0006	32.66	9210	0.0	0.0023
0.0014	32.77	9240	0.0	0.0023
0.0001	32.87	9270	0.0	0.0022
0.0001	32.98	9300	0.0	0.0021
0.0001	33.09	9330	0.0	0.0021
0.0002	33.19	9360	0.0	0.0025
0.0001	33.3	9390	0.0	0.0024
0.0001	33.4	9420	0.0	0.0022
0.0002	33.51	9450	0.0	0.0023
0.002	33.62	9480	0.0	0.0024
0.0003	33.72	9510	0.0	0.0022
0.0003	33.83	9540	0.0	0.0023
0.0006	33.94	9570	0.0	0.0023
0.0001	34.04	9600	0.0	0.0024
0.0001	34.15	9630	0.0	0.0023
0.0002	34.26	9660	0.0	0.0023
0.0002	34.36	9690	0.0	0.0021
0.0001	34.47	9720	0.0	0.0023
0.0001	34.57	9750	0.0	0.0023
0.0003	34.68	9780	0.0	0.0024
0.0002	34.79	9810	0.0	0.0023
0.0002	34.89	9840	0.0	0.0023
0.0001	35.0	9870	0.0	0.0023
0.0003	35.11	9900	0.0	0.0024
0.0001	35.21	9930	0.0	0.0024
0.0002	35.32	9960	0.0	0.0022
0.0001	35.43	9990	0.0	0.0024
0.0002	35.53	10020	0.0	0.0023
0.0001	35.64	10050	0.0	0.0024
0.0001	35.74	10080	0.0	0.0024
0.0001	35.85	10110	0.0	0.0024
0.0001	35.96	10140	0.0	0.0022
0.0002	36.06	10170	0.0	0.0023
0.0002	36.17	10200	0.0	0.0025
0.0001	36.28	10230	0.0	0.0025
0.0001	36.38	10260	0.0	0.0027
0.0001	36.49	10290	0.0	0.0024
0.0001	36.6	10320	0.0	0.0024
0.0004	36.7	10350	0.0	0.0024
0.0002	36.81	10380	0.0	0.0023
0.0002	36.91	10410	0.0	0.0024
0.0003	37.02	10440	0.0	0.0024
0.0004	37.13	10470	0.0	0.0025
0.0001	37.23	10500	0.0	0.0024
0.0001	37.34	10530	0.0	0.0024
0.0001	37.45	10560	0.0	0.0025
0.0002	37.55	10590	0.0	0.0025
0.0002	37.66	10620	0.0	0.0024
0.0001	37.77	10650	0.0	0.0025
0.0005	37.87	10680	0.0	0.0024
0.0001	37.98	10710	0.0	0.0025
0.0	38.09	10740	0.0	0.0025
0.0001	38.19	10770	0.0	0.0026
0.0001	38.3	10800	0.0	0.0025
0.0001	38.4	10830	0.0	0.0025
0.0001	38.51	10860	0.0	0.0025
0.0002	38.62	10890	0.0	0.0024
0.0001	38.72	10920	0.0	0.0025
0.0	38.83	10950	0.0	0.0025
0.0001	38.94	10980	0.0	0.0024
0.0001	39.04	11010	0.0	0.0024
0.0003	39.15	11040	0.0	0.0024
0.0005	39.26	11070	0.0	0.0025
0.0	39.36	11100	0.0	0.0024
0.0001	39.47	11130	0.0	0.0024
0.0002	39.57	11160	0.0	0.0024
0.0004	39.68	11190	0.0	0.0024
0.0002	39.79	11220	0.0	0.0025
0.0003	39.89	11250	0.0	0.0024
0.0	40.0	11280	0.0	0.0024

Framework versions

• Transformers 4.37.1
• Pytorch 2.1.2
• Datasets 2.16.1
• Tokenizers 0.15.1