segformer-b4-random-init-morphpadver1-hgo-coord-v9_mix_resample_40epochs

This model is a fine-tuned version of random_init on the NICOPOI-9/morphpad_coord_hgo_512_4class_v2 dataset. It achieves the following results on the evaluation set:

• Loss: 1.3834
• Mean Iou: 0.1030
• Mean Accuracy: 0.2500
• Overall Accuracy: 0.2754
• Accuracy 0-0: 0.0
• Accuracy 0-90: 0.7090
• Accuracy 90-0: 0.2908
• Accuracy 90-90: 0.0
• Iou 0-0: 0.0
• Iou 0-90: 0.2501
• Iou 90-0: 0.1618
• Iou 90-90: 0.0

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: 1
• eval_batch_size: 1
• seed: 42
• 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: linear
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy 0-0	Accuracy 0-90	Accuracy 90-0	Accuracy 90-90	Iou 0-0	Iou 0-90	Iou 90-0	Iou 90-90
1.3905	1.3638	4000	1.3824	0.1021	0.2504	0.2759	0.0006	0.7381	0.2614	0.0013	0.0006	0.2519	0.1546	0.0013
1.3909	2.7276	8000	1.3850	0.0902	0.2517	0.2764	0.0201	0.9055	0.0549	0.0264	0.0189	0.2694	0.0479	0.0247
1.3898	4.0914	12000	1.3876	0.1004	0.2495	0.2473	0.0002	0.0077	0.4529	0.5370	0.0002	0.0076	0.2046	0.1892
1.3848	5.4552	16000	1.3852	0.0762	0.2499	0.2758	0.0201	0.9673	0.0113	0.0007	0.0188	0.2745	0.0110	0.0007
1.3912	6.8190	20000	1.3829	0.1076	0.2500	0.2741	0.0	0.4934	0.5068	0.0	0.0	0.2165	0.2139	0.0
1.3394	8.1827	24000	1.3831	0.0698	0.2501	0.2773	0.0	0.9983	0.0022	0.0	0.0	0.2771	0.0022	0.0
1.4205	9.5465	28000	1.3845	0.1274	0.2499	0.2605	0.0	0.3064	0.4085	0.2845	0.0	0.1682	0.1966	0.1448
1.382	10.9103	32000	1.3839	0.0861	0.2487	0.2665	0.0713	0.0273	0.8963	0.0	0.0569	0.0254	0.2619	0.0
1.3715	12.2741	36000	1.3869	0.0884	0.2498	0.2762	0.0	0.8874	0.1116	0.0	0.0	0.2689	0.0847	0.0
1.3879	13.6379	40000	1.3818	0.0802	0.2526	0.2743	0.0	0.0602	0.9501	0.0	0.0	0.0539	0.2671	0.0
1.3847	15.0017	44000	1.3850	0.0678	0.25	0.2711	0.0	0.0	1.0	0.0	0.0	0.0	0.2711	0.0
1.3943	16.3655	48000	1.3833	0.0739	0.2497	0.2693	0.0302	0.0001	0.9684	0.0	0.0273	0.0001	0.2684	0.0
1.3877	17.7293	52000	1.3842	0.1075	0.2511	0.2749	0.0	0.4278	0.5767	0.0	0.0	0.2042	0.2257	0.0
1.3813	19.0931	56000	1.3833	0.1195	0.2532	0.2746	0.0	0.4714	0.4770	0.0643	0.0	0.2154	0.2104	0.0523
1.3934	20.4569	60000	1.3844	0.0700	0.2501	0.2712	0.0	0.0098	0.9904	0.0000	0.0	0.0096	0.2704	0.0000
1.3841	21.8207	64000	1.3828	0.0851	0.2524	0.2794	0.0	0.9286	0.0811	0.0	0.0	0.2730	0.0674	0.0
1.3697	23.1845	68000	1.3835	0.0939	0.2484	0.2743	0.0	0.8252	0.1682	0.0	0.0	0.2612	0.1145	0.0
1.3951	24.5482	72000	1.3836	0.1060	0.2484	0.2719	0.0	0.4205	0.5731	0.0	0.0	0.1985	0.2255	0.0
1.3842	25.9120	76000	1.3828	0.0700	0.2500	0.2771	0.0	0.9966	0.0030	0.0003	0.0	0.2769	0.0030	0.0003
1.3952	27.2758	80000	1.3825	0.1020	0.2456	0.2683	0.0	0.3268	0.6555	0.0	0.0	0.1705	0.2374	0.0
1.4017	28.6396	84000	1.3823	0.0912	0.2555	0.2826	0.0	0.9046	0.1175	0.0	0.0	0.2731	0.0917	0.0
1.3873	30.0034	88000	1.3823	0.0742	0.2487	0.2756	0.0	0.9701	0.0248	0.0	0.0	0.2738	0.0232	0.0
1.4049	31.3672	92000	1.3834	0.0890	0.2503	0.2722	0.0	0.1255	0.8756	0.0	0.0	0.0941	0.2619	0.0
1.3727	32.7310	96000	1.3829	0.0721	0.2498	0.2769	0.0	0.9865	0.0128	0.0	0.0	0.2759	0.0123	0.0
1.3856	34.0948	100000	1.3836	0.0748	0.2504	0.2775	0.0000	0.9762	0.0255	0.0000	0.0000	0.2753	0.0240	0.0000
1.3706	35.4586	104000	1.3829	0.0915	0.2524	0.2791	0.0	0.8800	0.1297	0.0	0.0	0.2689	0.0971	0.0
1.3861	36.8224	108000	1.3833	0.1078	0.2511	0.2758	0.0	0.5713	0.4331	0.0	0.0	0.2302	0.2008	0.0
1.3839	38.1862	112000	1.3831	0.0812	0.2508	0.2776	0.0	0.9429	0.0601	0.0	0.0	0.2729	0.0520	0.0
1.3839	39.5499	116000	1.3834	0.1030	0.2500	0.2754	0.0	0.7090	0.2908	0.0	0.0	0.2501	0.1618	0.0

Framework versions

• Transformers 4.48.3
• Pytorch 2.1.0
• Datasets 3.2.0
• Tokenizers 0.21.0