segformer-b0-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: 1.2528
• Mean Iou: 0.2977
• Mean Accuracy: 0.4609
• Overall Accuracy: 0.4591
• Accuracy 0-0: 0.3603
• Accuracy 0-90: 0.4502
• Accuracy 90-0: 0.4232
• Accuracy 90-90: 0.6098
• Iou 0-0: 0.2715
• Iou 0-90: 0.3233
• Iou 90-0: 0.3141
• Iou 90-90: 0.2821

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
1.3877	1.3638	4000	1.3825	0.1120	0.2581	0.2834	0.0074	0.6338	0.3902	0.0011	0.0073	0.2487	0.1909	0.0011
1.2102	2.7276	8000	1.3756	0.1484	0.2805	0.2694	0.2393	0.1558	0.1636	0.5633	0.1609	0.1086	0.1147	0.2092
1.3585	4.0914	12000	1.3637	0.1622	0.2963	0.2942	0.2999	0.0845	0.4896	0.3115	0.1795	0.0708	0.2194	0.1792
1.3062	5.4552	16000	1.3470	0.1825	0.3121	0.3212	0.2055	0.5129	0.2845	0.2453	0.1549	0.2359	0.1840	0.1554
1.3491	6.8190	20000	1.3312	0.1933	0.3303	0.3424	0.1899	0.5864	0.3057	0.2391	0.1462	0.2584	0.2009	0.1676
1.1479	8.1827	24000	1.3200	0.2086	0.3471	0.3523	0.2267	0.3377	0.4672	0.3570	0.1655	0.2167	0.2496	0.2026
1.4035	9.5465	28000	1.3137	0.2157	0.3594	0.3602	0.2554	0.4350	0.2812	0.4661	0.1857	0.2445	0.2053	0.2271
1.3416	10.9103	32000	1.3023	0.2201	0.3639	0.3660	0.2353	0.3488	0.4199	0.4515	0.1768	0.2346	0.2462	0.2228
1.095	12.2741	36000	1.2883	0.2310	0.3757	0.3795	0.3025	0.4730	0.3459	0.3813	0.2028	0.2700	0.2346	0.2165
1.2933	13.6379	40000	1.2834	0.2327	0.3785	0.3882	0.2849	0.4351	0.5287	0.2655	0.1996	0.2682	0.2804	0.1825
1.2888	15.0017	44000	1.2892	0.2325	0.3807	0.3921	0.2443	0.6177	0.3612	0.2995	0.1860	0.3028	0.2493	0.1918
1.2542	16.3655	48000	1.2823	0.2372	0.3865	0.3972	0.2325	0.6028	0.3713	0.3395	0.1804	0.3078	0.2554	0.2050
1.2767	17.7293	52000	1.2623	0.2481	0.3976	0.4070	0.2735	0.4473	0.5460	0.3236	0.2071	0.2865	0.3009	0.1978
1.2314	19.0931	56000	1.2697	0.2541	0.4045	0.4111	0.2902	0.5142	0.4228	0.3906	0.2086	0.3068	0.2825	0.2186
1.273	20.4569	60000	1.2814	0.2509	0.4011	0.4111	0.2550	0.5118	0.4926	0.3452	0.2010	0.3025	0.2933	0.2066
1.5948	21.8207	64000	1.2636	0.2561	0.4070	0.4150	0.2698	0.5138	0.4581	0.3862	0.2084	0.3075	0.2925	0.2160
0.8904	23.1845	68000	1.2795	0.2490	0.4008	0.4138	0.2460	0.5977	0.4621	0.2974	0.1932	0.3167	0.2907	0.1953
1.2066	24.5482	72000	1.2601	0.2555	0.4119	0.4259	0.2483	0.4454	0.6835	0.2703	0.2036	0.3017	0.3208	0.1959
1.284	25.9120	76000	1.2672	0.2615	0.4179	0.4220	0.2264	0.4455	0.4647	0.5349	0.1919	0.2995	0.3056	0.2490
1.3295	27.2758	80000	1.2713	0.2703	0.4265	0.4333	0.2465	0.5185	0.4634	0.4776	0.2047	0.3240	0.3087	0.2437
0.857	28.6396	84000	1.2813	0.2655	0.4286	0.4262	0.2374	0.4049	0.3834	0.6887	0.2048	0.2945	0.2891	0.2734
1.1713	30.0034	88000	1.2530	0.2789	0.4374	0.4473	0.3726	0.4646	0.6296	0.2827	0.2610	0.3141	0.3358	0.2046
0.7255	31.3672	92000	1.2424	0.2842	0.4424	0.4444	0.3358	0.4430	0.4794	0.5114	0.2469	0.3141	0.3177	0.2582
1.0961	32.7310	96000	1.2584	0.2760	0.4366	0.4447	0.2329	0.4157	0.6283	0.4696	0.2077	0.3070	0.3346	0.2548
1.1369	34.0948	100000	1.2632	0.2863	0.4469	0.4483	0.2992	0.4085	0.5100	0.5698	0.2409	0.3092	0.3272	0.2677
1.0328	35.4586	104000	1.2445	0.2921	0.4512	0.4568	0.3166	0.5220	0.4865	0.4798	0.2508	0.3353	0.3201	0.2621
1.2026	36.8224	108000	1.2417	0.2928	0.4522	0.4602	0.3099	0.4714	0.6024	0.4251	0.2529	0.3287	0.3358	0.2537
1.2102	38.1862	112000	1.3176	0.2773	0.4430	0.4419	0.2541	0.3309	0.5290	0.6580	0.2253	0.2724	0.3334	0.2782
1.1765	39.5499	116000	1.2528	0.2977	0.4609	0.4591	0.3603	0.4502	0.4232	0.6098	0.2715	0.3233	0.3141	0.2821

Framework versions

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