G17-AMFU-Net

This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2900
• Dice: 0.8320
• Iou: 0.7535
• Precision: 0.8675
• Recall: 0.8433
• Specificity: 0.9870

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: 0.0003
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• 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: 60

Training results

Training Loss	Epoch	Step	Dice	Iou	Validation Loss	Precision	Recall	Specificity
1.0718	1.1848	250	0.5860	0.4503	1.0382	0.5331	0.8221	0.9518
0.7947	2.3697	500	0.6467	0.5207	0.7440	0.5922	0.8447	0.9562
0.5498	3.5545	750	0.7286	0.6204	0.5669	0.7235	0.8263	0.9778
0.5368	4.7393	1000	0.7452	0.6478	0.5151	0.8479	0.7400	0.9908
0.4243	5.9242	1250	0.7608	0.6599	0.4263	0.7784	0.8140	0.9821
0.4262	7.1090	1500	0.7746	0.6779	0.3980	0.7891	0.8336	0.9797
0.3725	8.2938	1750	0.7567	0.6697	0.4304	0.8597	0.7570	0.9904
0.3821	9.4787	2000	0.7855	0.6966	0.3779	0.8407	0.8049	0.9869
0.3208	10.6635	2250	0.7872	0.6951	0.3735	0.8321	0.8071	0.9869
0.428	11.8483	2500	0.7978	0.7073	0.3502	0.8326	0.8180	0.9865
0.3897	13.0332	2750	0.7856	0.6926	0.3738	0.7879	0.8389	0.9814
0.3618	14.2180	3000	0.7787	0.6846	0.3871	0.8853	0.7413	0.9920
0.315	15.4028	3250	0.8045	0.7187	0.3467	0.8567	0.8101	0.9890
0.3798	16.5877	3500	0.8116	0.7224	0.3337	0.8230	0.8481	0.9849
0.3345	17.7725	3750	0.8109	0.7281	0.3244	0.8750	0.8128	0.9891
0.3678	18.9573	4000	0.8133	0.7271	0.3278	0.8281	0.8526	0.9846
0.3247	20.1422	4250	0.8107	0.7252	0.3216	0.8291	0.8504	0.9805
0.3474	21.3270	4500	0.7956	0.7096	0.3534	0.8763	0.7865	0.9921
0.2496	22.5118	4750	0.8153	0.7305	0.3227	0.8556	0.8224	0.9883
0.3256	23.6967	5000	0.8139	0.7242	0.3235	0.8041	0.8716	0.9811
0.3007	24.8815	5250	0.8170	0.7314	0.3164	0.8368	0.8499	0.9834
0.3147	26.0664	5500	0.8164	0.7296	0.3132	0.8430	0.8377	0.9867
0.32	27.2512	5750	0.8133	0.7320	0.3268	0.8551	0.8234	0.9888
0.3075	28.4360	6000	0.8147	0.7284	0.3280	0.8323	0.8503	0.9856
0.3115	29.6209	6250	0.8161	0.7299	0.3149	0.8580	0.8224	0.9845
0.289	30.8057	6500	0.8276	0.7438	0.2971	0.8677	0.8299	0.9867
0.3096	31.9905	6750	0.8331	0.7491	0.2892	0.8273	0.8792	0.9803
0.2563	33.1754	7000	0.8290	0.7431	0.2998	0.8391	0.8567	0.9843
0.2647	34.3602	7250	0.8112	0.7275	0.3341	0.8952	0.7875	0.9916
0.2489	35.5450	7500	0.8191	0.7331	0.3096	0.8294	0.8540	0.9829
0.252	36.7299	7750	0.8302	0.7484	0.2965	0.8591	0.8447	0.9872
0.2575	37.9147	8000	0.8188	0.7372	0.3077	0.8603	0.8227	0.9882
0.2733	39.0995	8250	0.8260	0.7461	0.2984	0.8715	0.8288	0.9887
0.2686	40.2844	8500	0.8307	0.7453	0.2935	0.8404	0.8634	0.9815
0.2413	41.4692	8750	0.8285	0.7446	0.2948	0.8438	0.8598	0.9846
0.2277	42.6540	9000	0.8304	0.7503	0.2972	0.8820	0.8298	0.9901
0.2468	43.8389	9250	0.8331	0.7515	0.2989	0.8791	0.8301	0.9897
0.2491	45.0237	9500	0.8307	0.7499	0.2967	0.8734	0.8394	0.9889
0.2199	46.2085	9750	0.8382	0.7565	0.2861	0.8598	0.8528	0.9866
0.2463	47.3934	10000	0.8397	0.7570	0.2822	0.8488	0.8703	0.9837
0.2285	48.5782	10250	0.8348	0.7539	0.2821	0.8715	0.8376	0.9878
0.2367	49.7630	10500	0.8339	0.7532	0.2920	0.8651	0.8413	0.9881
0.2387	50.9479	10750	0.8366	0.7558	0.2830	0.8514	0.8580	0.9858
0.2384	52.1327	11000	0.8382	0.7569	0.2768	0.8597	0.8537	0.9870
0.2108	53.3175	11250	0.8366	0.7570	0.2835	0.8662	0.8499	0.9873
0.1938	54.5024	11500	0.8306	0.7518	0.2925	0.8759	0.8382	0.9881
0.2203	55.6872	11750	0.8359	0.7565	0.2873	0.8767	0.8405	0.9886
0.1858	56.8720	12000	0.8342	0.7552	0.2918	0.8771	0.8369	0.9881
0.2523	58.0569	12250	0.2905	0.8319	0.7533	0.8659	0.8451	0.9870
0.221	59.2417	12500	0.2900	0.8320	0.7535	0.8675	0.8433	0.9870

Framework versions

• Transformers 4.57.1
• Pytorch 2.10.0+cu128
• Datasets 4.8.5
• Tokenizers 0.22.2