segformer-b1-finetuned-segments-pv_v1_normalized_p100_4batch

This model is a fine-tuned version of nvidia/segformer-b1-finetuned-ade-512-512 on the mouadenna/satellite_PV_dataset_train_test_v1 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0012
• Mean Iou: 0.9591
• Precision: 0.9785

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: 0.0004
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.001
• num_epochs: 40
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Precision
0.0122	0.9989	229	0.0078	0.8264	0.9201
0.007	1.9978	458	0.0068	0.8134	0.8659
0.0051	2.9967	687	0.0042	0.8781	0.9431
0.0057	4.0	917	0.0038	0.8851	0.9200
0.0044	4.9989	1146	0.0042	0.8738	0.8984
0.0042	5.9978	1375	0.0035	0.8848	0.9454
0.0043	6.9967	1604	0.0036	0.8847	0.9527
0.0044	8.0	1834	0.0032	0.8961	0.9469
0.0032	8.9989	2063	0.0039	0.8778	0.9144
0.0033	9.9978	2292	0.0028	0.9072	0.9458
0.0028	10.9967	2521	0.0025	0.9144	0.9593
0.0029	12.0	2751	0.0028	0.9069	0.9329
0.0029	12.9989	2980	0.0025	0.9148	0.9617
0.0027	13.9978	3209	0.0026	0.9130	0.9508
0.0024	14.9967	3438	0.0021	0.9255	0.9552
0.0023	16.0	3668	0.0034	0.8896	0.9616
0.0028	16.9989	3897	0.0029	0.9028	0.9420
0.0029	17.9978	4126	0.0022	0.9235	0.9508
0.0025	18.9967	4355	0.0021	0.9260	0.9621
0.0023	20.0	4585	0.0020	0.9306	0.9591
0.0022	20.9989	4814	0.0019	0.9334	0.9668
0.0023	21.9978	5043	0.0019	0.9318	0.9649
0.003	22.9967	5272	0.0021	0.9274	0.9517
0.0019	24.0	5502	0.0018	0.9363	0.9670
0.002	24.9989	5731	0.0018	0.9370	0.9571
0.0022	25.9978	5960	0.0019	0.9330	0.9558
0.0021	26.9967	6189	0.0018	0.9359	0.9593
0.0018	28.0	6419	0.0016	0.9421	0.9625
0.0017	28.9989	6648	0.0016	0.9447	0.9650
0.0016	29.9978	6877	0.0015	0.9452	0.9651
0.0017	30.9967	7106	0.0015	0.9478	0.9692
0.0016	32.0	7336	0.0014	0.9503	0.9697
0.0015	32.9989	7565	0.0014	0.9512	0.9720
0.0014	33.9978	7794	0.0013	0.9531	0.9721
0.0015	34.9967	8023	0.0013	0.9547	0.9716
0.0013	36.0	8253	0.0013	0.9542	0.9683
0.0014	36.9989	8482	0.0012	0.9573	0.9750
0.0013	37.9978	8711	0.0012	0.9579	0.9768
0.0014	38.9967	8940	0.0012	0.9582	0.9754
0.0014	39.9564	9160	0.0012	0.9591	0.9785

Framework versions

• Transformers 4.42.3
• Pytorch 2.1.2
• Datasets 2.20.0
• Tokenizers 0.19.1