segformer_b1_finetuned_segment_pv_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: 0.0072
• Mean Iou: 0.8692
• Precision: 0.9162

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: 4e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 40
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Precision
0.4877	1.0	917	0.2178	0.5326	0.6037
0.0992	2.0	1834	0.0258	0.7353	0.7913
0.022	3.0	2751	0.0119	0.7764	0.9267
0.0106	4.0	3668	0.0084	0.8093	0.8316
0.0069	5.0	4585	0.0070	0.8327	0.8698
0.0054	6.0	5502	0.0057	0.8431	0.8829
0.0044	7.0	6419	0.0055	0.8494	0.8877
0.0039	8.0	7336	0.0059	0.8474	0.9002
0.0036	9.0	8253	0.0056	0.8505	0.8894
0.0031	10.0	9170	0.0054	0.8575	0.8994
0.0031	11.0	10087	0.0051	0.8620	0.9190
0.0027	12.0	11004	0.0053	0.8646	0.9043
0.0026	13.0	11921	0.0059	0.8643	0.9275
0.0028	14.0	12838	0.0055	0.8641	0.9131
0.0025	15.0	13755	0.0053	0.8661	0.9120
0.0023	16.0	14672	0.0056	0.8644	0.9107
0.0022	17.0	15589	0.0052	0.8668	0.9125
0.0022	18.0	16506	0.0054	0.8700	0.9143
0.0022	19.0	17423	0.0059	0.8685	0.9180
0.002	20.0	18340	0.0057	0.8696	0.9157
0.0019	21.0	19257	0.0061	0.8682	0.9136
0.0019	22.0	20174	0.0069	0.8606	0.9262
0.0019	23.0	21091	0.0062	0.8700	0.9172
0.0018	24.0	22008	0.0061	0.8682	0.9258
0.0019	25.0	22925	0.0062	0.8669	0.9124
0.002	26.0	23842	0.0065	0.8672	0.9206
0.0017	27.0	24759	0.0062	0.8688	0.9108
0.0016	28.0	25676	0.0066	0.8686	0.9154
0.0016	29.0	26593	0.0066	0.8704	0.9175
0.0016	30.0	27510	0.0068	0.8664	0.9127
0.0015	31.0	28427	0.0069	0.8679	0.9150
0.0015	32.0	29344	0.0065	0.8691	0.9196
0.0015	33.0	30261	0.0069	0.8676	0.9130
0.0014	34.0	31178	0.0067	0.8691	0.9164
0.0014	35.0	32095	0.0071	0.8679	0.9153
0.0014	36.0	33012	0.0072	0.8687	0.9127
0.0014	37.0	33929	0.0073	0.8689	0.9151
0.0014	38.0	34846	0.0069	0.8696	0.9183
0.0014	39.0	35763	0.0070	0.8694	0.9163
0.0013	40.0	36680	0.0072	0.8692	0.9162

Framework versions

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