segformer-b5-finetuned-segments-chargers-full-v5.1

This model is a fine-tuned version of nvidia/mit-b5 on the dskong07/chargers-full-v0.1 dataset. It achieves the following results on the evaluation set:

• Loss: 0.3962
• Mean Iou: 0.7853
• Mean Accuracy: 0.8809
• Overall Accuracy: 0.9180
• Accuracy Unlabeled: nan
• Accuracy Screen: 0.8689
• Accuracy Body: 0.9342
• Accuracy Cable: 0.7464
• Accuracy Plug: 0.9272
• Accuracy Void-background: 0.9277
• Iou Unlabeled: nan
• Iou Screen: 0.7716
• Iou Body: 0.8025
• Iou Cable: 0.6299
• Iou Plug: 0.8233
• Iou Void-background: 0.8993

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: 16
• eval_batch_size: 16
• 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: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Unlabeled	Accuracy Screen	Accuracy Body	Accuracy Cable	Accuracy Plug	Accuracy Void-background	Iou Unlabeled	Iou Screen	Iou Body	Iou Cable	Iou Plug	Iou Void-background
0.6293	2.2222	20	0.9773	0.6257	0.7646	0.8358	nan	0.7319	0.9339	0.4731	0.8530	0.8314	nan	0.6239	0.6559	0.3567	0.6921	0.8000
0.4226	4.4444	40	0.5124	0.7130	0.8278	0.8860	nan	0.8381	0.9318	0.5989	0.8755	0.8949	nan	0.7153	0.7396	0.4883	0.7585	0.8632
0.2018	6.6667	60	0.4182	0.7468	0.8496	0.9023	nan	0.8413	0.9117	0.6632	0.9111	0.9209	nan	0.7288	0.7664	0.5540	0.8020	0.8828
0.1678	8.8889	80	0.4337	0.7622	0.8648	0.9083	nan	0.8578	0.9141	0.7229	0.9038	0.9252	nan	0.7428	0.7780	0.5838	0.8164	0.8899
0.1711	11.1111	100	0.3718	0.7632	0.8733	0.9096	nan	0.8756	0.9143	0.7393	0.9143	0.9228	nan	0.7393	0.7856	0.5889	0.8108	0.8917
0.1385	13.3333	120	0.4243	0.7655	0.8642	0.9087	nan	0.8161	0.9245	0.7292	0.9288	0.9221	nan	0.7507	0.7793	0.5982	0.8105	0.8890
0.1582	15.5556	140	0.4018	0.7712	0.8737	0.9117	nan	0.8464	0.9394	0.7324	0.9330	0.9175	nan	0.7659	0.7914	0.6003	0.8065	0.8921
0.1249	17.7778	160	0.4037	0.7726	0.8692	0.9142	nan	0.8531	0.9299	0.7032	0.9326	0.9270	nan	0.7503	0.7948	0.6020	0.8198	0.8963
0.1239	20.0	180	0.4015	0.7768	0.8752	0.9140	nan	0.8574	0.9293	0.7190	0.9470	0.9233	nan	0.7674	0.7912	0.6079	0.8228	0.8948
0.0895	22.2222	200	0.4355	0.7753	0.8771	0.9134	nan	0.8245	0.9352	0.7656	0.9393	0.9208	nan	0.7559	0.7914	0.6172	0.8167	0.8951
0.1014	24.4444	220	0.4007	0.7802	0.8862	0.9160	nan	0.8555	0.9317	0.7764	0.9452	0.9221	nan	0.7648	0.8008	0.6235	0.8141	0.8981
0.0782	26.6667	240	0.3852	0.7821	0.8763	0.9176	nan	0.8749	0.9311	0.7368	0.9079	0.9310	nan	0.7729	0.8030	0.6184	0.8164	0.8996
0.076	28.8889	260	0.4259	0.7767	0.8762	0.9148	nan	0.8671	0.9390	0.7260	0.9271	0.9220	nan	0.7621	0.7967	0.6144	0.8142	0.8963
0.0752	31.1111	280	0.4058	0.7850	0.8828	0.9184	nan	0.8832	0.9272	0.7436	0.9305	0.9295	nan	0.7726	0.8029	0.6235	0.8250	0.9007
0.0674	33.3333	300	0.3960	0.7845	0.8816	0.9179	nan	0.8838	0.9252	0.7370	0.9324	0.9297	nan	0.7732	0.8027	0.6254	0.8218	0.8993
0.0873	35.5556	320	0.4026	0.7864	0.8843	0.9177	nan	0.8807	0.9296	0.7488	0.9358	0.9265	nan	0.7827	0.8023	0.6274	0.8217	0.8978
0.0819	37.7778	340	0.4422	0.7832	0.8785	0.9170	nan	0.8622	0.9365	0.7359	0.9320	0.9258	nan	0.7770	0.7996	0.6221	0.8192	0.8983
0.1206	40.0	360	0.4247	0.7825	0.8767	0.9168	nan	0.8733	0.9329	0.7202	0.9297	0.9273	nan	0.7714	0.7987	0.6216	0.8233	0.8977
0.076	42.2222	380	0.3981	0.7861	0.8848	0.9174	nan	0.8767	0.9326	0.7508	0.9393	0.9246	nan	0.7790	0.8006	0.6283	0.8245	0.8979
0.056	44.4444	400	0.4297	0.7845	0.8842	0.9170	nan	0.8762	0.9359	0.7425	0.9439	0.9227	nan	0.7757	0.8010	0.6285	0.8196	0.8976
0.0666	46.6667	420	0.3950	0.7839	0.8793	0.9174	nan	0.8566	0.9372	0.7452	0.9313	0.9262	nan	0.7682	0.8010	0.6305	0.8213	0.8987
0.0689	48.8889	440	0.3962	0.7853	0.8809	0.9180	nan	0.8689	0.9342	0.7464	0.9272	0.9277	nan	0.7716	0.8025	0.6299	0.8233	0.8993

Framework versions

• Transformers 4.48.3
• Pytorch 2.6.0+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0