segformer-b0-finetuned-segments-sidewalk-oct-22

This model is a fine-tuned version of nvidia/mit-b0 on the maratuly/Pseudo-echo dataset. It achieves the following results on the evaluation set:

• Loss: 0.3463
• Mean Iou: 0.7622
• Mean Accuracy: 0.9528
• Overall Accuracy: 0.9581
• Accuracy Unlabeled: nan
• Accuracy Lv: 0.9931
• Accuracy Rv: 0.9354
• Accuracy La: 0.9533
• Accuracy Ra: 0.9293
• Iou Unlabeled: 0.0
• Iou Lv: 0.9931
• Iou Rv: 0.9354
• Iou La: 0.9533
• Iou Ra: 0.9293

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: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Unlabeled	Accuracy Lv	Accuracy Rv	Accuracy La	Accuracy Ra	Iou Unlabeled	Iou Lv	Iou Rv	Iou La	Iou Ra
1.0046	5.0	20	1.3540	0.4931	0.7263	0.7606	nan	0.9742	0.6789	0.5862	0.6657	0.0	0.6545	0.6789	0.5540	0.5783
0.7304	10.0	40	0.8262	0.6901	0.8908	0.9017	nan	0.9906	0.8147	0.9192	0.8387	0.0	0.9028	0.8145	0.9046	0.8287
0.5957	15.0	60	0.5683	0.7395	0.9277	0.9367	nan	0.9896	0.9141	0.9225	0.8845	0.0	0.9766	0.9141	0.9225	0.8845
0.5094	20.0	80	0.4881	0.7541	0.9426	0.9499	nan	0.9909	0.9368	0.9312	0.9116	0.0	0.9909	0.9368	0.9312	0.9116
0.4681	25.0	100	0.4384	0.7689	0.9612	0.9653	nan	0.9942	0.9434	0.9650	0.9421	0.0	0.9942	0.9434	0.9650	0.9421
0.4045	30.0	120	0.4078	0.7639	0.9549	0.9602	nan	0.9933	0.9418	0.9519	0.9328	0.0	0.9933	0.9418	0.9519	0.9328
0.3956	35.0	140	0.3844	0.7664	0.9580	0.9625	nan	0.9939	0.9406	0.9551	0.9425	0.0	0.9939	0.9406	0.9551	0.9425
0.3736	40.0	160	0.3736	0.7687	0.9609	0.9652	nan	0.9961	0.9409	0.9631	0.9436	0.0	0.9961	0.9409	0.9631	0.9436
0.3431	45.0	180	0.3528	0.7622	0.9528	0.9577	nan	0.9923	0.9321	0.9539	0.9327	0.0	0.9923	0.9321	0.9539	0.9327
0.3428	50.0	200	0.3463	0.7622	0.9528	0.9581	nan	0.9931	0.9354	0.9533	0.9293	0.0	0.9931	0.9354	0.9533	0.9293

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.15.0