Deepglobe_segformer_3

This model is a fine-tuned version of nvidia/mit-b0 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8406
• Mean Iou: 0.4620
• Mean Accuracy: 0.5756
• Overall Accuracy: 0.8155
• Per Category Iou: [0.6433628051500885, 0.8443710072700582, 0.05868084899551419, 0.6663364807256275, 0.5090198981201258, 0.5124226283624453, 0.0]
• Per Category Accuracy: [0.8883659557544211, 0.9159453906069126, 0.06770783080218444, 0.7777261715162402, 0.5203630889379521, 0.85919641595548, 0.0]

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Per Category Iou	Per Category Accuracy
1.4727	0.49	20	1.5749	0.2930	0.4344	0.7062	[0.47121379062001395, 0.7670396913155827, 0.09819894521882991, 0.5227229243539695, 0.0031194487593101524, 0.18902045042309878, 0.0]	[0.9424700572966612, 0.8218753490741639, 0.17839715076899326, 0.8786984834251509, 0.003125783241778855, 0.21592899892395664, 0.0]
1.1746	0.98	40	1.0074	0.3675	0.4978	0.7835	[0.6335804347395366, 0.81926412147752, 0.042360636232568284, 0.5516872946875934, 0.06391535031461955, 0.46156523081466366, 0.0]	[0.8835907055203469, 0.8959329160865661, 0.04826409660176376, 0.9133718981931649, 0.0641168704492156, 0.6795223711292329, 0.0]
0.9123	1.46	60	0.8961	0.4223	0.5476	0.8032	[0.6331637378113184, 0.8287686244851504, 0.06308541030241316, 0.6147403096624029, 0.27778067188553224, 0.5387135123290364, 0.0]	[0.8993383411101392, 0.8976098372522392, 0.07531755948360704, 0.9001059260136011, 0.28091500200350256, 0.7797697919353069, 0.0]
1.0879	1.95	80	0.8875	0.4412	0.5616	0.8054	[0.6004683670971223, 0.8332655416128979, 0.04572839646818258, 0.6549740332584124, 0.44062035905035596, 0.5130410733489108, 0.0]	[0.9154682866219158, 0.9036612725247445, 0.052284546805349184, 0.8106196696766435, 0.44475361708037797, 0.8046768856202466, 0.0]
1.2142	2.44	100	0.8355	0.4533	0.5722	0.8077	[0.6298554391449935, 0.8369662549593841, 0.050673209170535985, 0.6649442110070505, 0.4958035295244105, 0.49513095742822216, 0.0]	[0.8994176115323517, 0.902362104312013, 0.05909947222168619, 0.790401335237105, 0.5042568442360589, 0.8498825640463182, 0.0]
0.884	2.93	120	0.8406	0.4620	0.5756	0.8155	[0.6433628051500885, 0.8443710072700582, 0.05868084899551419, 0.6663364807256275, 0.5090198981201258, 0.5124226283624453, 0.0]	[0.8883659557544211, 0.9159453906069126, 0.06770783080218444, 0.7777261715162402, 0.5203630889379521, 0.85919641595548, 0.0]

Framework versions

• Transformers 4.31.0
• Pytorch 2.0.0+cu117
• Datasets 2.10.1
• Tokenizers 0.13.3