mobilenet_v2 trained on dushj98/aerial_real_only
This model is a fine-tuned version of mobilenet_v2 on the dushj98/aerial_real_only dataset.
Model Details
- Model Type: mobilenet_v2
- Training Dataset: dushj98/aerial_real_only
- Number of classes: 13
Training Metadata
- model: mobilenet_v2
- epochs: 25
- lr: 0.0005
- batch_size: 64
- dataset: dushj98/aerial_real_only
- random_seed: 42
- feature_extraction: False
- training_mode: ft
- data_augmentation: True
- num_classes: 13
- patience: 10
- scheduler: cosine
- best_val_acc: 0.48502994011976047
- best_val_loss: 1.5764319687491462
- best_epoch: 1
- val_acc_history: [0.4682174113311838, 0.48502994011976047, 0.4601566098572087, 0.4474896361123906, 0.4553201289728236, 0.45808383233532934, 0.4553201289728236, 0.4714417319207738, 0.46453247351450944, 0.4774297558728696, 0.4771994472593275, 0.4778903730999539]
- train_time: 403.73659375100397
- lr_history: [0.0005, 0.0004980483885753333, 0.0004922243323793362, 0.0004826196802573422, 0.0004693859033108562, 0.00045273170610779945, 0.0004329197352867993, 0.0004102624374628006, 0.00038511713175730155, 0.0003578803746623554, 0.0003289817061077994, 0.0002988768753599668]
- Downloads last month
- 185
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support