metadata
license: other
base_model: nvidia/segformer-b0-finetuned-ade-512-512
tags:
- vision
- image-segmentation
- generated_from_trainer
metrics:
- precision
model-index:
- name: segformer-b0-finetuned-segments-pv_v1_x3_normalized_p100_4batch
results: []
segformer-b0-finetuned-segments-pv_v1_x3_normalized_p100_4batch
This model is a fine-tuned version of nvidia/segformer-b0-finetuned-ade-512-512 on the mouadenna/satellite_PV_dataset_train_test_v1 dataset. It achieves the following results on the evaluation set:
- Loss: 0.0056
- Mean Iou: 0.8288
- Precision: 0.8928
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: 0.0004
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.001
- num_epochs: 40
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Precision |
|---|---|---|---|---|---|
| 0.0086 | 0.9993 | 687 | 0.0068 | 0.8080 | 0.8515 |
| 0.0061 | 2.0 | 1375 | 0.0056 | 0.8257 | 0.8862 |
| 0.0058 | 2.9993 | 2062 | 0.0056 | 0.8284 | 0.9154 |
| 0.0063 | 4.0 | 2750 | 0.0055 | 0.8212 | 0.9261 |
| 0.0051 | 4.9993 | 3437 | 0.0081 | 0.7851 | 0.9189 |
| 0.0042 | 6.0 | 4125 | 0.0062 | 0.8322 | 0.9034 |
| 0.004 | 6.9993 | 4812 | 0.0067 | 0.8262 | 0.8807 |
| 0.0049 | 8.0 | 5500 | 0.0061 | 0.8271 | 0.9135 |
| 0.0043 | 8.9993 | 6187 | 0.0056 | 0.8288 | 0.8928 |
Framework versions
- Transformers 4.42.3
- Pytorch 2.1.2
- Datasets 2.20.0
- Tokenizers 0.19.1