metadata
library_name: transformers
license: apache-2.0
base_model: facebook/detr-resnet-50
tags:
- generated_from_trainer
model-index:
- name: output
results: []
output
This model is a fine-tuned version of facebook/detr-resnet-50 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.9020
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: 0.0001
- 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: cosine
- lr_scheduler_warmup_steps: 375
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.8809 | 1.0 | 250 | 1.6853 |
| 1.4938 | 2.0 | 500 | 1.3327 |
| 1.3403 | 3.0 | 750 | 1.2940 |
| 1.3485 | 4.0 | 1000 | 1.2512 |
| 1.2645 | 5.0 | 1250 | 1.2189 |
| 1.2061 | 6.0 | 1500 | 1.1517 |
| 1.1632 | 7.0 | 1750 | 1.1042 |
| 1.179 | 8.0 | 2000 | 1.0796 |
| 1.1558 | 9.0 | 2250 | 1.0567 |
| 1.1294 | 10.0 | 2500 | 1.0661 |
| 1.0644 | 11.0 | 2750 | 1.0214 |
| 1.0545 | 12.0 | 3000 | 1.0508 |
| 1.0455 | 13.0 | 3250 | 0.9904 |
| 1.0242 | 14.0 | 3500 | 0.9863 |
| 1.0021 | 15.0 | 3750 | 0.9897 |
| 0.9947 | 16.0 | 4000 | 0.9774 |
| 0.9456 | 17.0 | 4250 | 0.9485 |
| 0.9339 | 18.0 | 4500 | 0.9553 |
| 0.9481 | 19.0 | 4750 | 0.9431 |
| 0.8992 | 20.0 | 5000 | 0.9348 |
| 0.8792 | 21.0 | 5250 | 0.9265 |
| 0.8946 | 22.0 | 5500 | 0.9271 |
| 0.8913 | 23.0 | 5750 | 0.9249 |
| 0.8813 | 24.0 | 6000 | 0.9072 |
| 0.885 | 25.0 | 6250 | 0.9115 |
| 0.8477 | 26.0 | 6500 | 0.9091 |
| 0.8792 | 27.0 | 6750 | 0.9045 |
| 0.8816 | 28.0 | 7000 | 0.9034 |
| 0.8633 | 29.0 | 7250 | 0.9005 |
| 0.8921 | 30.0 | 7500 | 0.9102 |
Framework versions
- Transformers 4.51.3
- Pytorch 2.6.0+cu124
- Datasets 3.5.0
- Tokenizers 0.21.1