convnextv2-pico-1k-224-finetuned-galaxy10-decals
This model is a fine-tuned version of facebook/convnextv2-pico-1k-224 on the matthieulel/galaxy10_decals dataset. It achieves the following results on the evaluation set:
- Loss: 0.5795
- Accuracy: 0.8546
- Precision: 0.8565
- Recall: 0.8546
- F1: 0.8545
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: 5e-05
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 256
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 30
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Precision | Recall | F1 |
|---|---|---|---|---|---|---|---|
| 1.0752 | 0.99 | 62 | 0.9584 | 0.6697 | 0.6820 | 0.6697 | 0.6581 |
| 0.814 | 2.0 | 125 | 0.6716 | 0.7728 | 0.7780 | 0.7728 | 0.7695 |
| 0.7289 | 2.99 | 187 | 0.6071 | 0.7959 | 0.8093 | 0.7959 | 0.7943 |
| 0.6444 | 4.0 | 250 | 0.5873 | 0.8055 | 0.8103 | 0.8055 | 0.8019 |
| 0.5855 | 4.99 | 312 | 0.5889 | 0.8106 | 0.8226 | 0.8106 | 0.8106 |
| 0.5778 | 6.0 | 375 | 0.5039 | 0.8281 | 0.8321 | 0.8281 | 0.8274 |
| 0.5575 | 6.99 | 437 | 0.5162 | 0.8140 | 0.8235 | 0.8140 | 0.8148 |
| 0.5011 | 8.0 | 500 | 0.5369 | 0.8207 | 0.8234 | 0.8207 | 0.8205 |
| 0.4968 | 8.99 | 562 | 0.5152 | 0.8292 | 0.8282 | 0.8292 | 0.8270 |
| 0.4593 | 10.0 | 625 | 0.4854 | 0.8382 | 0.8408 | 0.8382 | 0.8367 |
| 0.4442 | 10.99 | 687 | 0.4923 | 0.8416 | 0.8423 | 0.8416 | 0.8411 |
| 0.4071 | 12.0 | 750 | 0.5312 | 0.8377 | 0.8356 | 0.8377 | 0.8331 |
| 0.4057 | 12.99 | 812 | 0.4954 | 0.8433 | 0.8449 | 0.8433 | 0.8416 |
| 0.4074 | 14.0 | 875 | 0.4735 | 0.8534 | 0.8509 | 0.8534 | 0.8493 |
| 0.3709 | 14.99 | 937 | 0.4977 | 0.8461 | 0.8450 | 0.8461 | 0.8442 |
| 0.3467 | 16.0 | 1000 | 0.5364 | 0.8286 | 0.8278 | 0.8286 | 0.8274 |
| 0.3129 | 16.99 | 1062 | 0.5695 | 0.8422 | 0.8413 | 0.8422 | 0.8376 |
| 0.3242 | 18.0 | 1125 | 0.5131 | 0.8455 | 0.8469 | 0.8455 | 0.8450 |
| 0.3046 | 18.99 | 1187 | 0.5553 | 0.8399 | 0.8382 | 0.8399 | 0.8371 |
| 0.2805 | 20.0 | 1250 | 0.5871 | 0.8523 | 0.8532 | 0.8523 | 0.8456 |
| 0.2776 | 20.99 | 1312 | 0.5428 | 0.8433 | 0.8404 | 0.8433 | 0.8404 |
| 0.2975 | 22.0 | 1375 | 0.5624 | 0.8393 | 0.8344 | 0.8393 | 0.8359 |
| 0.268 | 22.99 | 1437 | 0.5485 | 0.8495 | 0.8518 | 0.8495 | 0.8498 |
| 0.2535 | 24.0 | 1500 | 0.6135 | 0.8382 | 0.8367 | 0.8382 | 0.8358 |
| 0.2543 | 24.99 | 1562 | 0.6103 | 0.8393 | 0.8389 | 0.8393 | 0.8375 |
| 0.2283 | 26.0 | 1625 | 0.5639 | 0.8484 | 0.8499 | 0.8484 | 0.8480 |
| 0.2341 | 26.99 | 1687 | 0.5795 | 0.8546 | 0.8565 | 0.8546 | 0.8545 |
| 0.2404 | 28.0 | 1750 | 0.5794 | 0.8534 | 0.8515 | 0.8534 | 0.8511 |
| 0.2168 | 28.99 | 1812 | 0.5652 | 0.8546 | 0.8525 | 0.8546 | 0.8524 |
| 0.2057 | 29.76 | 1860 | 0.5650 | 0.8546 | 0.8519 | 0.8546 | 0.8518 |
Framework versions
- Transformers 4.37.2
- Pytorch 2.3.0
- Datasets 2.19.1
- Tokenizers 0.15.1