Downloads · 30 days
12
2% of all-time downloads
clip-archdaily-5k
TheLitttleThings/clip-archdaily-5k
clip-archdaily-5k is a zero-shot image classification model from TheLitttleThings. Use it for the zero-shot image classification task on the model card, and read the license before you ship it in a product. It is set up for transformers.
should probably proofread and complete it, then remove this comment. --
All-time downloads
550
Public
Repo size
4.7 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.bin788 MB · 100%
At a glance
- Task
- Zero-Shot Image Classification
- Library
- transformers
- Model type
- clip
- Access
- Public
- Created
- Aug 5, 2022
- Updated
- Aug 9, 2022
- SHA
- ea82e08e
Model card
From the Hugging Face model README
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
clip-archdaily-5k
This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.5681
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: 3e-05
- train_batch_size: 56
- eval_batch_size: 56
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 2.533 | 0.34 | 200 | 2.4607 |
| 2.1012 | 0.68 | 400 | 1.9922 |
| 1.6059 | 1.02 | 600 | 1.7986 |
| 1.4557 | 1.36 | 800 | 1.6130 |
| 1.4268 | 1.7 | 1000 | 1.4073 |
| 0.8588 | 2.04 | 1200 | 1.2657 |
| 0.8191 | 2.38 | 1400 | 1.1214 |
| 0.81 | 2.72 | 1600 | 1.0418 |
| 0.5546 | 3.06 | 1800 | 0.9735 |
| 0.4905 | 3.4 | 2000 | 0.9006 |
| 0.5209 | 3.74 | 2200 | 0.8762 |
| 0.3127 | 4.08 | 2400 | 0.8457 |
| 0.3145 | 4.42 | 2600 | 0.7886 |
| 0.3265 | 4.76 | 2800 | 0.7853 |
| 0.2215 | 5.1 | 3000 | 0.7309 |
| 0.2351 | 5.44 | 3200 | 0.7082 |
| 0.2332 | 5.78 | 3400 | 0.6770 |
| 0.1793 | 6.12 | 3600 | 0.6856 |
| 0.1617 | 6.46 | 3800 | 0.6470 |
| 0.1468 | 6.8 | 4000 | 0.6700 |
| 0.1293 | 7.14 | 4200 | 0.6460 |
| 0.1257 | 7.48 | 4400 | 0.6415 |
| 0.0975 | 7.82 | 4600 | 0.6454 |
| 0.0835 | 8.16 | 4800 | 0.6111 |
| 0.0856 | 8.5 | 5000 | 0.6124 |
| 0.0887 | 8.84 | 5200 | 0.5956 |
| 0.069 | 9.18 | 5400 | 0.5877 |
| 0.0625 | 9.52 | 5600 | 0.5798 |
| 0.0599 | 9.86 | 5800 | 0.5681 |
Framework versions
- Transformers 4.21.1
- Pytorch 1.12.0+cu113
- Datasets 2.4.0
- Tokenizers 0.12.1
- Task
- Zero-Shot Image Classification
- Library
- transformers
- Type
- clip
- Created
- Aug 5, 2022
- Updated
- Aug 9, 2022