File size: 395 Bytes
9698229 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 |
---
base_model:
- openai/clip-vit-large-patch14
datasets:
- dpdl-benchmark/oxford_flowers102
metrics:
- accuracy
---
# Model Card
## Training Details
Adam Optimizer with a constant learning rate 1e-5 for 4000 steps training (batch_size=128).
Only the vision encoder is fine-tuned.
## Evaluation Results
Test set accuracy:
- pre-trained: 0.7924866080284119
- fine-tuned: 0.9767441749572754
|