Update best model - Epoch 0 - Acc: 0.0892
Browse files- README.md +15 -8
- config.json +6 -6
- model.safetensors +2 -2
README.md
CHANGED
|
@@ -18,8 +18,8 @@ for vision tasks. Trained on CIFAR-10.
|
|
| 18 |
|
| 19 |
- **Architecture**: Vision Transformer with fractal positional encoding
|
| 20 |
- **Dataset**: CIFAR-100 (100 classes)
|
| 21 |
-
- **Embedding Dimension**:
|
| 22 |
-
- **Depth**:
|
| 23 |
- **Patch Size**: 4x4
|
| 24 |
- **PE Levels**: 12
|
| 25 |
- **Simplex Dimension**: 5-simplex
|
|
@@ -27,8 +27,8 @@ for vision tasks. Trained on CIFAR-10.
|
|
| 27 |
## Training
|
| 28 |
|
| 29 |
- **Dataset**: CIFAR-100
|
| 30 |
-
- **Epochs**:
|
| 31 |
-
- **Best Accuracy**: 0.
|
| 32 |
- **Batch Size**: 512
|
| 33 |
- **Learning Rate**: 0.001
|
| 34 |
|
|
@@ -44,17 +44,24 @@ for vision tasks. Trained on CIFAR-10.
|
|
| 44 |
```python
|
| 45 |
from geovocab2.train.model.vit_beatrix import SimplifiedGeometricClassifier
|
| 46 |
from safetensors.torch import load_file
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 47 |
|
| 48 |
# Load model
|
| 49 |
model = SimplifiedGeometricClassifier(
|
| 50 |
num_classes=100, # CIFAR-100
|
| 51 |
img_size=32,
|
| 52 |
-
embed_dim=
|
| 53 |
-
depth=
|
| 54 |
)
|
| 55 |
|
| 56 |
-
# Load weights
|
| 57 |
-
state_dict = load_file(
|
| 58 |
model.load_state_dict(state_dict)
|
| 59 |
model.eval()
|
| 60 |
|
|
|
|
| 18 |
|
| 19 |
- **Architecture**: Vision Transformer with fractal positional encoding
|
| 20 |
- **Dataset**: CIFAR-100 (100 classes)
|
| 21 |
+
- **Embedding Dimension**: 256
|
| 22 |
+
- **Depth**: 12 layers
|
| 23 |
- **Patch Size**: 4x4
|
| 24 |
- **PE Levels**: 12
|
| 25 |
- **Simplex Dimension**: 5-simplex
|
|
|
|
| 27 |
## Training
|
| 28 |
|
| 29 |
- **Dataset**: CIFAR-100
|
| 30 |
+
- **Epochs**: 0
|
| 31 |
+
- **Best Accuracy**: 0.0892
|
| 32 |
- **Batch Size**: 512
|
| 33 |
- **Learning Rate**: 0.001
|
| 34 |
|
|
|
|
| 44 |
```python
|
| 45 |
from geovocab2.train.model.vit_beatrix import SimplifiedGeometricClassifier
|
| 46 |
from safetensors.torch import load_file
|
| 47 |
+
from huggingface_hub import hf_hub_download
|
| 48 |
+
|
| 49 |
+
# Download weights from Hub
|
| 50 |
+
weights_path = hf_hub_download(
|
| 51 |
+
repo_id="AbstractPhil/vit-beatrix",
|
| 52 |
+
filename="weights/beatrix-cifar100/20251007_181653/model.safetensors"
|
| 53 |
+
)
|
| 54 |
|
| 55 |
# Load model
|
| 56 |
model = SimplifiedGeometricClassifier(
|
| 57 |
num_classes=100, # CIFAR-100
|
| 58 |
img_size=32,
|
| 59 |
+
embed_dim=256,
|
| 60 |
+
depth=12
|
| 61 |
)
|
| 62 |
|
| 63 |
+
# Load weights
|
| 64 |
+
state_dict = load_file(weights_path)
|
| 65 |
model.load_state_dict(state_dict)
|
| 66 |
model.eval()
|
| 67 |
|
config.json
CHANGED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"num_classes": 100,
|
| 3 |
"img_size": 32,
|
| 4 |
"patch_size": 4,
|
| 5 |
-
"embed_dim":
|
| 6 |
"k_simplex": 5,
|
| 7 |
-
"depth":
|
| 8 |
"num_heads": 8,
|
| 9 |
"mlp_ratio": 4.0,
|
| 10 |
"dropout": 0.0,
|
|
@@ -13,10 +13,10 @@
|
|
| 13 |
"pe_smooth_tau": 0.25,
|
| 14 |
"simplex_feature_weight": 0.2,
|
| 15 |
"batch_size": 512,
|
| 16 |
-
"num_epochs":
|
| 17 |
"learning_rate": 0.001,
|
| 18 |
"weight_decay": 0.005,
|
| 19 |
-
"warmup_epochs":
|
| 20 |
"task_loss_weight": 1.0,
|
| 21 |
"flow_loss_weight": 0.5,
|
| 22 |
"coherence_loss_weight": 0.3,
|
|
@@ -34,10 +34,10 @@
|
|
| 34 |
"timestamp_dirs": true,
|
| 35 |
"push_to_hub": true,
|
| 36 |
"hub_model_id": "AbstractPhil/vit-beatrix",
|
|
|
|
| 37 |
"hub_upload_best_only": true,
|
| 38 |
"use_tensorboard": true,
|
| 39 |
"log_dir": "./logs",
|
| 40 |
"log_every": 50,
|
| 41 |
-
"train_baseline": false
|
| 42 |
-
"hub_token": null
|
| 43 |
}
|
|
|
|
| 2 |
"num_classes": 100,
|
| 3 |
"img_size": 32,
|
| 4 |
"patch_size": 4,
|
| 5 |
+
"embed_dim": 256,
|
| 6 |
"k_simplex": 5,
|
| 7 |
+
"depth": 12,
|
| 8 |
"num_heads": 8,
|
| 9 |
"mlp_ratio": 4.0,
|
| 10 |
"dropout": 0.0,
|
|
|
|
| 13 |
"pe_smooth_tau": 0.25,
|
| 14 |
"simplex_feature_weight": 0.2,
|
| 15 |
"batch_size": 512,
|
| 16 |
+
"num_epochs": 10,
|
| 17 |
"learning_rate": 0.001,
|
| 18 |
"weight_decay": 0.005,
|
| 19 |
+
"warmup_epochs": 2,
|
| 20 |
"task_loss_weight": 1.0,
|
| 21 |
"flow_loss_weight": 0.5,
|
| 22 |
"coherence_loss_weight": 0.3,
|
|
|
|
| 34 |
"timestamp_dirs": true,
|
| 35 |
"push_to_hub": true,
|
| 36 |
"hub_model_id": "AbstractPhil/vit-beatrix",
|
| 37 |
+
"hub_model_name": "beatrix-cifar100",
|
| 38 |
"hub_upload_best_only": true,
|
| 39 |
"use_tensorboard": true,
|
| 40 |
"log_dir": "./logs",
|
| 41 |
"log_every": 50,
|
| 42 |
+
"train_baseline": false
|
|
|
|
| 43 |
}
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1f25f11e2851e52c4ee631b191044282ff7e23236f4b6a7fed2388d32f045a29
|
| 3 |
+
size 38117196
|