Update README.md
Browse files
README.md
CHANGED
|
@@ -42,8 +42,29 @@ python -m pip install safetensors
|
|
| 42 |
### Load Model Checkpoint
|
| 43 |
|
| 44 |
```python
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 45 |
```
|
| 46 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 47 |
## Acknowledgements
|
| 48 |
|
| 49 |
This work is supported by the Helmholtz Association Initiative and Networking Fund through the Helmholtz AI platform, and the HAICORE@KIT grant.
|
|
|
|
| 42 |
### Load Model Checkpoint
|
| 43 |
|
| 44 |
```python
|
| 45 |
+
from safetensors.torch import load_file
|
| 46 |
+
|
| 47 |
+
model = OptoGPT()
|
| 48 |
+
|
| 49 |
+
safetensors_path = "optogpt-model.safetensors"
|
| 50 |
+
state_dict = load_file(safetensors_path)
|
| 51 |
+
model.load_state_dict(state_dict)
|
| 52 |
```
|
| 53 |
|
| 54 |
+
## Useful Information
|
| 55 |
+
|
| 56 |
+
| Stat | Value |
|
| 57 |
+
| :------------------ | ----------: |
|
| 58 |
+
| #Parameters. | 108,381,113 |
|
| 59 |
+
| Best validation MAE | 0.0379 |
|
| 60 |
+
| Epochs trained | 1,000 |
|
| 61 |
+
| Best epoch. | 996 |
|
| 62 |
+
| Batch size | 256 |
|
| 63 |
+
| n_blocks | 6 |
|
| 64 |
+
| n_heads | 8 |
|
| 65 |
+
| d_model | 1,024 |
|
| 66 |
+
| max_seq_length | 20 |
|
| 67 |
+
|
| 68 |
## Acknowledgements
|
| 69 |
|
| 70 |
This work is supported by the Helmholtz Association Initiative and Networking Fund through the Helmholtz AI platform, and the HAICORE@KIT grant.
|