Update README.md
Browse files
README.md
CHANGED
@@ -235,19 +235,16 @@ accuracy across different resolutions and representations:
|
|
235 |
|
236 |
1. **Primary Predictions Loss**:
|
237 |
- This term computes the L1 loss between the primary model predictions and the reference values. It ensures that the transformer's outputs
|
238 |
-
|
239 |
|
240 |
2. **Downsampled Predictions Loss**:
|
241 |
- Recognizing the importance of accuracy across varying resolutions, this term calculates the L1 loss between the downsampled versions of the
|
242 |
-
|
243 |
-
|
244 |
|
245 |
3. **Blurred Predictions Loss**:
|
246 |
- To ensure the model's robustness against small perturbations and noise, this term evaluates the L1 loss between blurred versions of the
|
247 |
-
|
248 |
-
|
249 |
-
By combining these loss terms, the Swin2 transformer is trained to produce accurate predictions across different resolutions and under various data transformations,
|
250 |
-
ensuring its versatility and robustness in diverse scenarios.
|
251 |
|
252 |
## Computing Infrastructure
|
253 |
|
|
|
235 |
|
236 |
1. **Primary Predictions Loss**:
|
237 |
- This term computes the L1 loss between the primary model predictions and the reference values. It ensures that the transformer's outputs
|
238 |
+
closely match the ground truth across the primary spatial resolution.
|
239 |
|
240 |
2. **Downsampled Predictions Loss**:
|
241 |
- Recognizing the importance of accuracy across varying resolutions, this term calculates the L1 loss between the downsampled versions of the
|
242 |
+
predictions and the reference values. By incorporating this term, the model is incentivized to preserve critical information even when the data is represented
|
243 |
+
at a coarser scale.
|
244 |
|
245 |
3. **Blurred Predictions Loss**:
|
246 |
- To ensure the model's robustness against small perturbations and noise, this term evaluates the L1 loss between blurred versions of the
|
247 |
+
predictions and the references. This encourages the model to produce predictions that maintain accuracy even under slight modifications in the data representation.
|
|
|
|
|
|
|
248 |
|
249 |
## Computing Infrastructure
|
250 |
|