Upload README.md with huggingface_hub

README.md

@@ -37,6 +37,11 @@ Install requirements and import necessary packages
 
 
 
+```python
+
+```
+
+
 ```python
 # Imports
 import torch
@@ -83,6 +88,9 @@ preprocess = Compose([
 ])
 ```
 
+    monai.transforms.croppad.array CropForeground.__init__:allow_smaller: Current default value of argument `allow_smaller=True` has been deprecated since version 1.2. It will be changed to `allow_smaller=False` in version 1.5.
+
+
 ## Run Inference
 Process an input CT scan and extract features
 
@@ -116,14 +124,15 @@ print(output.shape)
     torch.Size([2, 227, 181, 258])
 
 
-## Fine-tuning instructions
-
-The above model does not have a trained decoder which means the predictions you will get are nonsensical. 
-
-You can however use the pre-trained encoder and the model architecture to finetune on your own datasets - especially if they are small sized. 
+## Fine-tuning Instructions
 
-A very simple way to fit this into your pipelines is to take the model loaded above using model = SegResNet.from_pretrained('project-lighter/ct_fm_segresnet') and replace the model in your training pipeline with this. 
+The model above does not include a trained decoder, which means the predictions you receive will be nonsensical.
 
-Using Auto3DSeg with our model is the recommended approach, follow the instructions here: https://project-lighter.github.io/CT-FM/replication-guide/downstream/#tumor-segmentation-with-auto3dseg
+However, you can leverage the pre-trained encoder and model architecture to fine-tune on your own datasets—especially if they are small. A simple way to integrate this into your pipeline is to replace the model in your training process with the pre-trained version. For example:
+```python
+    model = SegResNet.from_pretrained('project-lighter/ct_fm_segresnet')
+```
+We recommend using Auto3DSeg in conjunction with our model. For detailed guidance, please refer to the instructions here:  
+https://project-lighter.github.io/CT-FM/replication-guide/downstream/#tumor-segmentation-with-auto3dseg