Merge branch 'main' of hf.co:jfang/mars-vit-base-ctx2m

README.md

@@ -18,6 +18,9 @@ Model Card for Mars ViT Base Model
 
 ## Usage Examples
 ### Using timm
+
+First download checkpoint-1199.pth (backbone only)
+
 ```python
 import timm
 import torch
@@ -27,7 +30,7 @@ model = timm.create_model(
     in_chans=1,
     num_classes=0,
     global_pool='',
-    checkpoint_path="https://huggingface.co/jfang/mars-vit-base-ctx2m/resolve/main/checkpoint-1199.pth"
+    checkpoint_path="./checkpoint-1199.pth" # must use local path
 )
 
 model.eval()
@@ -54,8 +57,6 @@ inputs = image_processor(image, return_tensors="pt")
 outputs = model(**inputs)
 ```
 
-### Model Performance
-The model is optimized for feature extraction from CTX images. Detailed performance metrics on specific tasks or datasets are not provided in this card.
 
 ### Limitations
 The model is trained specifically on CTX images and may not generalize well to other types of images without further fine-tuning.

preprocessor_config.json

@@ -0,0 +1,9 @@
+{
+  "do_normalize": true,
+  "do_resize": true,
+  "feature_extractor_type": "ViTFeatureExtractor",
+  "image_mean": [0.5],
+  "image_std": [0.5],
+  "resample": 2,
+  "size": 224
+}