Update README.md

README.md

@@ -8,21 +8,19 @@ license: apache-2.0
 ---
 # Model card for convnextv2-base-k5-in1k-128
 
-ConvNeXtV2-Base trained from scratch on ImageNet-1K @128x128 (k=5, 300ep, global batch 4096) using ConvNeXt (Liu et al., 2022, Table 5) supervised settings.
+ConvNeXtV2-Base (k=5) trained from scratch on ImageNet-1K @128x128.
+
+## Training Recipe: 
+- ConvNeXt supervised recipe -- Table 5.
+- BF16 AMP
 
 ## Model Details
-- **Run Dir:** /mnt/data/training-runs/convnextv2_base-128-k5-sup-e300-bf16-gb4096
-- **Experiment:** convnextv2_base-128-k5-sup-e300-bf16-gb4096
 - **Model:** convnextv2_base
 - **Model Kwargs:**
   - kernel_sizes: 5
 - **Input Size:**
-  - 3
-  - 128
-  - 128
-- **Checkpoint:** model_best.pth.tar
-- **Weights Key:** state_dict_ema
-- **Best Metric:** 82.09399995224
+  - 3x128x128
+- **Top-1 Accuracy:** 82.09399995224
 
 ## Model Usage
 ```python
@@ -30,4 +28,3 @@ import timm
 model = timm.create_model('hf-hub:PRadecki/convnextv2-base-k5-in1k-128', pretrained=True)
 model.eval()
 ```
-