Upload folder using huggingface_hub

README.md

@@ -9,30 +9,30 @@ This is a classification model trained on CIFAR10, specifically on the `train/`
 
 ### Performance
 
-It achieves **93.9% accuracy** on the `test/` set despite having only **only 97,530 trainable parameters**! Here is a full classification report generated with `sklearn.metrics.classification_report`:
+It achieves **94.6% accuracy** on the `test/` set (98.7% on `train/`) despite having only **only 97,530 trainable parameters**! Here is a full classification report generated with `sklearn.metrics.classification_report`:
 
 ```
               precision    recall  f1-score   support
 
-    airplane      0.945     0.946     0.946      1000
-  automobile      0.942     0.982     0.962      1000
-        bird      0.931     0.918     0.924      1000
-         cat      0.886     0.850     0.868      1000
-        deer      0.947     0.952     0.950      1000
-         dog      0.910     0.893     0.902      1000
-        frog      0.949     0.972     0.960      1000
-       horse      0.952     0.966     0.959      1000
-        ship      0.966     0.960     0.963      1000
-       truck      0.960     0.953     0.956      1000
-
-    accuracy                          0.939     10000
-   macro avg      0.939     0.939     0.939     10000
-weighted avg      0.939     0.939     0.939     10000
+    airplane      0.954     0.947     0.950      1000
+  automobile      0.950     0.984     0.967      1000
+        bird      0.929     0.933     0.931      1000
+         cat      0.891     0.877     0.884      1000
+        deer      0.960     0.951     0.955      1000
+         dog      0.913     0.910     0.911      1000
+        frog      0.964     0.973     0.969      1000
+       horse      0.966     0.963     0.964      1000
+        ship      0.966     0.972     0.969      1000
+       truck      0.969     0.953     0.961      1000
+
+    accuracy                          0.946     10000
+   macro avg      0.946     0.946     0.946     10000
+weighted avg      0.946     0.946     0.946     10000
 ```
 
 ### Architecture & training procedure
 
-The model and training procedure were taken *without modification* (except using our `train/` split and `num_epochs = 180`) from [github.com/xvel/cifar10-tiniest](https://github.com/xvel/cifar10-tiniest). Its author reportedly got inspirated by [github.com/soyflourbread/cifar10-tiny](https://github.com/soyflourbread/cifar10-tiny).
+The model and training procedure were taken *without modification* (except using our `train/` split and `num_epochs = 300`) from [github.com/xvel/cifar10-tiniest](https://github.com/xvel/cifar10-tiniest). Its author reportedly got inspirated by [github.com/soyflourbread/cifar10-tiny](https://github.com/soyflourbread/cifar10-tiny).
 
 Below is a *reformatted* layerwise overview of the model, generated with [torchinfo](https://github.com/tyleryep/torchinfo):
 

tiniest.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a5c6eed155922eac8df2511573408c23421cf8be8726070398f652e48d71e98
+size 419451