johko commited on
Commit
cec4bb3
1 Parent(s): 186c6f2

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +4 -0
README.md CHANGED
@@ -6,6 +6,8 @@ datasets:
6
  - STL-10
7
  license: apache-2.0
8
  ---
 
 
9
  ## Description
10
 
11
  This is a simple image classification model trained with **Semi-supervised image classification using contrastive pretraining with SimCLR**
@@ -13,6 +15,8 @@ The training procedure was done as seen in the example on <a href='https://keras
13
 
14
  The model was **trained on STL-10**, which includes ten classes: airplane, bird, car, cat, deer, dog, horse, monkey, ship, truck.
15
 
 
 
16
 
17
  ## Background
18
  (by András Béres on <a href='https://keras.io/examples/vision/semisupervised_simclr/' target='_blank'>**keras.io**</a> )
 
6
  - STL-10
7
  license: apache-2.0
8
  ---
9
+ # Semi-supervised image classification using contrastive pretraining with SimCLR
10
+
11
  ## Description
12
 
13
  This is a simple image classification model trained with **Semi-supervised image classification using contrastive pretraining with SimCLR**
 
15
 
16
  The model was **trained on STL-10**, which includes ten classes: airplane, bird, car, cat, deer, dog, horse, monkey, ship, truck.
17
 
18
+ ##Metrics
19
+ There is a public W&B dashboard available <a href='https://wandb.ai/johko-cel/semi-supervised-contrastive-learning-simclr'>here</a> which illustrates the difference in different metrics such as accuracy of a baseline supervised trained model, a purely unsupervised model (pretrain) and the supervised finetuned model based on the unsupervised.
20
 
21
  ## Background
22
  (by András Béres on <a href='https://keras.io/examples/vision/semisupervised_simclr/' target='_blank'>**keras.io**</a> )