Update README.md

README.md

@@ -28,9 +28,7 @@ We started with the goal of distilling 10T+ high quality tokens from FineWeb V1.
 
 These were applied in the order shown in the Fig 1
 
-![GneissWeb3.png](GneissWeb3.png)
-
-[comment]: <img src="https://huggingface.co/datasets/ibm-granite/GneissWeb/blob/main/GneissWeb3.png"  alt="BRB">                 
+![GneissWeb3.png](GneissWeb3.png)               
 
 Figure 1 :  GneissWeb recipe 
 
@@ -46,6 +44,7 @@ The baselines from which equivalent data was subsampled and used for this compar
  Fig 2 shows how the subsamples were created for the Fineweb baselines as well for GneissWeb.  A similar strategy as for the creation of the Fineweb baseline was used for other baselines too
 
 
+![ablation_strategy.png](ablation_strategy.png)
 
                                                     Figure 2:  Subsampling and Ablation Strategy