Update README.md (#6)

- Update README.md (2c55bfec4767eea478e4682cdaffe1837191f965)


Co-authored-by: Hajar Emami <[email protected]>

README.md

@@ -39,7 +39,7 @@ Hugging Face introduced FineWeb V1.1.0, a large-scale dataset for LLM pre-traini
 
 These were applied in the order shown in the Fig 1
 
-<img src="GneissWeb3.png" alt="GneissWeb3" style="width:1000px;"/>
+<img src="GneissWeb_recipe_new.png" alt="GneissWeb_recipe_new" style="width:1000px;"/>
 
 **Figure 1 :**  GneissWeb recipe 
 
@@ -51,7 +51,7 @@ To compare GneissWeb against the baselines, we trained decoder models of 7B para
 
 We used FineWeb 1.1.0 and FineWeb-Edu-score2 as our comparison baselines.  The former is of  15T tokens and the latter of 5.4T tokens. Fig 2 shows how the subsamples were created for the Fineweb baselines as well for GneissWeb.  
 
-<img src="ablation_strategy_new.png" alt="ablation_strategy.png" style="width:1000px;"/>
+<img src="ablation_strategy_new.png" alt="ablation_strategy_new.png" style="width:1000px;"/>
 
 **Figure 2:**  Subsampling and Ablation Strategy