Update README.md

README.md

@@ -43,10 +43,31 @@ tags:
 
 # LLM Leaderboard Data for Hendrycks MATH Dataset
 
-A papers with code attempt to aggregate yearly (2022-2024) LLM / foundation model performance on the Hendryck's MATH evaluation.
+A Papers with Code attempt to aggregate yearly (2022-2024) LLM/Foundation model performance on Hendrycks' MATH evaluation.
 
 Data converted from source: [Math Word Problem Solving on MATH](https://paperswithcode.com/sota/math-word-problem-solving-on-math)
 
 ## Evaluation
 
-Introduced by Hendrycks et al. in Measuring Mathematical Problem Solving With the MATH Dataset. MATH is a new dataset of 12,500 challenging competition mathematics problems. Each problem in MATH has a full step-by-step solution which can be used to teach models to generate answer derivations and explanations.
+Introduced by Hendrycks et al. in *Measuring Mathematical Problem Solving With the MATH Dataset*. MATH is a dataset comprising 12,500 challenging competition mathematics problems. Each problem in MATH has a full step-by-step solution useful for teaching models to generate answer derivations and explanations.
+
+## Visualizations
+
+### Model Accuracy Trends
+
+![Model Accuracy Trends](img/model_accuracy_trends.png)
+
+*Figure 1: Trends in model accuracy from 2022 to 2024, illustrating improvement rates over time.*
+
+### Top 20% Model Accuracy
+
+![Top 20% Model Accuracy](img/top_20_percent_accuracy.png)
+
+*Figure 2: Accuracy distribution among the top 20% performing models on the Hendrycks MATH dataset.*
+
+### Standard Deviation vs Median (Top 20%)
+
+![Top 20% Standard Deviation vs Median](img/top_20_std_vs_median.png)
+
+*Figure 3: Relationship between standard deviation and median accuracy scores for the top 20% models.*
+