Fixed plot

data/rag_eval/results/results.json

@@ -4,11 +4,6 @@
         "context_recall": 0,
         "context_relevancy": 0
     },
-    {
-        "context_precision": 0.9819999999999999,
-        "context_recall": 0.9400000000000002,
-        "context_relevancy": 0
-    },
     {
         "context_precision": 0.9819999999999999,
         "context_recall": 0.9400000000000002,

test/benchmarks/rag/evaluation.py

@@ -185,7 +185,10 @@ def evaluate(vdb: Store, qa_paths: list, endpoint: str, metrics: list,
     return metrics, eval_dataset
 
 
-def update_evaluation_plots(results_df: pd.DataFrame, metrics: list, modified=True):
+def update_evaluation_plots(results_df: pd.DataFrame, metrics: list,
+                            modified=True,
+                            rows: int = 1,
+                            cols: int = 3):
     if len(metrics) == 0:
         raise ValueError('No metrics specified.')
 
@@ -206,8 +209,9 @@ def update_evaluation_plots(results_df: pd.DataFrame, metrics: list, modified=Tr
 
             # Add new results
             res: pd.Series = results_df.mean()
-            new_results = {metric_name: res[metric_name] if metric_name in res else content[len(content) - 1][metric_name]
-                           for metric_name in content[0].keys()}
+            new_results = {
+                metric_name: res[metric_name] if metric_name in res else content[len(content) - 1][metric_name]
+                for metric_name in content[0].keys()}
             content.append(new_results)
 
             fp.seek(0)
@@ -216,33 +220,44 @@ def update_evaluation_plots(results_df: pd.DataFrame, metrics: list, modified=Tr
     else:
         history = results_df
 
-    def plot_eval(plot_df: pd.DataFrame, name: str):
+    # Ensure the grid has enough space for all metrics
+    total_metrics = len(metrics)
+    if rows * cols < total_metrics:
+        raise ValueError(f'Grid size ({rows}x{cols}) is too small for {total_metrics} metrics.')
+
+        # Create a single plot with subplots for each metric
+    fig, axes = plt.subplots(rows, cols, figsize=(cols * 5, rows * 5))
+    axes = axes.flatten()  # Flatten in case of a single row or column
+
+    def plot_eval(ax, plot_df: pd.DataFrame, name: str):
         """Create a plot for an evaluation metric, the columns should be named 'x' and 'y'"""
-        sns.lineplot(data=plot_df, x='x', y='y', zorder=0)
-        plt.scatter(
+        sns.lineplot(data=plot_df, x='x', y='y', ax=ax, zorder=0)
+        ax.scatter(
             plot_df.iloc[1:]['x'],
             plot_df.iloc[1:]['y'],
             color='#000000',
             s=15,
             zorder=1
         )
-
-        plt.ylim(0, 1)
-        plt.xticks(range(0, len(plot_df)))
-
-        plt.title(f'RAG Evaluation: {name}')
-        plt.ylabel(name)
-        plt.xlabel('')
-        return plt
+        ax.set_ylim(0, 1)
+        ax.set_xticks(range(0, len(plot_df)))
+        ax.set_title(f'RAG Evaluation: {name}')
+        ax.set_ylabel(name)
+        ax.set_xlabel('')
 
     # Output the updated evaluation plots
-    plots = {}
-    for col in history.columns:
+    for i, col in enumerate(history.columns):
         values = history[col].to_list()
-        plots[col] = [{'x': i, 'y': val} for i, val in enumerate(values)]
-        metric_plot_df = pd.DataFrame(plots[col])
-        plot = plot_eval(metric_plot_df, col)
-        plot.savefig(f'../../../data/rag_eval/results/plots/{col}.png')
+        metric_plot_df = pd.DataFrame([{'x': i, 'y': val} for i, val in enumerate(values)])
+        plot_eval(axes[i], metric_plot_df, col)
+
+    # Hide any unused subplots
+    for j in range(i + 1, len(axes)):
+        fig.delaxes(axes[j])
+
+    plt.tight_layout()
+    plt.savefig(f'../../../data/rag_eval/results/plots/plot.png')
+    plt.close()
 
 
 def main(plot_only=False):
@@ -279,4 +294,4 @@ def main(plot_only=False):
 
 
 if __name__ == '__main__':
-    main(plot_only=False)
+    main(plot_only=True)