Update results.py

results.py

@@ -6,6 +6,37 @@ from plotly import graph_objects as go
 import pandas as pd
 import plotly.express as px
 
+
+##upsampling validation loss graph
+
+
+# Load the CSV file
+data = pd.read_csv('data/validation_loss _txt360_vs_fineweb.csv')
+
+# Create a plotly line chart
+fig = go.Figure()
+
+# Add the line for FineWeb-1.5T
+fig.add_trace(go.Scatter(x=data['Step'], y=data['FineWeb-1.5T'], mode='lines', name='FineWeb-1.5T'))
+
+# Add the line for TxT360
+fig.add_trace(go.Scatter(x=data['Step'], y=data['TxT360'], mode='lines', name='TxT360'))
+
+# Update layout
+fig.update_layout(
+    title='Validation Loss Comparison: TxT360 vs FineWeb-1.5T',
+    xaxis_title='Steps',
+    yaxis_title='Validation Loss',
+    legend_title='Dataset',
+    template='plotly_white'
+)
+
+# Show the plot
+validation_loss_graph = fig
+
+
+
+
 #Perplexity Across Different Buckets (global)
 # The data you provided
 DATA = [
@@ -687,6 +718,7 @@ upsampling_exp = Div(
         P("To account for differing dataset sizes, the evaluation scores represent the final evaluation score after the entire dataset has been processed."),
         H3("Training Evaluations"),
         P("We also conducted full scale training using TxT360 and FineWeb-1.5T. Below are plots of the training and validation loss curves for each dataset. We can see that TxT360 achieves a lower training and validation loss compared to FineWeb-1.5T. "),
+        validation_loss_graph,
 )
 
 perp1_div = Div(