Update results.py

results.py

@@ -416,9 +416,57 @@ fig.update_layout(
 # Show the figure
 llama_graph2 = fig
 
-
 #llama graph 3
 
+# Data for different buckets and perplexity values across years
+data = {
+    "1-1": {
+        "years": ["2014", "2015", "2016", "2017", "2018", "2019", "2020", "2021", "2022", "2023"],
+        "perplexities": [10.036774097041135, 9.46310273785878, 9.41413706166537, 9.50318602661455, 9.007669062339426, 8.388255660116407, 10.112246017864624, 10.239269162661959, 9.931951075969451, 8.646614152066428]
+    },
+    "2-5": {
+        "years": ["2014", "2015", "2016", "2017", "2018", "2019", "2020", "2021", "2022", "2023"],
+        "perplexities": [9.306693996275795, 8.763464863196129, 8.645126825996691, 9.473904977192573, 10.95829859145081, 10.676105294328789, 10.255251179892559, 9.54987953569235, 9.12737570591033, 8.806922449908505]
+    },
+    "6-10": {
+        "years": ["2014", "2015", "2016", "2017", "2018", "2019", "2020", "2021", "2022", "2023"],
+        "perplexities": [9.442327622499175, 9.075851726027564, 9.527148465147846, 9.755998086072951, 10.128151243953157, 9.728353939624842, 9.233548505479437, 9.067380903629866, 8.995868137602248, 8.816629232137835]
+    },
+    "11-100": {
+        "years": ["2014", "2015", "2016", "2017", "2018", "2019", "2020", "2021", "2022", "2023"],
+        "perplexities": [9.015408185880002, 8.868392446242012, 9.120345162203675, 8.968012141869462, 9.451949410987668, 9.381837094065533, 9.25131862646364, 9.014261939731549, 9.00805668763514, 8.995152677487027]
+    },
+    "101-1000": {
+        "years": ["2014", "2015", "2016", "2017", "2018", "2019", "2020", "2021", "2022", "2023"],
+        "perplexities": [9.94583162297666, 9.113560631617027, 8.9228845723255, 8.895860780054043, 8.863879736723902, 8.401723232809463, 8.458532176757009, 8.14345667720481, 7.882044010499616, 7.737747701620713]
+    },
+    "1001-30000000": {
+        "years": ["2014", "2015", "2016", "2017", "2018", "2019", "2020", "2021", "2022", "2023"],
+        "perplexities": [9.141712571508352, 8.037411460181893, 9.14052983061081, 8.757970647106037, 8.440366034517687, 7.5705604983353325, 7.4808205167223525, 7.312019290288715, 7.538858258386088, 6.77703951001925]
+    }
+}
+
+# Create figure
+fig = go.Figure()
+
+# Add traces for each bucket
+for bucket, bucket_data in data.items():
+    fig.add_trace(go.Scatter(x=bucket_data["years"], y=bucket_data["perplexities"], mode='lines+markers', name=bucket))
+
+# Update layout
+fig.update_layout(
+    title="Perplexity Across Different Buckets (Global)",
+    xaxis_title="Year",
+    yaxis_title="Average Perplexity",
+    legend_title="Bucket (Duplicate Count Range)"
+)
+
+# Show the figure
+llama_graph3 = fig
+
+
+#llama graph 4
+
 import plotly.graph_objects as go
 
 # Data for different years and dump duplications
@@ -481,10 +529,10 @@ fig.update_layout(
 )
 
 # Show the figure
-llama_graph3 = fig
+llama_graph4 = fig
 
 
-#llama graph 4
+#llama graph 5
 
 # Data for different buckets and years
 import plotly.graph_objects as go
@@ -549,9 +597,9 @@ fig.update_layout(
 )
 
 # Show the figure
-llama_graph4 = fig
+llama_graph5 = fig
 
-##llama graph 5
+##llama graph 6
 
 # Data for different years and duplication counts
 data = {
@@ -613,7 +661,7 @@ fig.update_layout(
 )
 
 # Show the figure
-llama_graph5 = fig
+llama_graph6 = fig
 
 
 intro_div = Div(
@@ -684,16 +732,20 @@ llama_div = Div(
             H3("Perplexity vs Dump Duplication"),
             Img(src="images/prep-vs-dump-dup-global.png", height = "300", width = "600" ),
             plotly2fasthtml(llama_graph3),
+    Section(
+            H3("Perplexity vs Dump Duplication"),
+            Img(src="images/prep-vs-dump-dup-global.png", height = "300", width = "600" ),
+            plotly2fasthtml(llama_graph4),
         ),
     Section(
             H3("Perplexity vs Local Buckets"),
             Img(src="images/prep-diff-buckets-local.png", height = "300", width = "600" ),
-            plotly2fasthtml(llama_graph4),
+            plotly2fasthtml(llama_graph5),
         ),
     Section(
             H3("Perplexity vs Local Dump Duplication"),
             Img(src="images/prep-vs-dump-dup-global.png", height = "300", width = "600" ),
-            plotly2fasthtml(llama_graph5),
+            plotly2fasthtml(llama_graph6),
         ),
 )