victormiller
commited on
Commit
•
41451ff
1
Parent(s):
d594e25
Update results.py
Browse files- results.py +35 -1
results.py
CHANGED
@@ -381,6 +381,40 @@ fig.update_layout(
|
|
381 |
# Show the figure
|
382 |
llama_graph1 = fig
|
383 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
384 |
|
385 |
intro_div = Div(
|
386 |
H2("Perplexity Evaluation on Duplicate Data"),
|
@@ -444,7 +478,7 @@ llama_div = Div(
|
|
444 |
Section(
|
445 |
H3("Perplexity vs Years"),
|
446 |
Img(src="images/prep-across-diff-years-global.png", height = "300", width = "600" ),
|
447 |
-
|
448 |
),
|
449 |
Section(
|
450 |
H3("Perplexity vs Dump Duplication"),
|
|
|
381 |
# Show the figure
|
382 |
llama_graph1 = fig
|
383 |
|
384 |
+
##llama graph 2
|
385 |
+
|
386 |
+
import plotly.graph_objects as go
|
387 |
+
|
388 |
+
# Data for different years and buckets
|
389 |
+
data = {
|
390 |
+
"1-1": [10.036774097041135, 9.46310273785878, 9.41413706166537, 9.50318602661455, 9.007669062339426, 8.388255660116407, 10.112246017864624, 10.239269162661959, 9.931951075969451, 8.646614152066428],
|
391 |
+
"2-5": [9.306693996275795, 8.763464863196129, 8.645126825996691, 9.473904977192573, 10.95829859145081, 10.676105294328789, 10.255251179892559, 9.54987953569235, 9.12737570591033, 8.806922449908505],
|
392 |
+
"6-10": [9.442327622499175, 9.075851726027564, 9.527148465147846, 9.755998086072951, 10.128151243953157, 9.728353939624842, 9.233548505479437, 9.067380903629866, 8.995868137602248, 8.816629232137835],
|
393 |
+
"11-100": [9.015408185880002, 8.868392446242012, 9.120345162203675, 8.968012141869462, 9.451949410987668, 9.381837094065533, 9.25131862646364, 9.014261939731549, 9.00805668763514, 8.995152677487027],
|
394 |
+
"101-1000": [9.94583162297666, 9.113560631617027, 8.9228845723255, 8.895860780054043, 8.863879736723902, 8.401723232809463, 8.458532176757009, 8.14345667720481, 7.882044010499616, 7.737747701620713],
|
395 |
+
"1001-30000000": [9.141712571508352, 8.037411460181893, 9.14052983061081, 8.757970647106037, 8.440366034517687, 7.5705604983353325, 7.4808205167223525, 7.312019290288715, 7.538858258386088, 6.77703951001925]
|
396 |
+
}
|
397 |
+
|
398 |
+
# Years
|
399 |
+
years = ["2014", "2015", "2016", "2017", "2018", "2019", "2020", "2021", "2022", "2023"]
|
400 |
+
|
401 |
+
# Create the figure
|
402 |
+
fig = go.Figure()
|
403 |
+
|
404 |
+
# Add lines for each bucket
|
405 |
+
for bucket, perplexities in data.items():
|
406 |
+
fig.add_trace(go.Scatter(x=years, y=perplexities, mode='lines+markers', name=bucket))
|
407 |
+
|
408 |
+
# Update layout
|
409 |
+
fig.update_layout(
|
410 |
+
title="Perplexity Across Different Years (Global)",
|
411 |
+
xaxis_title="Year",
|
412 |
+
yaxis_title="Average Perplexity",
|
413 |
+
legend_title="Bucket (duplicate count range)"
|
414 |
+
)
|
415 |
+
|
416 |
+
# Show the figure
|
417 |
+
llama_graph2 = fig
|
418 |
|
419 |
intro_div = Div(
|
420 |
H2("Perplexity Evaluation on Duplicate Data"),
|
|
|
478 |
Section(
|
479 |
H3("Perplexity vs Years"),
|
480 |
Img(src="images/prep-across-diff-years-global.png", height = "300", width = "600" ),
|
481 |
+
plotly2fasthtml(llama_graph2),
|
482 |
),
|
483 |
Section(
|
484 |
H3("Perplexity vs Dump Duplication"),
|