Skip to main content
stacey
Projects
deep-drive
Reports
Learning rate has strong negative correlation with car accuracy
Log in
Sign up
Share
Comment
Star
Share
Comment
Star
Learning rate has strong negative correlation with car accuracy
Lukas Biewald
Created on October 5
|
Last edited on October 5
Comment
Section 1
Parameter importance with respect to
car_acc
Parameters
1-10
of 18
Config parameter
Importance
Correlation
learning_rate
learning_rate
num_train
num_train
num_valid
num_valid
weight_decay
weight_decay
encoder.value_resnet34
encoder.value_resnet34
batch_size
batch_size
encoder.value_resnet18
encoder.value_resnet18
bn_weight_decay
bn_weight_decay
epochs
epochs
training_stages
training_stages
Loading...
iou
iou
Showing first 10 runs
0
5
10
15
Step
0.5
0.6
0.7
0.8
human_acc, iou, car_acc
human_acc, iou, car_acc
50
100
150
Time (minutes)
0
0.2
0.4
0.6
0.8
road_acc, car_acc
road_acc, car_acc
0
10
20
30
Step
0.65
0.7
0.75
0.8
0.85
0.9
0.95
Run set
390
Not in every case but overall.
LaTeX
\LaTeX
L
A
T
E
X
Add a comment