lars/aiRtrafficNN
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Finish run 2

Browse Source
This commit is contained in:
Lars Bogner
2025-11-27 08:34:08 +01:00
parent 3c5772afa5
commit ed13aeb822
2 changed files with 69 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Code/python/notebooks/hyperparameter_scan.py
+4 -4
View File
@@ -3,8 +3,8 @@ import pathlib
import pandas as pd
from aiRNN import dataloader, models, losses
MODE = 1
DEVICE = "cuda"
MODE = 2
DEVICE = "cuda:2"
# ------------------------------------------------------------
@@ -181,7 +181,7 @@ def run_mode_1():
def run_mode_2():
results = []
step = 1
step = 10
base_ds = dataloader.SaveDataset(torch.load("dataset.pt"), step=step)
train_ds, val_ds = split_dataset(base_ds)
@@ -228,7 +228,7 @@ def run_mode_2():
def run_mode_3():
results = []
step = 1
step = 10
base_name = "LSTM"
cls = models.ThreeInputLSTM
hidden_size = 16
Code/python/notebooks/hyperparameter_scan_results_mode2.csv
+65
View File
@@ -0,0 +1,65 @@
Model,Hidden Layers,RNN Layers,Dropout,Final Loss
LSTM,0,2,0.0,13.57315312304967
LSTM,0,2,0.05,19.794176068104488
LSTM,0,2,0.1,17.698180574766347
LSTM,0,2,0.15,23.32098652611316
LSTM,0,3,0.0,13.700146980688606
LSTM,0,3,0.05,17.857194296071228
LSTM,0,3,0.1,20.897490259627222
LSTM,0,3,0.15,20.09943933218298
LSTM,0,4,0.0,20.451897950239584
LSTM,0,4,0.05,30.697136825239156
LSTM,0,4,0.1,35.30218719428694
LSTM,0,4,0.15,56.88415065281828
LSTM,0,5,0.0,32.88772487640381
LSTM,0,5,0.05,41.01707866829886
LSTM,0,5,0.1,25.020204705251775
LSTM,0,5,0.15,31.874271876375442
LSTM,1,2,0.0,15.143101020598076
LSTM,1,2,0.05,17.294107027456793
LSTM,1,2,0.1,20.339507552939402
LSTM,1,2,0.15,20.261145202206894
LSTM,1,3,0.0,18.612104852434616
LSTM,1,3,0.05,21.64210480031833
LSTM,1,3,0.1,33.828577202810365
LSTM,1,3,0.15,24.81157889836271
LSTM,1,4,0.0,25.49259500100579
LSTM,1,4,0.05,43.26106494581196
LSTM,1,4,0.1,109.03877604847223
LSTM,1,4,0.15,26.519190546492457
LSTM,1,5,0.0,24.345332991908972
LSTM,1,5,0.05,2160.6643607985807
LSTM,1,5,0.1,2160.6641690966107
LSTM,1,5,0.15,70.0674764337674
LSTM,2,2,0.0,96.50166923899046
LSTM,2,2,0.05,22.740719331821925
LSTM,2,2,0.1,27.11947537811709
LSTM,2,2,0.15,26.377998459507044
LSTM,2,3,0.0,29.63593423870248
LSTM,2,3,0.05,40.430933052385356
LSTM,2,3,0.1,102.70769267015054
LSTM,2,3,0.15,39.901821069314444
LSTM,2,4,0.0,16.42642393246503
LSTM,2,4,0.05,103.43423487434924
LSTM,2,4,0.1,132.6539063252194
LSTM,2,4,0.15,109.47043222776601
LSTM,2,5,0.0,26.904185483153437
LSTM,2,5,0.05,2160.663507166043
LSTM,2,5,0.1,2160.6676799075703
LSTM,2,5,0.15,2160.6770200057767
LSTM,4,2,0.0,96.90064337555792
LSTM,4,2,0.05,44.17875827198297
LSTM,4,2,0.1,107.32435410459277
LSTM,4,2,0.15,36.28804776366328
LSTM,4,3,0.0,100.72823489551813
LSTM,4,3,0.05,114.38032405477175
LSTM,4,3,0.1,220.88371201636087
LSTM,4,3,0.15,38.124560718805014
LSTM,4,4,0.0,50.08498817766216
LSTM,4,4,0.05,2160.668174206371
LSTM,4,4,0.1,317.8842515542474
LSTM,4,4,0.15,380.57344377544564
LSTM,4,5,0.0,2160.663039516395
LSTM,4,5,0.05,2160.6706989986797
LSTM,4,5,0.1,158.177048481686
LSTM,4,5,0.15,704.9843010700924
1 Model Hidden Layers RNN Layers Dropout Final Loss
2 LSTM 0 2 0.0 13.57315312304967
3 LSTM 0 2 0.05 19.794176068104488
4 LSTM 0 2 0.1 17.698180574766347
5 LSTM 0 2 0.15 23.32098652611316
6 LSTM 0 3 0.0 13.700146980688606
7 LSTM 0 3 0.05 17.857194296071228
8 LSTM 0 3 0.1 20.897490259627222
9 LSTM 0 3 0.15 20.09943933218298
10 LSTM 0 4 0.0 20.451897950239584
11 LSTM 0 4 0.05 30.697136825239156
12 LSTM 0 4 0.1 35.30218719428694
13 LSTM 0 4 0.15 56.88415065281828
14 LSTM 0 5 0.0 32.88772487640381
15 LSTM 0 5 0.05 41.01707866829886
16 LSTM 0 5 0.1 25.020204705251775
17 LSTM 0 5 0.15 31.874271876375442
18 LSTM 1 2 0.0 15.143101020598076
19 LSTM 1 2 0.05 17.294107027456793
20 LSTM 1 2 0.1 20.339507552939402
21 LSTM 1 2 0.15 20.261145202206894
22 LSTM 1 3 0.0 18.612104852434616
23 LSTM 1 3 0.05 21.64210480031833
24 LSTM 1 3 0.1 33.828577202810365
25 LSTM 1 3 0.15 24.81157889836271
26 LSTM 1 4 0.0 25.49259500100579
27 LSTM 1 4 0.05 43.26106494581196
28 LSTM 1 4 0.1 109.03877604847223
29 LSTM 1 4 0.15 26.519190546492457
30 LSTM 1 5 0.0 24.345332991908972
31 LSTM 1 5 0.05 2160.6643607985807
32 LSTM 1 5 0.1 2160.6641690966107
33 LSTM 1 5 0.15 70.0674764337674
34 LSTM 2 2 0.0 96.50166923899046
35 LSTM 2 2 0.05 22.740719331821925
36 LSTM 2 2 0.1 27.11947537811709
37 LSTM 2 2 0.15 26.377998459507044
38 LSTM 2 3 0.0 29.63593423870248
39 LSTM 2 3 0.05 40.430933052385356
40 LSTM 2 3 0.1 102.70769267015054
41 LSTM 2 3 0.15 39.901821069314444
42 LSTM 2 4 0.0 16.42642393246503
43 LSTM 2 4 0.05 103.43423487434924
44 LSTM 2 4 0.1 132.6539063252194
45 LSTM 2 4 0.15 109.47043222776601
46 LSTM 2 5 0.0 26.904185483153437
47 LSTM 2 5 0.05 2160.663507166043
48 LSTM 2 5 0.1 2160.6676799075703
49 LSTM 2 5 0.15 2160.6770200057767
50 LSTM 4 2 0.0 96.90064337555792
51 LSTM 4 2 0.05 44.17875827198297
52 LSTM 4 2 0.1 107.32435410459277
53 LSTM 4 2 0.15 36.28804776366328
54 LSTM 4 3 0.0 100.72823489551813
55 LSTM 4 3 0.05 114.38032405477175
56 LSTM 4 3 0.1 220.88371201636087
57 LSTM 4 3 0.15 38.124560718805014
58 LSTM 4 4 0.0 50.08498817766216
59 LSTM 4 4 0.05 2160.668174206371
60 LSTM 4 4 0.1 317.8842515542474
61 LSTM 4 4 0.15 380.57344377544564
62 LSTM 4 5 0.0 2160.663039516395
63 LSTM 4 5 0.05 2160.6706989986797
64 LSTM 4 5 0.1 158.177048481686
65 LSTM 4 5 0.15 704.9843010700924