Joseph Hopfmüller
|
884d9f73c9
|
add default to birefsteps
|
2024-12-05 23:50:49 +01:00 |
|
Joseph Hopfmüller
|
08c81428ef
|
add new training data
|
2024-12-05 23:50:14 +01:00 |
|
Joseph Hopfmüller
|
9299316f81
|
add new training_data
|
2024-12-05 23:43:24 +01:00 |
|
Joseph Hopfmüller
|
2fcf621a50
|
update submodule to latest commit
|
2024-12-05 23:29:53 +01:00 |
|
Joseph Hopfmüller
|
36f89f379d
|
refactor signal generation to support single-channel mode and adjust birefringence parameters
|
2024-12-05 23:29:48 +01:00 |
|
Joseph Hopfmüller
|
950516197c
|
add new training data
|
2024-12-04 13:22:50 +01:00 |
|
Joseph Hopfmüller
|
b7eecea1ae
|
add new training data
|
2024-12-04 10:07:39 +01:00 |
|
Joseph Hopfmüller
|
d3626f9a44
|
remove newly generated training data
|
2024-12-04 10:00:50 +01:00 |
|
Joseph Hopfmüller
|
7f89c02d8d
|
update submodule to latest commit
|
2024-12-02 18:54:15 +01:00 |
|
Joseph Hopfmüller
|
a8a1c49c00
|
Merge branch 'machine_learning' of git.suuppl.dev:seppl/optical-regeneration into machine_learning
|
2024-12-02 18:51:30 +01:00 |
|
Joseph Hopfmüller
|
297e9e8d7f
|
update submodule configuration and enhance model settings; add eye diagram functionality
|
2024-12-02 18:50:43 +01:00 |
|
Joseph Hopfmüller
|
aa2e7a4cb4
|
move sliced dataset testing script
|
2024-12-02 18:50:13 +01:00 |
|
Joseph Hopfmüller
|
1dcefecf59
|
update signal generation patterns in generate_signal.py
|
2024-12-02 18:49:38 +01:00 |
|
Joseph Hopfmüller
|
a5f2f49360
|
enhance data loading and processing in FiberRegenerationDataset; add timestamps and support for multiple file paths
|
2024-12-02 18:49:14 +01:00 |
|
Joseph Hopfmüller
|
e20aa9bfb1
|
add eye_diagram analysis
|
2024-12-02 18:48:43 +01:00 |
|
Joseph Hopfmüller
|
010889af13
|
generate training data with finer length steps
|
2024-12-02 18:45:10 +01:00 |
|
Joseph Hopfmüller
|
c145d58df8
|
update subproject reference to latest commit
|
2024-11-29 15:53:08 +01:00 |
|
Joseph Hopfmüller
|
487288c923
|
define new activation functions and parametrizations
|
2024-11-29 15:51:25 +01:00 |
|
Joseph Hopfmüller
|
bdf6f5bfb8
|
clean up regen_no_hyper.py
|
2024-11-29 15:50:34 +01:00 |
|
Joseph Hopfmüller
|
e02662ed4f
|
new optuna studies
|
2024-11-29 15:49:59 +01:00 |
|
Joseph Hopfmüller
|
fd7a0b9c31
|
using latest knowledge for hyperparameter search
|
2024-11-29 15:49:46 +01:00 |
|
Joseph Hopfmüller
|
ff32aefd52
|
minor fixes and changes
|
2024-11-29 15:49:10 +01:00 |
|
Joseph Hopfmüller
|
b156b9ceaf
|
refactor hypertraining.py to improve model layer handling and response plotting; adjust data settings for batch processing
|
2024-11-29 15:48:27 +01:00 |
|
Joseph Hopfmüller
|
cfa08aae4e
|
add training.py for defining and running models without hyperparametertuning
|
2024-11-29 15:48:18 +01:00 |
|
Joseph Hopfmüller
|
0422c81f3b
|
update single_core_regen settings new runs
|
2024-11-24 01:56:01 +01:00 |
|
Joseph Hopfmüller
|
7343ccb3a5
|
refactor complex loss functions for improved readability; update settings and dataset classes for consistency
|
2024-11-24 01:55:32 +01:00 |
|
Joseph Hopfmüller
|
9a16a5637d
|
add optional parameter suggestion methods for Optuna trials
|
2024-11-24 01:55:12 +01:00 |
|
Joseph Hopfmüller
|
80e9a3379e
|
add autosampler support
|
2024-11-20 23:10:14 +01:00 |
|
Joseph Hopfmüller
|
8d4d0468bd
|
complexhalf (complex32) isn't supported by torch.linalg.qr
|
2024-11-20 22:56:26 +01:00 |
|
Joseph Hopfmüller
|
6358c95c42
|
new hyperparameter db
|
2024-11-20 22:49:40 +01:00 |
|
Joseph Hopfmüller
|
674033ac2e
|
move hypertraining class into separate file;
move settings dataclasses into separate file;
add SemiUnitaryLayer;
clean up model response plotting code;
cnt hyperparameter search
|
2024-11-20 22:49:31 +01:00 |
|
Joseph Hopfmüller
|
cdca5de473
|
training loop speedup
|
2024-11-20 11:29:18 +01:00 |
|
Joseph Hopfmüller
|
1622c38582
|
refactor: remove unused Optuna visualization utility
|
2024-11-17 22:23:37 +01:00 |
|
Joseph Hopfmüller
|
2bba760378
|
add: implement Optuna visualization utility with Dash
|
2024-11-17 22:23:01 +01:00 |
|
Joseph Hopfmüller
|
9ec548757d
|
add: regen.py (main hyperparameter training file)
feat: add utility functions for fiber dataset visualization and hyperparameter training;
housekeeping: rename dataset.py -> datasets.py
|
2024-11-17 22:22:37 +01:00 |
|
Joseph Hopfmüller
|
05a3ee9394
|
refactor: clean up .gitignore, remove unused scripts
|
2024-11-17 22:18:44 +01:00 |
|
Joseph Hopfmüller
|
086240489a
|
minor edits on notes
|
2024-11-17 22:16:52 +01:00 |
|
Joseph Hopfmüller
|
87f40fc37c
|
add SlicedDataset class and utility scripts; refactor: remove _path_fix.py and update imports;
|
2024-11-17 01:04:33 +01:00 |
|
Joseph Hopfmüller
|
90aa6dbaf8
|
housekeeping
|
2024-11-17 01:04:14 +01:00 |
|
Joseph Hopfmüller
|
744c5f5166
|
rename dir;
add torch import test script
|
2024-11-16 00:39:19 +01:00 |
|
Joseph Hopfmüller
|
939a511625
|
add installation notes for CUDA 12.4, pyenv, and PyTorch; include useful links
|
2024-11-16 00:03:43 +01:00 |
|
Joseph Hopfmüller
|
2bf47dc0c0
|
more precise clone instructions
|
2024-11-15 23:54:33 +01:00 |
|
Joseph Hopfmüller
|
5871c79501
|
update README with cloning notes and adjust path in _path_fix.py for submodule import
|
2024-11-15 21:21:50 +01:00 |
|
Joseph Hopfmüller
|
f7217a93bb
|
add pypho as submodule
|
2024-11-15 21:00:19 +01:00 |
|
Joseph Hopfmüller
|
5e2d3dd6b7
|
change pypho symlink to submodule and handle new location in _path_fix.py
|
2024-11-15 20:56:34 +01:00 |
|
Joseph Hopfmüller
|
c42da6ca58
|
add license EUPL-1.2
|
2024-11-15 18:28:45 +01:00 |
|
Joseph Hopfmüller
|
e929991b3d
|
add datasets to LFS
|
2024-11-15 18:27:19 +01:00 |
|
Joseph Hopfmüller
|
fa85fca978
|
add data to lfs
|
2024-11-15 17:01:29 +01:00 |
|
Joseph Hopfmüller
|
9c43ce4c85
|
git lfs setup cont'd 2
|
2024-11-15 16:54:03 +01:00 |
|
Joseph Hopfmüller
|
c76bfd20fe
|
lfs setup cont'd
|
2024-11-15 16:49:09 +01:00 |
|