issues
search
mlverse
/
luz
Higher Level API for torch
https://mlverse.github.io/luz/
Other
83
stars
12
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Release luz 0.3.0
#99
dfalbel
closed
2 years ago
0
progress bar for `lr_finder`.
#98
dfalbel
closed
2 years ago
0
Lr finder improvements
#97
dfalbel
closed
2 years ago
1
Fix typos in fit help.
#96
jonthegeek
closed
2 years ago
1
Add callback to use the best model
#95
dfalbel
closed
2 years ago
0
Fix README typos.
#94
jonthegeek
closed
2 years ago
3
Add backward argument to `setup`.
#93
dfalbel
closed
2 years ago
0
Weights
#92
dfalbel
opened
2 years ago
1
FIgure out how to modify how `backward()` is called.
#91
dfalbel
closed
2 years ago
1
Add callback for gradient clipping inspired by FastAI's.
#90
dfalbel
closed
2 years ago
0
Learning rates tested by lr_finder() run into saturation
#89
EmKarr
closed
2 years ago
1
Stop/resume model fitting
#88
pradosj
closed
2 years ago
0
lr_finder changes minibatch size
#87
jtrecenti
closed
2 years ago
1
Examples dont run on their own
#86
GitHubGeniusOverlord
closed
2 years ago
1
Update learning rate exponentially and maybe, add smoothing to loss plot
#85
skeydan
closed
2 years ago
2
move weight reshaping inside nnf_mixup
#84
skeydan
closed
2 years ago
0
Consider only computing metrics for the validation set.
#83
dfalbel
closed
2 years ago
0
add mixup callback and mixup cross entropy loss
#82
skeydan
closed
2 years ago
9
Clean up wording in get-started vignette.
#81
jonthegeek
closed
3 years ago
1
add label smoothing
#80
skeydan
closed
3 years ago
1
return the mod instead due to testthat changes.
#79
dfalbel
closed
3 years ago
0
Error with last batch that is smaller than the others
#78
skeydan
closed
3 years ago
4
Fix add = TRUE order in `on.exit`
#77
dfalbel
closed
3 years ago
0
Bugfix/early stopping
#76
dfalbel
closed
3 years ago
2
Move `master` branch to `main`
#75
jennybc
closed
3 years ago
1
Leak the `fit` function environment
#74
dfalbel
closed
2 years ago
8
Implements `evaluate`
#73
dfalbel
closed
3 years ago
0
Implement `evaluate` so we can quickly compute metrics in a specific dataset.
#72
dfalbel
closed
3 years ago
1
Feature/dataloader options
#71
dfalbel
closed
3 years ago
0
Training data should be shuffled by default
#70
dfalbel
closed
3 years ago
0
support for proportion of valid_data + documentation improvements.
#69
dfalbel
closed
3 years ago
0
Support for `valid_data = p`
#68
dfalbel
closed
3 years ago
0
Feature/numerics
#67
dfalbel
closed
3 years ago
0
Adds `as_dataloader` to cast data into dataloader
#66
dfalbel
closed
3 years ago
0
Unet example
#65
dfalbel
closed
3 years ago
0
Combining early stopping with csv logger callback, metrics for last epoch don't get logged
#64
skeydan
closed
3 years ago
1
Early stopping with patience=1 shouldn't stop in the first epoch
#63
dfalbel
closed
3 years ago
0
Release luz 0.2.0
#62
dfalbel
closed
3 years ago
0
set old seed behavior to make tests pass with newer torch version
#61
dfalbel
closed
3 years ago
0
round batch size usign floor() and change order of start and end lr
#60
cmcmaster1
closed
3 years ago
1
add lr_finder
#59
cmcmaster1
closed
3 years ago
2
support for choosing CUDA device index if multiple GPUs present
#58
cmcmaster1
closed
3 years ago
0
Initial work on tracing callback
#57
dfalbel
opened
3 years ago
0
Implement luz_callback_validation_check
#56
mattwarkentin
opened
3 years ago
0
Provide support for minimum and maximum number of epochs
#55
mattwarkentin
closed
3 years ago
1
Fixes progress bars for predictions.
#54
dfalbel
closed
3 years ago
0
Minimum and maximum number of epochs
#53
mattwarkentin
closed
3 years ago
2
Fixes csv logger and validation progress bar length
#52
mattwarkentin
closed
3 years ago
1
Issue with length of progress bar for validation loop
#51
mattwarkentin
closed
3 years ago
0
`luz_callback_csv_logger()` saves a space-delimited file, not a comma-separated one. I think `sep = ","` needs to be added here:
#50
dfalbel
closed
3 years ago
0
Previous
Next