Package: luz 0.4.0.9002

Daniel Falbel

luz: Higher Level 'API' for 'torch'

A high level interface for 'torch' providing utilities to reduce the the amount of code needed for common tasks, abstract away torch details and make the same code work on both the 'CPU' and 'GPU'. It's flexible enough to support expressing a large range of models. It's heavily inspired by 'fastai' by Howard et al. (2020) <arxiv:2002.04688>, 'Keras' by Chollet et al. (2015) and 'PyTorch Lightning' by Falcon et al. (2019) <doi:10.5281/zenodo.3828935>.

Authors:Daniel Falbel [aut, cre, cph], Christophe Regouby [ctb], RStudio [cph]

luz_0.4.0.9002.tar.gz
luz_0.4.0.9002.zip(r-4.5)luz_0.4.0.9002.zip(r-4.4)luz_0.4.0.9002.zip(r-4.3)
luz_0.4.0.9002.tgz(r-4.4-any)luz_0.4.0.9002.tgz(r-4.3-any)
luz_0.4.0.9002.tar.gz(r-4.5-noble)luz_0.4.0.9002.tar.gz(r-4.4-noble)
luz_0.4.0.9002.tgz(r-4.4-emscripten)luz_0.4.0.9002.tgz(r-4.3-emscripten)
luz.pdf |luz.html✨
luz/json (API)
NEWS

# Install 'luz' in R:

install.packages('luz', repos = c('https://mlverse.r-universe.dev', 'https://cloud.r-project.org'))

Bug tracker:https://github.com/mlverse/luz/issues

On CRAN:

9.38 score 84 stars 1 packages 295 scripts 1.1k downloads 44 exports 35 dependencies

Last updated 2 months agofrom:bc41ef1053. Checks:OK: 5 NOTE: 2. Indexed: yes.

Target	Result	Date
Doc / Vignettes	OK	Nov 16 2024
R-4.5-win	NOTE	Nov 16 2024
R-4.5-linux	NOTE	Nov 16 2024
R-4.4-win	OK	Nov 16 2024
R-4.4-mac	OK	Nov 16 2024
R-4.3-win	OK	Nov 16 2024
R-4.3-mac	OK	Nov 16 2024

Exports:%>%accelerator as_dataloader evaluate fit get_metrics lr_finder luz_callback luz_callback_auto_resume luz_callback_csv_logger luz_callback_early_stopping luz_callback_gradient_clip luz_callback_interrupt luz_callback_keep_best_model luz_callback_lr_scheduler luz_callback_metrics luz_callback_mixed_precision luz_callback_mixup luz_callback_model_checkpoint luz_callback_profile luz_callback_progress luz_callback_resume_from_checkpoint luz_callback_tfevents luz_callback_train_valid luz_load luz_load_checkpoint luz_load_model_weights luz_metric luz_metric_accuracy luz_metric_binary_accuracy luz_metric_binary_accuracy_with_logits luz_metric_binary_auroc luz_metric_mae luz_metric_mse luz_metric_multiclass_auroc luz_metric_rmse luz_metric_set luz_save luz_save_model_weights nn_mixup_loss nnf_mixup set_hparams set_opt_hparams setup

Dependencies:bit bit64 callr cli colorspace coro crayon desc farver fs generics glue hms jsonlite labeling lifecycle magrittr munsell pkgconfig prettyunits processx progress ps purrr R6 RColorBrewer Rcpp rlang safetensors scales torch vctrs viridisLite withr zeallot

Accelerator API

Rendered fromaccelerator.Rmdusingknitr::rmarkdownon Nov 16 2024.

Last update: 2022-05-03
Started: 2021-05-24

Custom loops with luz

Rendered fromcustom-loop.Rmdusingknitr::rmarkdownon Nov 16 2024.

Last update: 2023-05-14
Started: 2021-05-10

Get started with luz

Rendered fromget-started.Rmdusingknitr::rmarkdownon Nov 16 2024.

Last update: 2023-04-14
Started: 2021-04-23

Citation

Development and contributors

Readme and manuals

Help Manual

Help page	Topics
Create an accelerator	accelerator
Creates a dataloader from its input	as_dataloader as_dataloader.array as_dataloader.dataloader as_dataloader.dataset as_dataloader.iterable_dataset as_dataloader.list as_dataloader.matrix as_dataloader.numeric as_dataloader.torch_tensor
Context object	context
Context object	ctx
Evaluates a fitted model on a dataset	evaluate
Fit a 'nn_module'	fit.luz_module_generator
Get metrics from the object	get_metrics get_metrics.luz_module_fitted
Learning Rate Finder	lr_finder
Create a new callback	luz_callback
Resume training callback	luz_callback_auto_resume
CSV logger callback	luz_callback_csv_logger
Early stopping callback	luz_callback_early_stopping
Gradient clipping callback	luz_callback_gradient_clip
Interrupt callback	luz_callback_interrupt
Keep the best model	luz_callback_keep_best_model
Learning rate scheduler callback	luz_callback_lr_scheduler
Metrics callback	luz_callback_metrics
Automatic Mixed Precision callback	luz_callback_mixed_precision
Mixup callback	luz_callback_mixup
Checkpoints model weights	luz_callback_model_checkpoint
Profile callback	luz_callback_profile
Progress callback	luz_callback_progress
Allow resume model training from a specific checkpoint	luz_callback_resume_from_checkpoint
tfevents callback	luz_callback_tfevents
Train-eval callback	luz_callback_train_valid
Load trained model	luz_load
Loads a checkpoint	luz_load_checkpoint
Loads model weights into a fitted object.	luz_load_model_weights luz_save_model_weights
Creates a new luz metric	luz_metric
Accuracy	luz_metric_accuracy
Binary accuracy	luz_metric_binary_accuracy
Binary accuracy with logits	luz_metric_binary_accuracy_with_logits
Computes the area under the ROC	luz_metric_binary_auroc
Mean absolute error	luz_metric_mae
Mean squared error	luz_metric_mse
Computes the multi-class AUROC	luz_metric_multiclass_auroc
Root mean squared error	luz_metric_rmse
Creates a metric set	luz_metric_set
Saves luz objects to disk	luz_save
Loss to be used with 'callbacks_mixup()'.	nn_mixup_loss
Mixup logic	nnf_mixup
Create predictions for a fitted model	predict.luz_module_fitted
Set hyper-parameter of a module	set_hparams
Set optimizer hyper-parameters	set_opt_hparams
Set's up a 'nn_module' to use with luz	setup