Metadata-Version: 2.1
Name: lpd
Version: 0.0.8
Summary: A Fast, Flexible Trainer and Extensions for Pytorch
Home-page: https://github.com/roysadaka/lpd
Author: Roy Sadaka
Maintainer: lpd developers
Maintainer-email: torch.lpd@gmail.com
License: MIT Licences
Description: ![Logo](https://raw.githubusercontent.com/RoySadaka/lpd/master/images/logo.png)
        
        # lpd
        
        A Fast, Flexible Trainer and Extensions for Pytorch
        
        ``lpd`` derives from the Hebrew word *lapid* (לפיד) which means "torch".
        
        ## For latest PyPI stable release
        ```sh
            pip install lpd
        ```
        
        ## Usage
        
        ``lpd`` intended to properly structure your pytorch model training. The main usages are given below.
        
        ### Training your model
        
        ```python
            from lpd.trainer import Trainer
            import lpd.utils.torch_utils as tu
            import lpd.callbacks as cbs 
            from lpd.callbacks import EpochEndStats, ModelCheckPoint, Tensorboard, EarlyStopping
            from lpd.extensions.custom_metrics import binary_accuracy_with_logits
        
            device = tu.get_gpu_device_if_available() # with fallback to CPU if GPU not avilable
            model = TestModel(config, num_embeddings).to(device) #this is your model class, and its being sent to the relevant device
            optimizer = optim.SGD(params=model.parameters())
            scheduler = optim.lr_scheduler.ReduceLROnPlateau(optimizer, mode='min', patience=5, verbose=True)
            loss_func = nn.BCEWithLogitsLoss().to(device) #this is your loss class, already sent to the relevant device
            metric_name_to_func = {"acc":binary_accuracy_with_logits} # add as much metrics as you like
        
            # you can use some of the defined callbacks, or you can create your own
            callbacks = [
                        SchedulerStep(scheduler_parameters_func=lambda trainer: trainer.val_stats.get_loss()), # notice lambda for scheduler that takes loss in step()
                        ModelCheckPoint(checkpoint_dir, checkpoint_file_name, monitor='val_loss', save_best_only=True, round_values_on_print_to=7), 
                        Tensorboard(summary_writer_dir=summary_writer_dir),
                        EarlyStopping(patience=10, monitor='val_loss'),
                        EpochEndStats(cb_phase=cbs.CB_ON_EPOCH_END, round_values_on_print_to=7) # better to put it last on the list (makes better sense in the log prints)
                    ]
        
            trainer = Trainer(model, 
                              device, 
                              loss_func, 
                              optimizer,
                              scheduler,
                              metric_name_to_func, 
                              train_data_loader,  #iterable or generator
                              val_data_loader,    #iterable or generator
                              train_steps,
                              val_steps,
                              num_epochs,
                              callbacks)
            
            trainer.train()
        ```
        
        ### Evaluating your model
        ```python
            trainer.evaluate(test_data_loader, test_steps)
        ```
        
        
        ### Callbacks
        Some common callbacks are available under ``lpd.callbacks``. 
        
        Notice that ``cb_phase`` will determine the execution phase.
        
        These are the current available phases, more will be added soon
        ```python
            CB_ON_TRAIN_BEGIN
            CB_ON_TRAIN_END  
            CB_ON_EPOCH_BEGIN
            CB_ON_EPOCH_END  
        ```
        
        ``EpochEndStats`` callback will print an epoch summary at the end of every epoch
        
        ![EpochSummary](https://raw.githubusercontent.com/RoySadaka/lpd/master/images/epoch_summary.png)
        
        You can also create your own callbacks
        
        ```python
            import lpd.callbacks as cbs
            from lpd.callbacks import CallbackBase
        
            class MyAwesomeCallback(CallbackBase):
                def __init__(self, cb_phase=cbs.CB_ON_TRAIN_BEGIN):
                    super(MyAwesomeCallback, self).__init__(cb_phase)
        
                def __call__(self, callback_context): # <=== implement this method!
                    # your implementation here
                    # using callback_context, you can access anything in your trainer
                    # below are some examples to get the hang of it
                    val_loss = callback_context.val_stats.get_loss()
                    train_loss = callback_context.train_stats.get_loss()
                    train_metrics = callback_context.train_stats.get_metrics()
                    val_metrics = callback_context.val_stats.get_metrics()
                    opt = callback_context.trainer.optimizer
                    scheduler = callback_context.trainer.scheduler
        ```
        
        ### Extensions
        ``lpd.extensions`` provides some custom pytorch layers, these are just some layers we like using when we create our models, to gain better flexibility.
        
        So you can use them at your own will, there youll also find custom metrics and schedulers.
        We will add more layers, metrics and schedulers from time to time.
        
        
        ## TODOS (more added frequently)
        * Add support for multiple schedulers 
        * Add support for multiple losses
        * EpochEndStats - save and print best accuracies
        
        ## Something is missing?! please share with us
        You can open an issue, but also feel free to email us at torch.lpd@gmail.com
        
Keywords: pytorch trainer extensions machine deep learning
Platform: UNKNOWN
Classifier: Development Status :: 5 - Production/Stable
Classifier: Intended Audience :: Developers
Classifier: Intended Audience :: Education
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: MacOS
Classifier: Operating System :: MacOS :: MacOS X
Classifier: Operating System :: Microsoft
Classifier: Operating System :: Microsoft :: MS-DOS
Classifier: Operating System :: Microsoft :: Windows
Classifier: Operating System :: POSIX
Classifier: Operating System :: POSIX :: BSD
Classifier: Operating System :: POSIX :: BSD :: FreeBSD
Classifier: Operating System :: POSIX :: Linux
Classifier: Operating System :: Unix
Classifier: Programming Language :: Python :: 3
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Classifier: Topic :: Utilities
Requires-Python: >=3.6
Description-Content-Type: text/markdown
