torchbearer: A model fitting library for PyTorch

Last update: Dec 13, 2022

Overview

Note: We're moving to PyTorch Lightning! Read about the move here. From the end of February, torchbearer will no longer be actively maintained. We'll continue to fix bugs when they are found and ensure that torchbearer runs on new versions of pytorch. However, we won't plan or implement any new functionality (if there's something you'd like to see in a training library, consider creating an issue on PyTorch Lightning).

Website • Docs • Examples • Install • Citing • Related

A PyTorch model fitting library designed for use by researchers (or anyone really) working in deep learning or differentiable programming. Specifically, we aim to dramatically reduce the amount of boilerplate code you need to write without limiting the functionality and openness of PyTorch.

Examples

General

	Quickstart: Get up and running with torchbearer, training a simple CNN on CIFAR-10.


	Callbacks: A detailed exploration of callbacks in torchbearer, with some useful visualisations.


	Imaging: A detailed exploration of the imaging sub-package in torchbearer, useful for showing visualisations during training.


Serialization: This guide gives an introduction to serializing and restarting training in torchbearer.


History and Replay: This guide gives an introduction to the history returned by a trial and the ability to replay training.


Custom Data Loaders: This guide gives an introduction on how to run custom data loaders in torchbearer.


Data Parallel: This guide gives an introduction to using torchbearer with DataParrallel.


	LiveLossPlot: A demonstration of the LiveLossPlot callback included in torchbearer.


	PyCM: A demonstration of the PyCM callback included in torchbearer for generating confusion matrices.


NVIDIA Apex: A guide showing how to perform half and mixed precision training in torchbearer with NVIDIA Apex.

Deep Learning

	Training a VAE: A demonstration of how to train (add do a simple visualisation of) a Variational Auto-Encoder (VAE) on MNIST with torchbearer.


	Training a GAN: A demonstration of how to train (add do a simple visualisation of) a Generative Adversarial Network (GAN) on MNIST with torchbearer.


	Generating Adversarial Examples: A demonstration of how to perform a simple adversarial attack with torchbearer.


	Transfer Learning with Torchbearer: A demonstration of how to perform transfer learning on STL10 with torchbearer.


	Regularisers in Torchbearer: A demonstration of how to use all of the built-in regularisers in torchbearer (Mixup, CutOut, CutMix, Random Erase, Label Smoothing and Sample Pairing).


Manifold Mixup: A demonstration of how to use the Manifold Mixup callback in Torchbearer.


	Class Appearance Model: A demonstration of the Class Appearance Model (CAM) callback in torchbearer.

Differentiable Programming

	Optimising Functions: An example (and some fun visualisations) showing how torchbearer can be used for the purpose of optimising functions with respect to their parameters using gradient descent.


	Linear SVM: Train a linear support vector machine (SVM) using torchbearer, with an interactive visualisation!


	Breaking Adam: The Adam optimiser doesn't always converge, in this example we reimplement some of the function optimisations from the AMSGrad paper showing this empirically.

Install

The easiest way to install torchbearer is with pip:

pip install torchbearer

Alternatively, build from source with:

pip install git+https://github.com/pytorchbearer/torchbearer

Citing Torchbearer

If you find that torchbearer is useful to your research then please consider citing our preprint: Torchbearer: A Model Fitting Library for PyTorch, with the following BibTeX entry:

@article{torchbearer2018,
  author = {Ethan Harris and Matthew Painter and Jonathon Hare},
  title = {Torchbearer: A Model Fitting Library for PyTorch},
  journal  = {arXiv preprint arXiv:1809.03363},
  year = {2018}
}

Torchbearer isn't the only library for training PyTorch models. Here are a few others that might better suit your needs (this is by no means a complete list, see the awesome pytorch list or the incredible pytorch for more):

skorch, model wrapper that enables use with scikit-learn - crossval etc. can be very useful
PyToune, simple Keras style API
ignite, advanced model training from the makers of PyTorch, can need a lot of code for advanced functions (e.g. Tensorboard)
TorchNetTwo (TNT), can be complex to use but well established, somewhat replaced by ignite
Inferno, training utilities and convenience classes for PyTorch
Pytorch Lightning, lightweight wrapper on top of PyTorch with advanced multi-gpu and cluster support
Pywick, high-level training framework, based on torchsample, support for various segmentation models

Comments

RuntimeError: CUDA error: CUBLAS_STATUS_ALLOC_FAILED when calling `cublasCreate(handle

Dear all,

It seems that torchbearer does not want to work for me. I am trying to simply classify images using resnet. You can find my code here (https://github.com/FrancescoSaverioZuppichini/PyTorch-Deep-Learning-Template/tree/feature/cuda-error), the main training logic is:

import time
from comet_ml import Experiment
import torchbearer
import torch.optim as optim
import torch.nn as nn
from torchsummary import summary
from Project import Project
from data import get_dataloaders
from data.transformation import train_transform, val_transform
from models import MyCNN, resnet18
from utils import device, show_dl
from torchbearer import Trial
from torchbearer.callbacks import CSVLogger, ModelCheckpoint, EarlyStopping, ReduceLROnPlateau
from callbacks import CometCallback
from logger import logging

if __name__ == '__main__':
    project = Project()
    # our hyperparameters
    params = {
        'lr': 0.001,
        'batch_size': 64,
        'epochs': 1,
        'model': 'resnet18-finetune',
        'id': time.time()
    }

    logging.info(f'Using device={device} 🚀')
    # everything starts with the data
    train_dl, val_dl, test_dl = get_dataloaders(
        project.data_dir,
        val_transform=val_transform,
        train_transform=train_transform,
        batch_size=params['batch_size'],
        num_workers=4,
    )
    # is always good practice to visualise some of the train and val images to be sure data-aug
    # is applied properly
    # show_dl(train_dl)
    # show_dl(test_dl)
    # define our comet experiment
    experiment = Experiment(api_key='8THqoAxomFyzBgzkStlY95MOf',
                            project_name="dl-pytorch-template", workspace="francescosaveriozuppichini")
    experiment.log_parameters(params)
    # create our special resnet18
    cnn = resnet18(n_classes=2).to(device)
    loss = nn.CrossEntropyLoss()
    # print the model summary to show useful information
    logging.info(summary(cnn, (3, 224, 244)))
    # define custom optimizer and instantiace the trainer `Model`
    optimizer = optim.Adam(cnn.parameters(), lr=params['lr'])
    # create our Trial object to train and evaluate the model
    trial = Trial(cnn, optimizer, loss, metrics=['acc', 'loss'],
                  callbacks=[
                      CometCallback(experiment),
                      ReduceLROnPlateau(monitor='val_loss',
                                        factor=0.1, patience=5),
                      EarlyStopping(monitor='val_acc', patience=5, mode='max'),
                      CSVLogger(str(project.checkpoint_dir / 'history.csv')),
                      ModelCheckpoint(str(project.checkpoint_dir / f'{params["id"]}-best.pt'), monitor='val_acc', mode='max')
    ]).to(device)
    trial.with_generators(train_generator=train_dl,
                          val_generator=val_dl, test_generator=test_dl)
    history = trial.run(epochs=params['epochs'], verbose=1)
    logging.info(history)
    preds = trial.evaluate(data_key=torchbearer.TEST_DATA)
    logging.info(f'test preds=({preds})')
    # experiment.log_metric('test_acc', test_acc)

I am running the same logic (same model) with poutyne and I have no problems. I really would like to switch to torchbearer

Error is:

2020-02-03 13:32:03,386 - [INFO] - None
  0%|                                                                                                                                                             | 0/1 [00:00<?, ?it/s]C:/w/1/s/tmp_conda_3.7_100118/conda/conda-bld/pytorch_1579082551706/work/aten/src/THCUNN/ClassNLLCriterion.cu:106: block: [0,0,0], thread: [2,0,0] Assertion `t >= 0 && t < n_classes` failed.
C:/w/1/s/tmp_conda_3.7_100118/conda/conda-bld/pytorch_1579082551706/work/aten/src/THCUNN/ClassNLLCriterion.cu:106: block: [0,0,0], thread: [13,0,0] Assertion `t >= 0 && t < n_classes` failed.
C:/w/1/s/tmp_conda_3.7_100118/conda/conda-bld/pytorch_1579082551706/work/aten/src/THCUNN/ClassNLLCriterion.cu:106: block: [0,0,0], thread: [17,0,0] Assertion `t >= 0 && t < n_classes` failed.
C:/w/1/s/tmp_conda_3.7_100118/conda/conda-bld/pytorch_1579082551706/work/aten/src/THCUNN/ClassNLLCriterion.cu:106: block: [0,0,0], thread: [20,0,0] Assertion `t >= 0 && t < n_classes` failed.
C:/w/1/s/tmp_conda_3.7_100118/conda/conda-bld/pytorch_1579082551706/work/aten/src/THCUNN/ClassNLLCriterion.cu:106: block: [0,0,0], thread: [21,0,0] Assertion `t >= 0 && t < n_classes` failed.
C:/w/1/s/tmp_conda_3.7_100118/conda/conda-bld/pytorch_1579082551706/work/aten/src/THCUNN/ClassNLLCriterion.cu:106: block: [0,0,0], thread: [22,0,0] Assertion `t >= 0 && t < n_classes` failed.
C:/w/1/s/tmp_conda_3.7_100118/conda/conda-bld/pytorch_1579082551706/work/aten/src/THCUNN/ClassNLLCriterion.cu:106: block: [0,0,0], thread: [23,0,0] Assertion `t >= 0 && t < n_classes` failed.
C:/w/1/s/tmp_conda_3.7_100118/conda/conda-bld/pytorch_1579082551706/work/aten/src/THCUNN/ClassNLLCriterion.cu:106: block: [0,0,0], thread: [25,0,0] Assertion `t >= 0 && t < n_classes` failed.
C:/w/1/s/tmp_conda_3.7_100118/conda/conda-bld/pytorch_1579082551706/work/aten/src/THCUNN/ClassNLLCriterion.cu:106: block: [0,0,0], thread: [29,0,0] Assertion `t >= 0 && t < n_classes` failed.
C:/w/1/s/tmp_conda_3.7_100118/conda/conda-bld/pytorch_1579082551706/work/aten/src/THCUNN/ClassNLLCriterion.cu:106: block: [0,0,0], thread: [30,0,0] Assertion `t >= 0 && t < n_classes` failed.
Traceback (most recent call last):
  File "c:/Users/Francesco/Documents/PyTorch-Deep-Learning-Template/main.py", line 64, in <module>
    history = trial.run(epochs=params['epochs'], verbose=1)
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\trial.py", line 133, in wrapper
    res = func(self, *args, **kwargs)
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\trial.py", line 988, in run
    final_metrics = self._fit_pass(state)[torchbearer.METRICS]
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\trial.py", line 298, in wrapper
    res = func(self, *args, **kwargs)
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\trial.py", line 1033, in _fit_pass
    state[torchbearer.OPTIMIZER].step(lambda: self.closure(state))
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torch\optim\adam.py", line 58, in step
    loss = closure()
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\trial.py", line 1033, in <lambda>
    state[torchbearer.OPTIMIZER].step(lambda: self.closure(state))
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\bases.py", line 382, in closure
    state[loss].backward(**state[torchbearer.BACKWARD_ARGS])
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\comet_ml\monkey_patching.py", line 246, in wrapper
    return_value = original(*args, **kwargs)
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torch\tensor.py", line 195, in backward
    torch.autograd.backward(self, gradient, retain_graph, create_graph)
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torch\autograd\__init__.py", line 99, in backward
    allow_unreachable=True)  # allow_unreachable flag
RuntimeError: CUDA error: CUBLAS_STATUS_ALLOC_FAILED when calling `cublasCreate(handle)`

Do your library work for you? Do you use it in your daily workflow?

Thank you.

Cheers,

Francesco Saverio

opened by FrancescoSaverioZuppichini 7

Accuracy computation for seq2seq model

0/10(t): 100%|██████████| 1000/1000 [02:30<00:00,  6.75it/s, running_loss=0.0338, running_acc=0.326, loss=0.689, loss_std=1.17, acc=35.9, acc_std=0]
0/10(v): 100%|██████████| 20/20 [00:01<00:00, 19.90it/s, val_loss=0.0341, val_loss_std=0.0122, val_acc=42, val_acc_std=0]
1/10(t): 100%|██████████| 1000/1000 [02:30<00:00,  6.76it/s, running_loss=0.00997, running_acc=0.327, loss=0.019, loss_std=0.0166, acc=41.8, acc_std=0]
1/10(v): 100%|██████████| 20/20 [00:01<00:00, 19.98it/s, val_loss=0.0126, val_loss_std=0.00798, val_acc=42.1, val_acc_std=0]
2/10(t): 100%|██████████| 1000/1000 [02:30<00:00,  6.75it/s, running_loss=0.00493, running_acc=0.328, loss=0.00837, loss_std=0.00938, acc=41.8, acc_std=0]
2/10(v): 100%|██████████| 20/20 [00:01<00:00, 19.89it/s, val_loss=0.00783, val_loss_std=0.00716, val_acc=42.2, val_acc_std=0]
3/10(t):  45%|████▌     | 454/1000 [01:08<01:21,  6.73it/s, running_loss=0.00458, running_acc=0.316]

Are the accuracies correct? (running_acc=.326, acc=35.9?)

I may be misunderstanding something, but shouldn't running_acc and acc be the same at the end of each epoch?

bug

opened by kl2792 6

Tqdm for Jupyter Notebook

Each iteration of TQDM starts a new line in Jupyter Notebook -- is there any way to integrate one of the suggested fixes into torchbearer?

(ref: https://github.com/tqdm/tqdm/issues/375, https://stackoverflow.com/a/47200571)
bug

opened by kl2792 6

ReduceLROnPlateau

Dear all,

first of all, I love this library.

The ReduceLROnPlateau is not working when I call trail.evaluate.

...
    trial = Trial(cnn, optimizer, loss, metrics=['acc', 'loss'],
                  callbacks=[
                    #   CometCallback(experiment),
                      ReduceLROnPlateau(monitor='val_loss',
                                        factor=0.1, patience=5),
                    #   EarlyStopping(monitor='val_acc', patience=5, mode='max'),
                    #   CSVLogger('history.csv'),
                    #   ModelCheckpoint('best.pt', monitor='val_acc', mode='max')
    ]).to(device)
    trial.with_generators(train_generator=train_dl,
                          val_generator=val_dl, test_generator=test_dl)
    # history = trial.run(params['epochs'], verbose=1)
    preds = trial.evaluate(data_key=torchbearer.TEST_DATA)

error:

0/1(e): 100%|███████████████████████████████████| 1/1 [00:00<00:00,  2.18it/s, test_acc=0.4667, test_loss=0.6516]
Traceback (most recent call last):
  File "c:/Users/Francesco/Documents/PyTorch-Deep-Learning-Template/main.py", line 62, in <module>
    preds = trial.evaluate(data_key=torchbearer.TEST_DATA)
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\trial.py", line 298, in wrapper
    res = func(self, *args, **kwargs)
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\trial.py", line 133, in wrapper
    res = func(self, *args, **kwargs)
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\trial.py", line 1131, in evaluate
    state[torchbearer.CALLBACK_LIST].on_end_epoch(state)
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\callbacks\callbacks.py", line 221, in on_end_epoch
    self._for_list(lambda callback: callback.on_end_epoch(state))
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\trial.py", line 105, in _for_list
    function(self.callback_list)
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\callbacks\callbacks.py", line 221, in <lambda>
    self._for_list(lambda callback: callback.on_end_epoch(state))
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\callbacks\callbacks.py", line 221, in on_end_epoch
    self._for_list(lambda callback: callback.on_end_epoch(state))
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\callbacks\callbacks.py", line 66, in _for_list
    function(callback)
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\callbacks\callbacks.py", line 221, in <lambda>
    self._for_list(lambda callback: callback.on_end_epoch(state))
  File "C:\Users\Francesco\Anaconda3\envs\dl\lib\site-packages\torchbearer\callbacks\torch_scheduler.py", line 32, in on_end_epoch
    self._scheduler.step(state[torchbearer.METRICS][self._monitor], epoch=state[torchbearer.EPOCH])
KeyError: 'val_loss'

Probably you need to disable the callback when evaluating (or just checking if the monitored metrics is in state['metrics'].

Thank you!

Best Regards,

Francesco Saverio

opened by FrancescoSaverioZuppichini 5

Trial predict fails with the given example
After training the model, I want to get the prediction on the test set. Not the accuracy. I know that should be Trial.evaluate(). It works well. Therefore I used Trial.predict(). Is that right?

But the error says that AttributeError: 'dict' object has no attribute 'data'.

I read the instruction of Trial that provides an example,

# Simple trial to predict on some validation and test data >>> from torchbearer import Trial >>> val_data = torch.rand(5, 5) >>> test_data = torch.rand(5, 5) >>> t = Trial(None).with_test_data(test_data) >>> test_predictions = t.predict(data_key=torchbearer.TEST_DATA)

I ran it but got an error AttributeError: 'NoneType' object has no attribute 'eval'

So, is there any problem in this method?
bug
opened by danielhuoo 5
Lean Model Checkpointing
Hi,

I ran a Trial and have my model saves my model using torchbearer.callbacks.checkpointers.Best to a file model.pt.

When I load the file with torch.load and run try to make a forward pass with it, I get the following error:

model = MyModule() state_dict = torch.load('vae.pt') model.load_state_dict(state_dict) # <== I get the error here

AttributeError: 'StateKey' object has no attribute 'startswith'

I get that model is being saved so that I can be recovered to be ready for torchbearer, but how can we save the model lean?

It seems like here, the model is only saved for reusability by torchbearer.

Thanks a lot!
docs
opened by dorukhansergin 5
loss_std resulting in complex number and breaking Tensorboard

I'm using torchbearer with PyTorch 0.4 and TensorboardX 1.2. Previously, I was using PyTorch 0.4.1, but I had to downgrade to use the TensorboardX because of a incompatibility with them. After adding the Tensorboard callback, the following error is raised after training for some time:

{TypeError}can't convert complex to float

When debugging, I noticed that the add_scalar() of TensorboardX tried to convert the scalar to float and, somehow, the val_loss_std was a complex number. Is there and error in how the std is calculated in order to result in a complex number?
bug

opened by fernandocamargoai 5
Support multi input and output
Right now, it's not possible to:

Have a Module multiple inputs (eg. forward(x1, x2)).

Have a Module with multiple outputs (returning a tuple).

I worked around the first problem by creating a Module with a single input and indexing each individual input. But the second problem makes it impossible to use the TripletMarginLoss, for example, since it expects 3 outputs from the Module.
opened by fernandocamargoai 3
Model checkpointers save_weights_only

As per the discussion in #504 it would be good if the checkpointers had an option to just save the model state dict, rather than the trial one. Not sure what the argument should be, something like save_model_only / save_weights_only? @dorukhansergin @MattPainter01 any thoughts on this?
enhancement

opened by ethanwharris 3
Running indefinitely?

Currently there is no way to ask torchbearer to run until stopped. This would be useful for reinforcement learning where we don't know how long an episode will be.
enhancement

opened by ethanwharris 3

lr_scheduler order changed since PyTorch 1.1.0

Many thanks for the wonderful library.

A warning message emerged when a scheduler was used in the callbacks: scheduler = torchbearer.callbacks.torch_scheduler.StepLR(step_size=5, gamma=0.1)

Hope it could be considered in later update if not yet included.

Python37\lib\site-packages\torch\optim\lr_scheduler.py:100: UserWarning: Detected call of `lr_scheduler.step()` before `optimizer.step()`. In PyTorch 1.1.0 and later, you should call them in the opposite order: `optimizer.step()` before `lr_scheduler.step()`.  Failure to do this will result in PyTorch skipping the first value of the learning rate schedule.See more details at https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate
  "https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate", UserWarning)

bug

opened by yongduek 2

Queston about training loop

Hi! I'm trying to fork the repo and add some functionality for an experiment. But that requires an addition in the training loop. I've read the documentation and the code but I can't seem to understand where the training loop itself is defined. Can somebody point me in the right direction?

Thanks in advance!

opened by AnabetsyR 4
GradientNormClipping callback error

When I insert this callback in the trial I get the following error. Is this some kind of bug? It seems like the gradients are not passed in the callback.

""" File "/home/dimitris/.local/lib/python3.6/site-packages/torch/nn/utils/clip_grad.py", line 30, in clip_grad_norm_ total_norm = torch.norm(torch.stack([torch.norm(p.grad.detach(), norm_type) for p in parameters]), norm_type) RuntimeError: stack expects a non-empty TensorList """

opened by dimimal 0
Native automatic mixed precision for torchbearer

Native automatic mixed precision support (torch.cuda.amp) is now in master: https://pytorch.org/docs/master/amp.html https://pytorch.org/docs/master/notes/amp_examples.html

Not sure if you ever tried Nvidia's (our) experimental Apex Amp, but I know it has many pain points (extension builds, forward/backward compatibilty, DataParallel support, flaky checkpointing, i don’t even know if it can be hacked to handle double backward/gradient penalty, others…). torch.cuda.amp fixes all these, the interface is more flexible and intuitive, and the tighter integration with pytorch brings more future optimizations into scope.

I think the torch.cuda.amp API is a good fit for a higher-level library because its style is more functional (as in, it doesn't statefully alter anything outside itself). The necessary torch.cuda.amp calls don't have silent/weird effects elsewhere.

If you want to talk about adding torch.cuda.amp to torchbearer, with an eye towards it becoming the future-proof source of mixed precision, message me on Pytorch slack anytime (or ask me for invites if you're not signed up). I'll check this issue periodically but I'm on Pytorch slack a greater fraction of the time than I care to admit.

opened by mcarilli 0
Y_pred tuple behaviour changed

When y_pred is a tuple (i.e. model returns multiple outputs) the criterion now recieves the tuple unpacked. This should either be reverted or more clearly documented.
bug

opened by ethanwharris 0

Releases(0.5.3)

0.5.3(Jan 31, 2020)
[0.5.3] - 2020-01-31

Added

Method in bases to access metrics

Changed

Deprecated

Removed

Fixed

Metric access bugs in various callbacks

Source code(tar.gz)
Source code(zip)
0.5.2(Jan 28, 2020)
[0.5.2] - 2020-28-01

Added

Added option to use mixup loss with cutmix

Support for PyTorch 1.4.0

Changed

Changed PyCM save methods to use *args and **kwargs

Deprecated

Removed

Fixed

Fixed a bug where the PyCM callback would fail when saving

Source code(tar.gz)
Source code(zip)
0.5.1(Nov 6, 2019)
[0.5.1] - 2019-11-06

Added

Added BCPlus callback for between-class learning

Added support for PyTorch 1.3

Added a show flag to the ImagingCallback.to_pyplot method, set to false to stop it from calling plt.show

Added manifold mixup

Changed

Changed the default behaviour of ImagingCallback.to_pyplot to turn off the axis

Deprecated

Removed

Fixed

Fixed a bug when resuming an old state dict with tqdm enabled

Fixed a bug in imaging where passing a title to to_pyplot was not possible

Source code(tar.gz)
Source code(zip)
0.5.0(Sep 17, 2019)
[0.5.0] - 2019-09-17

Added

Added PyTorch CyclicLR scheduler

Changed

Torchbearer now supports Modules with multiple inputs and multiple outputs

Deprecated

Removed

Cyclic LR callback in favour of torch cyclic lr scheduler

Removed support for PyTorch 0.4.x

Fixed

Fixed bug where aggregate predictions couldn't handle empty list

Fixed a bug where Runtime Errors on forward weren't handled properly

Fixed a bug where exceptions on forward wouldn't print the traceback properly

Fixed a documentation mistake whereby ReduceLROnPlateau was said to increase learning rate

Source code(tar.gz)
Source code(zip)
0.4.0(Sep 17, 2019)
[0.4.0] - 2019-07-05

Added

Added with_loader trial method that allows running of custom batch loaders

Added a Mock Model which is set when None is passed as the model to a Trial. Mock Model always returns None.

Added __call__(state) to StateKey so that they can now be used as losses

Added a callback to do cutout regularisation

Added a with_data trial method that allows passing of train, val and test data in one call

Added the missing on_init callback decorator

Added a step_on_batch flag to the early stopping callback

Added multi image support to imaging

Added a callback to unpack state into torchbearer.X at sample time for specified keys and update state after the forward pass based on model outputs. This is useful for using DataParallel which pass the main state dict directly.

Added callback for generating confusion matrices with PyCM

Added a mixup callback with associated loss

Added Label Smoothing Regularisation (LSR) callback

Added CutMix regularisation

Added default metric from paper for when Mixup loss is used

Changed

Changed history to now just be a list of records

Categorical Accuracy metric now also accepts tensors of size (B, C) and gets the max over C for the taget class

Deprecated

Removed

Removed the variational sub-package, this will now be packaged separately

Removed verbose argument from the early stopping callback

Fixed

Fixed a bug where list or dictionary metrics would cause the tensorboard callback to error

Fixed a bug where running a trial without training steps would error

Fixed a bug where the caching imaging callback didn't reset data so couldn't be run in multiple trials

Fixed a bug in the ClassAppearanceModel callback

Fixed a bug where the state given to predict was not a State object

Fixed a bug with Cutout on gpu

Fixed a bug where MakeGrid callback wasn't passing all arguments correctly

Fixed a bug in ImagingCallback that would sometimes cause make_grid to throw an error

Fixed a bug where the verbose argument would not work unless given as a keyword argument

Fixed a bug where the data_key argument would sometimes not work as expected

Fixed a bug where cutout required a seed

Fixed a bug where cutmix wasn't sendign the beta distribution sample to the device

Source code(tar.gz)
Source code(zip)
0.3.2(May 28, 2019)
[0.3.2] - 2019-05-28

Added

Changed

Deprecated

Removed

Fixed

Fixed a bug where for_steps would sometimes not work as expected if called in the wrong order

Fixed a bug where torchbearer installed via pip would crash on import

Source code(tar.gz)
Source code(zip)
0.3.1(May 24, 2019)
[0.3.1] - 2019-05-24

Added

Added cyclic learning rate finder

Added on_init callback hook to run at the end of trial init

Added callbacks for weight initialisation in torchbearer.callbacks.init

Added with_closure trial method that allows running of custom closures

Added base_closure function to bases that allows creation of standard training loop closures

Added ImagingCallback class for callbacks which produce images that can be sent to tensorboard, visdom or a file

Added CachingImagingCallback and MakeGrid callback to make a grid of images

Added the option to give the only_if callback decorator a function of self and state rather than just state

Added Layer-sequential unit-variance (LSUV) initialization

Added ClassAppearanceModel callback and example page for visualising CNNs

Added on_checkpoint callback decorator

Added support for PyTorch 1.1.0

Changed

No_grad and enable_grad decorators are now also context managers

Deprecated

Removed

Removed the fluent decorator, just use return self

Removed install dependency on torchvision, still required for some functionality

Fixed

Fixed bug where replay errored when train or val steps were None

Fixed a bug where mock optimser wouldn't call it's closure

Fixed a bug where the notebook check raised ModuleNotFoundError when IPython not installed

Fixed a memory leak with metrics that causes issues with very long epochs

Fixed a bug with the once and once_per_epoch decorators

Fixed a bug where the test criterion wouldn't accept a function of state

Fixed a bug where type inference would not work correctly when chaining Trial methods

Fixed a bug where checkpointers would error when they couldn't find the old checkpoint to overwrite

Fixed a bug where the 'test' label would sometimes not populate correctly in the default accuracy metric

Source code(tar.gz)
Source code(zip)
0.3.0(Feb 28, 2019)
[0.3.0] - 2019-02-28

Added

Added torchbearer.variational, a sub-package for implementations of state of the art variational auto-encoders

Added SimpleUniform and SimpleExponential distributions

Added a decorator which can be used to cite a research article as part of a doc string

Added an optional dimension argument to the mean, std and running_mean metric aggregators

Added a var metric and decorator which can be used to calculate the variance of a metric

Added an unbiased flag to the std and var metrics to optionally not apply Bessel's correction (consistent with torch.std / torch.var)

Added support for rounding 1D lists to the Tqdm callback

Added SimpleWeibull distribution

Added support for Python 2.7

Added SimpleWeibullSimpleWeibullKL

Added SimpleExponentialSimpleExponentialKL

Added the option for model parameters only saving to Checkpointers.

Added documentation about serialization.

Added support for indefinite data loading. Iterators can now be run until complete independent of epochs or iterators can be refreshed during an epoch if complete.

Added support for batch intervals in interval checkpointer

Added line magic %torchbearer notebook

Added 'accuracy' variants of 'acc' default metrics

Changed

Changed the default behaviour of the std metric to compute the sample std, in line with torch.std

Tqdm precision argument now rounds to decimal places rather than significant figures

Trial will now simply infer if the model has an argument called 'state'

Torchbearer now infers if inside a notebook and will use the appropriate tqdm module if not set

Deprecated

Removed

Removed the old Model API (deprecated since version 0.2.0)

Removed the 'pass_state' argument from Trial, this will now be inferred

Removed the 'std' decorator from the default metrics

Fixed

Fixed a bug in the weight decay callback which would result in potentially negative decay (now just uses torch.norm)

Fixed a bug in the cite decorator causing the citation to not show up correctly

Fixed a memory leak in the mse primitive metric

Source code(tar.gz)
Source code(zip)
0.2.6.1(Feb 25, 2019)
[0.2.6.1] - 2019-02-25

Fixed

Fixed a bug where predictions would multiply when predict was called more than once

Source code(tar.gz)
Source code(zip)
0.2.6(Dec 19, 2018)
[0.2.6] - 2018-12-19

Added

Changed

Y_PRED, Y_TRUE and X can now equivalently be accessed as PREDICTION, TARGET and INPUT respectively

Deprecated

Removed

Fixed

Fixed a bug where the LiveLossPlot callback would trigger an error if run and evaluate were called separately

Fixed a bug where state key errors would report to the wrong stack level

Fixed a bug where the user would wrongly get a state key error in some cases

Source code(tar.gz)
Source code(zip)
0.2.5(Dec 19, 2018)
[0.2.5] - 2018-12-19

Added

Added flag to replay to replay only a single batch per epoch

Added support for PyTorch 1.0.0 and Python 3.7

MetricTree can now unpack dictionaries from root, this is useful if you want to get a mean of a metric. However, this should be used with caution as it extracts only the first value in the dict and ignores the rest.

Added a callback for the livelossplot visualisation tool for notebooks

Changed

All error / accuracy metrics can now optionally take state keys for predictions and targets as arguments

Deprecated

Removed

Fixed

Fixed a bug with the EpochLambda metric which required y_true / y_pred to have specific forms

Source code(tar.gz)
Source code(zip)
0.2.4(Nov 16, 2018)
[0.2.4] - 2018-11-16

Added

Added metric functionality to state keys so that they can be used as metrics if desired

Added customizable precision to the printer callbacks

Added threshold to binary accuracy. Now it will appropriately handle any values in [0, 1]

Changed

Changed the default printer precision to 4s.f.

Tqdm on_epoch now shows metrics immediately when resuming

Deprecated

Removed

Fixed

Fixed a bug which would incorrectly trigger version warnings when loading in models

Fixed bugs where the Trial would not fail gracefully if required objects were not in state

Fixed a bug where none criterion didn't work with the add_to_loss callback

Fixed a bug where tqdm on_epoch always started at 0

Source code(tar.gz)
Source code(zip)
0.2.3(Oct 12, 2018)
[0.2.3] - 2018-10-12

Added

Added string representation of Trial to give summary

Added option to log Trial summary to TensorboardText

Added a callback point ('on_checkpoint') which can be used for model checkpointing after the history ios updated

Changed

When resuming training checkpointers no longer delete the state file the trial was loaded from

Changed the metric eval to include a data_key which tells us what data we are evaluating on

Deprecated

Removed

Fixed

Fixed a bug where callbacks weren't handled correctly in the predict and evaluate methods of Trial

Fixed a bug where the history wasn't updated when new metrics were calculated with the evaluate method of Trial

Fixed a bug where tensorboard writers couldn't be reused

Fixed a bug where the none criterion didn't require gradient

Fix bug where tqdm wouldn't get correct iterator length when evaluating on test generator

Fixed a bug where evaluating before training tried to update history before it existed

Fixed a bug where the metrics would output 'val_acc' even if evaluating on test or train data

Fixed a bug where roc metric didn't detach y_pred before sending to numpy

Fixed a bug where resuming from a checkpoint saved with one of the callbacks didn't populate the epoch number correctly

Source code(tar.gz)
Source code(zip)
0.2.2(Sep 18, 2018)
[0.2.2] - 2018-09-18

Added

The default_for_key metric decorator can now be used to pass arguments to the init of the inner metric

The default metric for the key 'top_10_acc' is now the TopKCategoricalAccuracy metric with k set to 10

Added global verbose flag for trial that can be overridden by run, evaluate, predict

Added an LR metric which retrieves the current learning rate from the optimizer, default for key 'lr'

Fixed

Fixed a bug where the DefaultAccuracy metric would not put the inner metric in eval mode if the first call to reset was after the call to eval

Fixed a bug where trying to load a state dict in a different session to where it was saved didn't work properly

Fixed a bug where the empty criterion would trigger an error if no Y_TRUE was put in state

Source code(tar.gz)
Source code(zip)
0.2.1(Sep 11, 2018)
[0.2.1] - 2018-09-11

Added

Evaluation and prediction can now be done on any data using data_key keywork arg

Text tensorboard/visdom logger that writes epoch/batch metrics to text

Changed

TensorboardX, Numpy, Scikit-learn and Scipy are no longer dependancies and only required if using the tensorboard callbacks or roc metric

Deprecated

Removed

Fixed

Model class setting generator incorrectly leading to stop iterations.

Argument ordering is consistent in Trial.with_generators and Trial.__init__

Added a state dict for the early stopping callback

Fixed visdom parameters not getting set in some cases

Source code(tar.gz)
Source code(zip)
0.2.0(Aug 21, 2018)
See [NEW!] in README.md for new key features

[0.2.0] - 2018-08-21

Added

Added the ability to pass custom arguments to the tqdm callback

Added an ignore_index flag to the categorical accuracy metric, similar to nn.CrossEntropyLoss. Usage: metrics=[CategoricalAccuracyFactory(ignore_index=0)]

Added TopKCategoricalAccuracy metric (default for key: top_5_acc)

Added BinaryAccuracy metric (default for key: binary_acc)

Added MeanSquaredError metric (default for key: mse)

Added DefaultAccuracy metric (use with 'acc' or 'accuracy') - infers accuracy from the criterion

New Trial api torchbearer.Trial to replace the Model api. Trial api is more atomic and uses the fluent pattern to allow chaining of methods.

torchbearer.Trial has with_x_generator and with_x_data methods to add training/validation/testing generators to the trial. There is a with_generators method to allow passing of all generators in one call.

torchbearer.Trial has for_x_steps and for_steps to allow running of trails without explicit generators or data tensors

torchbearer.Trial keeps a history of run calls which tracks number of epochs ran and the final metrics at each epoch. This allows seamless resuming of trial running.

torchbearer.Trial.state_dict now returns the trial history and callback list state allowing for full resuming of trials

torchbearer.Trial has a replay method that can replay training (with callbacks and display) from the history. This is useful when loading trials from state.

The backward call can now be passed args by setting state[torchbearer.BACKWARD_ARGS]

torchbearer.Trial implements the forward pass, loss calculation and backward call as a optimizer closure

Metrics are now explicitly calculated with no gradient

Changed

Callback decorators can now be chained to allow construction with multiple methods filled

Callbacks can now implement state_dict and ``load_state_dict` to allow callbacks to resume with state

State dictionary is now accepts StateKey objects which are unique and generated through torchbearer.state.get_state

State dictionary now warns when accessed with strings as this allows for collisions

Checkpointer callbacks will now resume from a state dict when resume=True in Trial

Deprecated

torchbearer.Model has been deprecated in favour of the new torchbearer.Trial api

Removed

Removed the MetricFactory class. Decorators still work in the same way but the Factory is no longer needed.

Fixed
Source code(tar.gz)
Source code(zip)
0.1.7(Aug 14, 2018)
[0.1.7] - 2018-08-14

Added

Added visdom logging support to tensorbard callbacks

Added option to choose tqdm module (tqdm, tqdm_notebook, ...) to Tqdm callback

Added some new decorators to simplify custom callbacks that must only run under certain conditions (or even just once).

Changed

Instantiation of Model will now trigger a warning pending the new Trial API in the next version

TensorboardX dependancy now version 1.4

Deprecated

Removed

Fixed

Mean and standard deviation calculations now work correctly for network outputs with many dimensions

Callback list no longer shared between fit calls, now a new copy is made each fit

Source code(tar.gz)
Source code(zip)
0.1.6(Aug 10, 2018)
[0.1.6] - 2018-08-10

Added

Added a verbose level (options are now 0,1,2) which will print progress for the entire fit call, updating every epoch. Useful when doing dynamic programming with little data.

Added support for dictionary outputs of dataloader

Added abstract superclass for building TensorBoardX based callbacks

Changed

Timer callback can now also be used as a metric which allows display of specified timings to printers and has been moved to metrics.

The loss_criterion is renamed to criterion in torchbearer.Model arguments.

The criterion in torchbearer.Model is now optional and will provide a zero loss tensor if it is not given.

TensorBoard callbacks refactored to be based on a common super class

TensorBoard callbacks refactored to use a common SummaryWriter for each log directory

Deprecated

Removed

Fixed

Standard deviation calculation now returns 0 instead of complex value when given very close samples

Source code(tar.gz)
Source code(zip)
0.1.5(Jul 30, 2018)
[0.1.5] - 2018-07-30

Added

Added a on_validation_criterion callback hook

Added a DatasetValidationSplitter which can be used to create a validation split if required for datasets like Cifar10 or MNIST

Added simple timer callback

Changed

Deprecated

Removed

Fixed

Fixed a bug where checkpointers would not save the model in some cases

Fixed a bug with the ROC metric causing it to not work

Source code(tar.gz)
Source code(zip)
0.1.4(Jul 23, 2018)
[0.1.4] - 2018-07-23

Added

Added a decorator API for metrics which allows decorators to be used for metric construction

Added a default_for_key decorator which can be used to associate a string with a given metric in metric lists

Added a decorator API for callbacks which allows decorators to be used for simple callback construction

Added a add_to_loss callback decorator which allows quicker constructions of callbacks that add values to the loss

Changed

Changed the API for running metrics and aggregators to no longer wrap a metric but instead receive input

Deprecated

Removed

Fixed
Source code(tar.gz)
Source code(zip)
0.1.3(Jul 18, 2018)
[0.1.3] - 2018-07-17

Added

Added a flag (step_on_batch) to the LR Scheduler callbacks which allows for step() to be called on each iteration instead of each epoch

Added on_sample_validation and on_forward_validation calls for validation callbacks

Added GradientClipping callback which simply clips the absolute gradient of the model parameters

Changed

Changed the order of the arguments to the lambda function in the EpochLambda metric for consistency with pytorch and other metrics

Checkpointers now create directory to savepath if it doesn't exist

Changed the 'on_forward_criterion' callback method to 'on_criterion'

Changed epoch number in printer callbacks to be consistent with the rest of torchbearer

Deprecated

Removed

Fixed

Fixed tests which were failing as of version 0.1.2

Fixed validation_steps not being added to state

Fixed checkpointer bug when path contained only filename and no directory path

Fixed console printer bug not printing validation statistics

Fixed console printer bug calling final_metrics before they existed in state

Source code(tar.gz)
Source code(zip)
v0.1.2(Jun 8, 2018)
[0.1.2] - 2018-06-08

Added

Added support for tuple outputs from generators, bink expects output to be length 2. Specifically, x, y = next() is possible, where x and y can be tuples of arbitrary size or depth

Added support for torch dtypes in bink Model.to(...)

Added pickle_module and pickle_protocol to checkpointers for consistency with torch.save

Changed

Changed the learning rate scheduler callbacks to no longer require an optimizer and to have the proper arguments

Deprecated

Removed

Fixed

Fixed an issue in GradientNormClipping which raised a warning in PyTorch >= 0.4

Source code(tar.gz)
Source code(zip)
v0.1.1(May 30, 2018)
Add support for tuples in data loader

Source code(tar.gz)
Source code(zip)
v0.1.0(May 24, 2018)

Source code(tar.gz)
Source code(zip)