| Type: | Package |
| Title: | Interface to 'fastai' |
| Version: | 2.2.2 |
| Maintainer: | Turgut Abdullayev <turqut.a.314@gmail.com> |
| Description: | The 'fastai' https://docs.fast.ai/index.html library simplifies training fast and accurate neural networks using modern best practices. It is based on research in to deep learning best practices undertaken at 'fast.ai', including 'out of the box' support for vision, text, tabular, audio, time series, and collaborative filtering models. |
| License: | Apache License 2.0 |
| URL: | https://github.com/EagerAI/fastai |
| BugReports: | https://github.com/EagerAI/fastai/issues |
| Encoding: | UTF-8 |
| RoxygenNote: | 7.3.1 |
| Imports: | reticulate, generics, png, ggplot2, ggpubr, glue |
| Suggests: | knitr, testthat, rmarkdown, curl, magrittr, data.table, vctrs, stats, utils, R.utils, viridis, zeallot |
| VignetteBuilder: | knitr |
| NeedsCompilation: | no |
| Packaged: | 2024-04-11 13:04:07 UTC; turqu |
| Author: | Turgut Abdullayev [ctb, cre, cph, aut] |
| Repository: | CRAN |
| Date/Publication: | 2024-04-11 14:40:07 UTC |
Floor mod
Description
Floor mod
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
x %% y
Arguments
x |
tensor |
y |
tensor |
Value
tensor
Floor divide
Description
Floor divide
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
x %/% y
Arguments
x |
tensor |
y |
tensor |
Value
tensor
Fastai assignment
Description
The assignment has to be used for safe modification of the values inside tensors/layers
Usage
left %f% right
Arguments
left |
left side object |
right |
right side object |
Value
None
Logical_and
Description
Logical_and
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
x & y
Arguments
x |
tensor |
y |
tensor |
Value
tensor
Greater
Description
Greater
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
a > b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Greater or equal
Description
Greater or equal
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
a >= b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Less
Description
Less
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
a < b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Less or equal
Description
Less or equal
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
a <= b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Multiply
Description
Multiply
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
a * b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Add
Description
Add
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
a + b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Add layers to Sequential
Description
Add layers to Sequential
Usage
## S3 method for class 'torch.nn.modules.container.Sequential'
a + b
Arguments
a |
sequential model |
b |
layer |
Value
model
Div
Description
Div
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
a / b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Equal
Description
Equal
Usage
## S3 method for class 'fastai.torch_core.TensorImage'
a == b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Equal
Description
Equal
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
a == b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Equal
Description
Equal
Usage
## S3 method for class 'torch.Tensor'
a == b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
APScoreBinary
Description
Average Precision for single-label binary classification problems
Usage
APScoreBinary(
axis = -1,
average = "macro",
pos_label = 1,
sample_weight = NULL
)
Arguments
axis |
axis |
average |
average |
pos_label |
pos_label |
sample_weight |
sample_weight |
Value
None
APScoreMulti
Description
Average Precision for multi-label classification problems
Usage
APScoreMulti(
sigmoid = TRUE,
average = "macro",
pos_label = 1,
sample_weight = NULL
)
Arguments
sigmoid |
sigmoid |
average |
average |
pos_label |
pos_label |
sample_weight |
sample_weight |
Value
None
AWD_LSTM
Description
AWD-LSTM inspired by https://arxiv.org/abs/1708.02182
Usage
AWD_LSTM(
vocab_sz,
emb_sz,
n_hid,
n_layers,
pad_token = 1,
hidden_p = 0.2,
input_p = 0.6,
embed_p = 0.1,
weight_p = 0.5,
bidir = FALSE
)
Arguments
vocab_sz |
vocab_sz |
emb_sz |
emb_sz |
n_hid |
n_hid |
n_layers |
n_layers |
pad_token |
pad_token |
|
hidden_p | |
input_p |
input_p |
embed_p |
embed_p |
weight_p |
weight_p |
bidir |
bidir |
Value
None
AWD_QRNN
Description
Same as an AWD-LSTM, but using QRNNs instead of LSTMs
Usage
AWD_QRNN(
vocab_sz,
emb_sz,
n_hid,
n_layers,
pad_token = 1,
hidden_p = 0.2,
input_p = 0.6,
embed_p = 0.1,
weight_p = 0.5,
bidir = FALSE
)
Arguments
vocab_sz |
vocab_sz |
emb_sz |
emb_sz |
n_hid |
n_hid |
n_layers |
n_layers |
pad_token |
pad_token |
|
hidden_p | |
input_p |
input_p |
embed_p |
embed_p |
weight_p |
weight_p |
bidir |
bidir |
Value
None
AccumMetric
Description
Stores predictions and targets on CPU in accumulate to perform final calculations with 'func'.
Usage
AccumMetric(
func,
dim_argmax = NULL,
activation = "no",
thresh = NULL,
to_np = FALSE,
invert_arg = FALSE,
flatten = TRUE,
...
)
Arguments
func |
function |
dim_argmax |
dimension argmax |
activation |
activation |
thresh |
threshold point |
to_np |
to matrix or not |
invert_arg |
invert arguments |
flatten |
flatten |
... |
additional arguments to pass |
Value
None
Adam
Description
Adam
Usage
Adam(...)
Arguments
... |
parameters to pass |
Value
None
AdaptiveAvgPool
Description
nn()$AdaptiveAvgPool layer for 'ndim'
Usage
AdaptiveAvgPool(sz = 1, ndim = 2)
Arguments
sz |
size |
ndim |
dimension size |
AdaptiveConcatPool1d
Description
Layer that concats 'AdaptiveAvgPool1d' and 'AdaptiveMaxPool1d'
Usage
AdaptiveConcatPool1d(size = NULL)
Arguments
size |
output size |
Value
None
AdaptiveConcatPool2d
Description
Layer that concats 'AdaptiveAvgPool2d' and 'AdaptiveMaxPool2d'
Usage
AdaptiveConcatPool2d(size = NULL)
Arguments
size |
output size |
Value
None
Adaptive GAN Switcher
Description
Switcher that goes back to generator/critic when the loss goes below 'gen_thresh'/'crit_thresh'.
Usage
AdaptiveGANSwitcher(gen_thresh = NULL, critic_thresh = NULL)
Arguments
gen_thresh |
generator threshold |
critic_thresh |
discriminator threshold |
Value
None
AdaptiveLoss
Description
Expand the 'target' to match the 'output' size before applying 'crit'.
Usage
AdaptiveLoss(crit)
Arguments
crit |
critic |
Value
Loss object
Add Channels
Description
Add 'n_dim' channels at the end of the input.
Usage
AddChannels(n_dim)
Arguments
n_dim |
number of dimensions |
Add Noise
Description
Adds noise of specified color and level to the audio signal
Usage
AddNoise(noise_level = 0.05, color = 0)
Arguments
noise_level |
noise level |
color |
int, color |
Value
None
AffineCoordTfm
Description
Combine and apply affine and coord transforms
Usage
AffineCoordTfm(
aff_fs = NULL,
coord_fs = NULL,
size = NULL,
mode = "bilinear",
pad_mode = "reflection",
mode_mask = "nearest",
align_corners = NULL
)
Arguments
aff_fs |
aff fs |
coord_fs |
coordinate fs |
size |
size |
mode |
mode |
pad_mode |
padding mode |
mode_mask |
mode mask |
align_corners |
align corners |
Value
None
AudioBlock
Description
A 'TransformBlock' for audios
Usage
AudioBlock(
cache_folder = NULL,
sample_rate = 16000,
force_mono = TRUE,
crop_signal_to = NULL
)
Arguments
cache_folder |
cache folder |
sample_rate |
sample rate |
force_mono |
force mono or not |
crop_signal_to |
int, crop signal |
Value
None
AudioBlock from folder
Description
Build a 'AudioBlock' from a 'path' and caches some intermediary results
Usage
AudioBlock_from_folder(
path,
sample_rate = 16000,
force_mono = TRUE,
crop_signal_to = NULL
)
Arguments
path |
directory, path |
sample_rate |
sample rate |
force_mono |
force mono or not |
crop_signal_to |
int, crop signal |
Value
None
AudioGetter
Description
Create 'get_audio_files' partial function that searches path suffix 'suf'
Usage
AudioGetter(suf = "", recurse = TRUE, folders = NULL)
Arguments
suf |
suffix |
recurse |
recursive or not |
folders |
vector, folders |
Details
and passes along 'kwargs', only in 'folders', if specified.
Value
None
AudioPadType module
Description
AudioPadType module
Usage
AudioPadType()
Value
None
AudioSpectrogram module
Description
AudioSpectrogram module
Usage
AudioSpectrogram()
Value
None
Audio Tensor
Description
Semantic torch tensor that represents an audio.
Usage
AudioTensor(x, sr = NULL)
Arguments
x |
tensor |
sr |
sr |
Value
tensor
AudioTensor create
Description
Creates audio tensor from file
Usage
AudioTensor_create(
fn,
cache_folder = NULL,
frame_offset = 0,
num_frames = -1,
normalize = TRUE,
channels_first = TRUE
)
Arguments
fn |
function |
cache_folder |
cache folder |
frame_offset |
offset |
num_frames |
number of frames |
normalize |
apply normalization or not |
channels_first |
channels first/last |
Value
None
AudioToMFCC
Description
Transform to create MFCC features from audio tensors.
Usage
AudioToMFCC(
sample_rate = 16000,
n_mfcc = 40,
dct_type = 2,
norm = "ortho",
log_mels = FALSE,
melkwargs = NULL
)
Arguments
sample_rate |
sample rate |
n_mfcc |
number of mel-frequency cepstral coefficients |
dct_type |
dct type |
norm |
normalization type |
log_mels |
apply log to mels |
melkwargs |
additional arguments for mels |
Value
None
AudioToMFCC from cfg
Description
Creates AudioToMFCC from configuration file
Usage
AudioToMFCC_from_cfg(audio_cfg)
Arguments
audio_cfg |
audio configuration |
Value
None
AudioToSpec from cfg
Description
Creates AudioToSpec from configuration file
Usage
AudioToSpec_from_cfg(audio_cfg)
Arguments
audio_cfg |
audio configuration |
Value
None
Auto configuration
Description
Auto configuration
Usage
AutoConfig()
Value
None
AvgLoss
Description
Flattens input and output, same as nn$AvgLoss
Usage
AvgLoss(...)
Arguments
... |
parameters to pass |
Value
Loss object
AvgPool
Description
nn$AvgPool layer for 'ndim'
Usage
AvgPool(ks = 2, stride = NULL, padding = 0, ndim = 2, ceil_mode = FALSE)
Arguments
ks |
kernel size |
stride |
the stride of the window. Default value is kernel_size |
padding |
implicit zero padding to be added on both sides |
ndim |
dimension number |
ceil_mode |
when True, will use ceil instead of floor to compute the output shape |
Value
None
AvgSmoothLoss
Description
Smooth average of the losses (exponentially weighted with 'beta')
Usage
AvgSmoothLoss(beta = 0.98)
Arguments
beta |
beta, defaults to 0.98 |
Value
Loss object
BBoxBlock
Description
A 'TransformBlock' for bounding boxes in an image
Usage
BBoxBlock()
Value
None
BBoxLabeler
Description
Delegates ('__call__','decode','setup') to ('encodes','decodes','setups') if 'split_idx' matches
Usage
BBoxLabeler(enc = NULL, dec = NULL, split_idx = NULL, order = NULL)
Arguments
enc |
encoder |
dec |
decoder |
split_idx |
split by index |
order |
order |
Value
None
BBoxLblBlock
Description
A 'TransformBlock' for labeled bounding boxes, potentially with 'vocab'
Usage
BBoxLblBlock(vocab = NULL, add_na = TRUE)
Arguments
vocab |
vocabulary |
add_na |
add NA |
Value
None'
Examples
## Not run:
URLs_COCO_TINY()
c(images, lbl_bbox) %<-% get_annotations('coco_tiny/train.json')
timg = Transform(ImageBW_create)
idx = 49
c(coco_fn,bbox) %<-% list(paste('coco_tiny/train',images[[idx]],sep = '/'),
lbl_bbox[[idx]])
coco_img = timg(coco_fn)
tbbox = LabeledBBox(TensorBBox(bbox[[1]]), bbox[[2]])
coco_bb = function(x) {
TensorBBox_create(bbox[[1]])
}
coco_lbl = function(x) {
bbox[[2]]
}
coco_dsrc = Datasets(c(rep(coco_fn,10)),
list(Image_create(), list(coco_bb),
list( coco_lbl, MultiCategorize(add_na = TRUE) )
), n_inp = 1)
coco_tdl = TfmdDL(coco_dsrc, bs = 9,
after_item = list(BBoxLabeler(), PointScaler(),
ToTensor()),
after_batch = list(IntToFloatTensor(), aug_transforms())
)
coco_tdl %>% show_batch(dpi = 200)
## End(Not run)
BCELossFlat
Description
Flattens input and output, same as nn$BCELoss
Usage
BCELossFlat(...)
Arguments
... |
parameters to pass |
Value
Loss object
BCEWithLogitsLossFlat
Description
BCEWithLogitsLossFlat
Usage
BCEWithLogitsLossFlat(...)
Arguments
... |
parameters to pass |
Value
Loss object
BalancedAccuracy
Description
Balanced Accuracy for single-label binary classification problems
Usage
BalancedAccuracy(axis = -1, sample_weight = NULL, adjusted = FALSE)
Arguments
axis |
axis |
sample_weight |
sample_weight |
adjusted |
adjusted |
References
None
BaseLoss
Description
Flattens input and output, same as nn$BaseLoss
Usage
BaseLoss(...)
Arguments
... |
parameters to pass |
Value
Loss object
BaseTokenizer
Description
Basic tokenizer that just splits on spaces
Usage
BaseTokenizer(split_char = " ")
Arguments
split_char |
separator |
Value
None
Basic MFCC
Description
Basic MFCC
Usage
BasicMFCC(
sample_rate = 16000,
n_mfcc = 40,
dct_type = 2,
norm = "ortho",
log_mels = FALSE,
melkwargs = NULL
)
Arguments
sample_rate |
sample rate |
n_mfcc |
number of mel-frequency cepstral coefficients |
dct_type |
dct type |
norm |
normalization type |
log_mels |
apply log to mels |
melkwargs |
additional arguments for mels |
Value
None
BasicMelSpectrogram
Description
BasicMelSpectrogram
Usage
BasicMelSpectrogram(
sample_rate = 16000,
n_fft = 400,
win_length = NULL,
hop_length = NULL,
f_min = 0,
f_max = NULL,
pad = 0,
n_mels = 128,
window_fn = torch()$hann_window,
power = 2,
normalized = FALSE,
wkwargs = NULL,
mel = TRUE,
to_db = TRUE
)
Arguments
sample_rate |
sample rate |
n_fft |
number of fast fourier transforms |
win_length |
windowing length |
hop_length |
hopping length |
f_min |
minimum frequency |
f_max |
maximum frequency |
pad |
padding |
n_mels |
number of mel-spectrograms |
window_fn |
window function |
power |
power |
normalized |
normalized or not |
wkwargs |
additional arguments |
mel |
mel-spectrogram or not |
to_db |
to decibels |
Value
None
BasicSpectrogram
Description
BasicSpectrogram
Usage
BasicSpectrogram(
n_fft = 400,
win_length = NULL,
hop_length = NULL,
pad = 0,
window_fn = torch()$hann_window,
power = 2,
normalized = FALSE,
wkwargs = NULL,
mel = FALSE,
to_db = TRUE
)
Arguments
n_fft |
number of fast fourier transforms |
win_length |
windowing length |
hop_length |
hopping length |
pad |
padding mode |
window_fn |
window function |
power |
power |
normalized |
normalized or not |
wkwargs |
additional arguments |
mel |
mel-spectrogram or not |
to_db |
to decibels |
Value
None
BatchNorm
Description
BatchNorm layer with 'nf' features and 'ndim' initialized depending on 'norm_type'.
Usage
BatchNorm(
nf,
ndim = 2,
norm_type = 1,
eps = 1e-05,
momentum = 0.1,
affine = TRUE,
track_running_stats = TRUE
)
Arguments
nf |
input shape |
ndim |
dimension number |
norm_type |
normalization type |
eps |
epsilon |
momentum |
momentum |
affine |
affine |
track_running_stats |
track running statistics |
Value
None
BatchNorm1dFlat
Description
'nn.BatchNorm1d', but first flattens leading dimensions
Usage
BatchNorm1dFlat(
num_features,
eps = 1e-05,
momentum = 0.1,
affine = TRUE,
track_running_stats = TRUE
)
Arguments
num_features |
number of features |
eps |
epsilon |
momentum |
momentum |
affine |
affine |
track_running_stats |
track running statistics |
Value
None
BrierScore
Description
Brier score for single-label classification problems
Usage
BrierScore(axis = -1, sample_weight = NULL, pos_label = NULL)
Arguments
axis |
axis |
sample_weight |
sample_weight |
pos_label |
pos_label |
Value
None
BrierScoreMulti
Description
Brier score for multi-label classification problems
Usage
BrierScoreMulti(
thresh = 0.5,
sigmoid = TRUE,
sample_weight = NULL,
pos_label = NULL
)
Arguments
thresh |
thresh |
sigmoid |
sigmoid |
sample_weight |
sample_weight |
pos_label |
pos_label |
Value
None
COCOMetric
Description
Wrapper around [cocoapi evaluator](https://github.com/cocodataset/cocoapi)
Usage
COCOMetric(
metric_type = COCOMetricType()$bbox,
print_summary = FALSE,
show_pbar = FALSE
)
Arguments
metric_type |
Dependent on the task you're solving. |
print_summary |
If 'TRUE', prints a table with statistics. |
show_pbar |
If 'TRUE' shows pbar when preparing the data for evaluation. |
Details
Calculates average precision. # Arguments metric_type: Dependent on the task you're solving. print_summary: If 'TRUE', prints a table with statistics. show_pbar: If 'TRUE' shows pbar when preparing the data for evaluation.
Value
None
COCOMetricType
Description
Available options for 'COCOMetric'
Usage
COCOMetricType()
Value
None
CSVLogger
Description
Basic class handling tweaks of the training loop by changing a 'Learner' in various events
Usage
CSVLogger(fname = "history.csv", append = FALSE)
Arguments
fname |
file name |
append |
append or not |
Value
None
Examples
## Not run:
URLs_MNIST_SAMPLE()
# transformations
tfms = aug_transforms(do_flip = FALSE)
path = 'mnist_sample'
bs = 20
#load into memory
data = ImageDataLoaders_from_folder(path, batch_tfms = tfms, size = 26, bs = bs)
learn = cnn_learner(data, resnet18(), metrics = accuracy, path = getwd())
learn %>% fit_one_cycle(2, cbs = CSVLogger())
## End(Not run)
Callback module
Description
Callback module
Usage
Callback()
Value
None
Cat
Description
Concatenate layers outputs over a given dim
Usage
Cat(layers, dim = 1)
Arguments
layers |
layers |
dim |
dimension size |
Value
None
Categorify
Description
Transform the categorical variables to that type.
Usage
Categorify(cat_names, cont_names)
Arguments
cat_names |
The names of the categorical variables |
cont_names |
The names of the continuous variables |
Value
None
CategoryBlock
Description
'TransformBlock' for single-label categorical targets
Usage
CategoryBlock(vocab = NULL, sort = TRUE, add_na = FALSE)
Arguments
vocab |
vocabulary |
sort |
sort or not |
add_na |
add NA |
Value
Block object
Change Volume
Description
Changes the volume of the signal
Usage
ChangeVolume(p = 0.5, lower = 0.5, upper = 1.5)
Arguments
p |
probability |
lower |
lower bound |
upper |
upper bound |
Value
None
ClassificationInterpretation_from_learner
Description
Construct interpretation object from a learner
Usage
ClassificationInterpretation_from_learner(
learn,
ds_idx = 1,
dl = NULL,
act = NULL
)
Arguments
learn |
learner/model |
ds_idx |
ds by index |
dl |
dataloader |
act |
activation |
Value
interpretation object
CohenKappa
Description
Cohen kappa for single-label classification problems
Usage
CohenKappa(axis = -1, labels = NULL, weights = NULL, sample_weight = NULL)
Arguments
axis |
axis |
labels |
labels |
weights |
weights |
sample_weight |
sample_weight |
Value
None
ColReader
Description
Read 'cols' in 'row' with potential 'pref' and 'suff'
Usage
ColReader(cols, pref = "", suff = "", label_delim = NULL)
Arguments
cols |
columns |
pref |
pref |
suff |
suffix |
label_delim |
label separator |
Value
None
ColSplitter
Description
Split 'items' (supposed to be a dataframe) by value in 'col'
Usage
ColSplitter(col = "is_valid")
Arguments
col |
column |
Value
None
CollabDataLoaders_from_dblock
Description
Create a dataloaders from a given 'dblock'
Usage
CollabDataLoaders_from_dblock(
dblock,
source,
path = ".",
bs = 64,
val_bs = NULL,
shuffle_train = TRUE,
device = NULL
)
Arguments
dblock |
dblock |
source |
source |
path |
The folder where to work |
bs |
The batch size |
val_bs |
The batch size for the validation DataLoader (defaults to bs) |
shuffle_train |
If we shuffle the training DataLoader or not |
device |
device |
Value
None
CollabDataLoaders_from_df
Description
Create a 'DataLoaders' suitable for collaborative filtering from 'ratings'.
Usage
CollabDataLoaders_from_df(
ratings,
valid_pct = 0.2,
user_name = NULL,
item_name = NULL,
rating_name = NULL,
seed = NULL,
path = ".",
bs = 64,
val_bs = NULL,
shuffle_train = TRUE,
device = NULL
)
Arguments
ratings |
ratings |
valid_pct |
The random percentage of the dataset to set aside for validation (with an optional seed) |
user_name |
The name of the column containing the user (defaults to the first column) |
item_name |
The name of the column containing the item (defaults to the second column) |
rating_name |
The name of the column containing the rating (defaults to the third column) |
seed |
random seed |
path |
The folder where to work |
bs |
The batch size |
val_bs |
The batch size for the validation DataLoader (defaults to bs) |
shuffle_train |
If we shuffle the training DataLoader or not |
device |
the device, e.g. cpu, cuda, and etc. |
Value
None
Examples
## Not run:
URLs_MOVIE_LENS_ML_100k()
c(user,item,title) %<-% list('userId','movieId','title')
ratings = fread('ml-100k/u.data', col.names = c(user,item,'rating','timestamp'))
movies = fread('ml-100k/u.item', col.names = c(item, 'title', 'date', 'N', 'url',
paste('g',1:19,sep = '')))
rating_movie = ratings[movies[, .SD, .SDcols=c(item,title)], on = item]
dls = CollabDataLoaders_from_df(rating_movie, seed = 42, valid_pct = 0.1, bs = 64,
item_name=title, path='ml-100k')
## End(Not run)
CollectDataCallback
Description
Collect all batches, along with pred and loss, into self.data. Mainly for testing
Usage
CollectDataCallback(...)
CollectDataCallback(...)
Arguments
... |
arguments to pass |
Value
None
None
Contrast
Description
Apply change in contrast of 'max_lighting' to batch of images with probability 'p'.
Usage
Contrast(max_lighting = 0.2, p = 0.75, draw = NULL, batch = FALSE)
Arguments
max_lighting |
maximum lighting |
p |
probability |
draw |
draw |
batch |
batch |
Value
None
ConvLayer
Description
Create a sequence of convolutional ('ni' to 'nf'), ReLU (if 'use_activ') and 'norm_type' layers.
Usage
ConvLayer(
ni,
nf,
ks = 3,
stride = 1,
padding = NULL,
bias = NULL,
ndim = 2,
norm_type = 1,
bn_1st = TRUE,
act_cls = nn()$ReLU,
transpose = FALSE,
init = "auto",
xtra = NULL,
bias_std = 0.01,
dilation = 1,
groups = 1,
padding_mode = "zeros"
)
Arguments
ni |
number of inputs |
nf |
outputs/ number of features |
ks |
kernel size |
stride |
stride |
padding |
padding |
bias |
bias |
ndim |
dimension number |
norm_type |
normalization type |
bn_1st |
batch normalization 1st |
act_cls |
activation |
transpose |
transpose |
init |
initializer |
xtra |
xtra |
bias_std |
bias standard deviation |
dilation |
specify the dilation rate to use for dilated convolution |
groups |
groups size |
padding_mode |
padding mode, e.g 'zeros' |
Value
None
CorpusBLEUMetric
Description
Blueprint for defining a metric
Usage
CorpusBLEUMetric(vocab_sz = 5000, axis = -1)
Arguments
vocab_sz |
vocab_sz |
axis |
axis |
Value
None
CropPad
Description
Center crop or pad an image to 'size'
Usage
CropPad(size, pad_mode = "zeros", ...)
Arguments
size |
size |
pad_mode |
padding mode |
... |
additional arguments |
Value
None
Crop Time
Description
Random crops full spectrogram to be length specified in ms by crop_duration
Usage
CropTime(duration, pad_mode = AudioPadType()$Zeros)
Arguments
duration |
int, duration |
pad_mode |
padding mode, by default 'AudioPadType$Zeros' |
Value
None
CrossEntropyLossFlat
Description
Same as 'nn$Module', but no need for subclasses to call 'super().__init__'
Usage
CrossEntropyLossFlat(...)
Arguments
... |
parameters to pass |
Value
Loss object
CudaCallback
Description
Move data to CUDA device
Usage
CudaCallback(device = NULL)
Arguments
device |
device name |
Value
None
CutMix
Description
Implementation of 'https://arxiv.org/abs/1905.04899'
Usage
CutMix(alpha = 1)
Arguments
alpha |
alpha |
Value
None
CycleGAN
Description
CycleGAN model.
Usage
CycleGAN(
ch_in = 3,
ch_out = 3,
n_features = 64,
disc_layers = 3,
gen_blocks = 9,
lsgan = TRUE,
drop = 0,
norm_layer = NULL
)
Arguments
ch_in |
input |
ch_out |
output |
n_features |
number of features |
disc_layers |
discriminator layers |
gen_blocks |
generator blocks |
lsgan |
ls gan |
drop |
dropout rate |
norm_layer |
normalziation layer |
Details
When called, takes in input batch of real images from both domains and outputs fake images for the opposite domains (with the generators). Also outputs identity images after passing the images into generators that outputs its domain type (needed for identity loss). Attributes: 'G_A' ('nn.Module'): takes real input B and generates fake input A 'G_B' ('nn.Module'): takes real input A and generates fake input B 'D_A' ('nn.Module'): trained to make the difference between real input A and fake input A 'D_B' ('nn.Module'): trained to make the difference between real input B and fake input B
Value
None
CycleGANLoss
Description
CycleGAN loss function. The individual loss terms are also atrributes of this class that are accessed by fastai for recording during training.
Usage
CycleGANLoss(cgan, l_A = 10, l_B = 10, l_idt = 0.5, lsgan = TRUE)
Arguments
cgan |
The CycleGAN model. |
l_A |
lambda_A, weight of domain A losses. (default=10) |
l_B |
lambda_B, weight of domain B losses. (default=10) |
l_idt |
lambda_idt, weight of identity lossees. (default=0.5) |
lsgan |
Whether or not to use LSGAN objective (default=True) |
Details
Attributes: 'self.cgan' ('nn.Module'): The CycleGAN model. 'self.l_A' ('float'): lambda_A, weight of domain A losses. 'self.l_B' ('float'): lambda_B, weight of domain B losses. 'self.l_idt' ('float'): lambda_idt, weight of identity lossees. 'self.crit' ('AdaptiveLoss'): The adversarial loss function (either a BCE or MSE loss depending on 'lsgan' argument) 'self.real_A' and 'self.real_B' ('fastai.torch_core.TensorImage'): Real images from domain A and B. 'self.id_loss_A' ('torch.FloatTensor'): The identity loss for domain A calculated in the forward function 'self.id_loss_B' ('torch.FloatTensor'): The identity loss for domain B calculated in the forward function 'self.gen_loss' ('torch.FloatTensor'): The generator loss calculated in the forward function 'self.cyc_loss' ('torch.FloatTensor'): The cyclic loss calculated in the forward function
CycleGANTrainer
Description
Learner Callback for training a CycleGAN model.
Usage
CycleGANTrainer(...)
Arguments
... |
parameters to pass |
Value
None
DataBlock
Description
Generic container to quickly build 'Datasets' and 'DataLoaders'
Usage
DataBlock(
blocks = NULL,
dl_type = NULL,
getters = NULL,
n_inp = NULL,
item_tfms = NULL,
batch_tfms = NULL,
...
)
Arguments
blocks |
input blocks |
dl_type |
DL application |
getters |
how to get dataet |
n_inp |
n_inp is the number of elements in the tuples that should be considered part of the input and will default to 1 if tfms consists of one set of transforms |
item_tfms |
One or several transforms applied to the items before batching them |
batch_tfms |
One or several transforms applied to the batches once they are formed |
... |
additional parameters to pass |
Value
Block object
Data Loaders
Description
Data Loaders
Usage
Data_Loaders(...)
Arguments
... |
parameters to pass |
Value
loader object
Examples
## Not run:
data = Data_Loaders(train_loader, test_loader)
learn = Learner(data, Net(), loss_func = F$nll_loss,
opt_func = Adam(), metrics = accuracy, cbs = CudaCallback())
learn %>% fit_one_cycle(1, 1e-2)
## End(Not run)
Datasets
Description
A dataset that creates a list from each 'tfms', passed thru 'item_tfms'
Usage
Datasets(
items = NULL,
tfms = NULL,
tls = NULL,
n_inp = NULL,
dl_type = NULL,
use_list = NULL,
do_setup = TRUE,
split_idx = NULL,
train_setup = TRUE,
splits = NULL,
types = NULL,
verbose = FALSE
)
Arguments
items |
items |
tfms |
transformations |
tls |
tls |
n_inp |
n_inp |
dl_type |
DL type |
use_list |
use list |
do_setup |
do setup |
split_idx |
split by index |
train_setup |
train setup |
splits |
splits |
types |
types |
verbose |
verbose |
Value
None
Debugger
Description
A module to debug inside a model
Usage
Debugger(...)
Arguments
... |
parameters to pass |
Value
None
Delta
Description
Creates delta with order 1 and 2 from spectrogram and concatenate with the original
Usage
Delta(width = 9)
Arguments
width |
int, width |
Value
None
Dense Res Block
Description
Resnet block of 'nf' features. 'conv_kwargs' are passed to 'conv_layer'.
Usage
DenseResBlock(
nf,
norm_type = 1,
ks = 3,
stride = 1,
padding = NULL,
bias = NULL,
ndim = 2,
bn_1st = TRUE,
act_cls = nn()$ReLU,
transpose = FALSE,
init = "auto",
xtra = NULL,
bias_std = 0.01,
dilation = 1,
groups = 1,
padding_mode = "zeros"
)
Arguments
nf |
number of features |
norm_type |
normalization type |
ks |
kernel size |
stride |
stride |
padding |
padding |
bias |
bias |
ndim |
number of dimensions |
bn_1st |
batch normalization 1st |
act_cls |
activation |
transpose |
transpose |
init |
initizalier |
xtra |
xtra |
bias_std |
bias standard deviation |
dilation |
dilation number |
groups |
groups number |
padding_mode |
padding mode |
Value
block
DeterministicDihedral
Description
Apply a random dihedral transformation to a batch of images with a probability 'p'
Usage
DeterministicDihedral(
size = NULL,
mode = "bilinear",
pad_mode = "reflection",
align_corners = NULL
)
Arguments
size |
size |
mode |
mode |
pad_mode |
padding mode |
align_corners |
align corners |
Value
None
DeterministicDraw
Description
DeterministicDraw
Usage
DeterministicDraw(vals)
Arguments
vals |
values |
Value
None
DeterministicFlip
Description
Flip the batch every other call
Usage
DeterministicFlip(
size = NULL,
mode = "bilinear",
pad_mode = "reflection",
align_corners = TRUE,
...
)
Arguments
size |
size |
mode |
mode |
pad_mode |
padding mode |
align_corners |
align corners |
... |
parameters to pass |
Value
None
Dice coefficient
Description
Dice coefficient metric for binary target in segmentation
Usage
Dice(axis = 1)
Arguments
axis |
axis |
Value
None
Dicom class
Description
Dicom class
Usage
Dicom()
Value
None
Dihedral
Description
Apply a random dihedral transformation to a batch of images with a probability 'p'
Apply a random dihedral transformation to a batch of images with a probability 'p'
Usage
Dihedral(
p = 0.5,
draw = NULL,
size = NULL,
mode = "bilinear",
pad_mode = "reflection",
align_corners = NULL,
batch = FALSE
)
Dihedral(
p = 0.5,
draw = NULL,
size = NULL,
mode = "bilinear",
pad_mode = "reflection",
align_corners = NULL,
batch = FALSE
)
Arguments
p |
probability |
draw |
draw |
size |
size |
mode |
mode |
pad_mode |
padding mode |
align_corners |
align corners |
batch |
batch |
Value
None
None
DihedralItem
Description
Randomly flip with probability 'p'
Usage
DihedralItem(p = 1, nm = NULL, before_call = NULL)
Arguments
p |
probability |
nm |
nm |
before_call |
before call |
Value
None
Downmix Mono
Description
Transform multichannel audios into single channel
Usage
DownmixMono(enc = NULL, dec = NULL, split_idx = NULL, order = NULL)
Arguments
enc |
encoder |
dec |
decoder |
split_idx |
split by index |
order |
order, by default is NULL |
Value
None
DynamicUnet
Description
Create a U-Net from a given architecture.
Usage
DynamicUnet(
encoder,
n_classes,
img_size,
blur = FALSE,
blur_final = TRUE,
self_attention = FALSE,
y_range = NULL,
last_cross = TRUE,
bottle = FALSE,
act_cls = nn()$ReLU,
init = nn()$init$kaiming_normal_,
norm_type = NULL
)
Arguments
encoder |
encoder |
n_classes |
number of classes |
img_size |
image size |
blur |
blur is used to avoid checkerboard artifacts at each layer. |
blur_final |
blur final is specific to the last layer. |
self_attention |
self_attention determines if we use a self attention layer at the third block before the end. |
y_range |
If y_range is passed, the last activations go through a sigmoid rescaled to that range. |
last_cross |
last cross |
bottle |
bottle |
act_cls |
activation |
init |
initializer |
norm_type |
normalization type |
Value
None
EarlyStoppingCallback
Description
EarlyStoppingCallback
Usage
EarlyStoppingCallback(...)
Arguments
... |
parameters to pass |
Value
None
Embedding
Description
Embedding layer with truncated normal initialization
Usage
Embedding(ni, nf)
Arguments
ni |
inputs |
nf |
outputs / number of features |
Value
None
EmbeddingDropout
Description
Apply dropout with probability 'embed_p' to an embedding layer 'emb'.
Usage
EmbeddingDropout(emb, embed_p)
Arguments
emb |
emb |
embed_p |
embed_p |
Value
None
Explained Variance
Description
Explained variance between predictions and targets
Usage
ExplainedVariance(sample_weight = NULL)
Arguments
sample_weight |
sample_weight |
Value
None
F1Score
Description
F1 score for single-label classification problems
Usage
F1Score(
axis = -1,
labels = NULL,
pos_label = 1,
average = "binary",
sample_weight = NULL
)
Arguments
axis |
axis |
labels |
labels |
pos_label |
pos_label |
average |
average |
sample_weight |
sample_weight |
Value
None
F1ScoreMulti
Description
F1 score for multi-label classification problems
Usage
F1ScoreMulti(
thresh = 0.5,
sigmoid = TRUE,
labels = NULL,
pos_label = 1,
average = "macro",
sample_weight = NULL
)
Arguments
thresh |
thresh |
sigmoid |
sigmoid |
labels |
labels |
pos_label |
pos_label |
average |
average |
sample_weight |
sample_weight |
Value
None
FBeta
Description
FBeta score with 'beta' for single-label classification problems
Usage
FBeta(
beta,
axis = -1,
labels = NULL,
pos_label = 1,
average = "binary",
sample_weight = NULL
)
Arguments
beta |
beta |
axis |
axis |
labels |
labels |
pos_label |
pos_label |
average |
average |
sample_weight |
sample_weight |
Value
None
FBetaMulti
Description
FBeta score with 'beta' for multi-label classification problems
Usage
FBetaMulti(
beta,
thresh = 0.5,
sigmoid = TRUE,
labels = NULL,
pos_label = 1,
average = "macro",
sample_weight = NULL
)
Arguments
beta |
beta |
thresh |
thresh |
sigmoid |
sigmoid |
labels |
labels |
pos_label |
pos_label |
average |
average |
sample_weight |
sample_weight |
Value
None
FetchPredsCallback
Description
A callback to fetch predictions during the training loop
Usage
FetchPredsCallback(
ds_idx = 1,
dl = NULL,
with_input = FALSE,
with_decoded = FALSE,
cbs = NULL,
reorder = TRUE
)
Arguments
ds_idx |
dataset index |
dl |
DL application |
with_input |
with input or not |
with_decoded |
with decoded or not |
cbs |
callbacks |
reorder |
reorder or not |
Value
None
File Splitter
Description
Split 'items' by providing file 'fname' (contains names of valid items separated by newline).
Usage
FileSplitter(fname)
Arguments
fname |
file name |
Value
None
Fill Missing
Description
Fill the missing values in continuous columns.
Usage
FillMissing(
cat_names,
cont_names,
fill_strategy = FillStrategy_MEDIAN(),
add_col = TRUE,
fill_val = 0
)
Arguments
cat_names |
The names of the categorical variables |
cont_names |
The names of the continuous variables |
fill_strategy |
The strategy of filling |
add_col |
add_col |
fill_val |
fill_val |
Value
None
Examples
## Not run:
procs = list(FillMissing(),Categorify(),Normalize())
## End(Not run)
COMMON
Description
An enumeration.
Usage
FillStrategy_COMMON()
Value
None
CONSTANT
Description
An enumeration.
Usage
FillStrategy_CONSTANT()
Value
None
MEDIAN
Description
An enumeration.
Usage
FillStrategy_MEDIAN()
Value
None
Fixed GAN Switcher
Description
Switcher to do 'n_crit' iterations of the critic then 'n_gen' iterations of the generator.
Usage
FixedGANSwitcher(n_crit = 1, n_gen = 1)
Arguments
n_crit |
number of discriminator |
n_gen |
number of generator |
Value
None
Flatten
Description
Flatten 'x' to a single dimension, e.g. at end of a model. 'full' for rank-1 tensor
Usage
Flatten(full = FALSE)
Arguments
full |
bool, full or not |
Flip
Description
Randomly flip a batch of images with a probability 'p'
Usage
Flip(
p = 0.5,
draw = NULL,
size = NULL,
mode = "bilinear",
pad_mode = "reflection",
align_corners = TRUE,
batch = FALSE
)
Arguments
p |
probability |
draw |
draw |
size |
size of image |
mode |
mode |
pad_mode |
reflection, zeros, border as string parameter |
align_corners |
align corners ot not |
batch |
batch or not |
Value
None
FlipItem
Description
Randomly flip with probability 'p'
Usage
FlipItem(p = 0.5)
Arguments
p |
probability |
Value
None
FolderDataset
Description
A PyTorch Dataset class that can be created from a folder 'path' of images, for the sole purpose of inference. Optional 'transforms'
Usage
FolderDataset(path, transforms = NULL)
Arguments
path |
path to dir |
transforms |
transformations |
Details
can be provided. Attributes: 'self.files': A list of the filenames in the folder. 'self.totensor': 'torchvision.transforms.ToTensor' transform. 'self.transform': The transforms passed in as 'transforms' to the constructor.
Value
None
ForgetMultGPU
Description
Wrapper around the CUDA kernels for the ForgetMult gate.
Usage
ForgetMultGPU(...)
Arguments
... |
parameters to pass |
Value
None
FuncSplitter
Description
Split 'items' by result of 'func' ('TRUE' for validation, 'FALSE' for training set).
Usage
FuncSplitter(func)
Arguments
func |
function |
Value
None
GAN Discriminative LR
Description
'Callback' that handles multiplying the learning rate by 'mult_lr' for the critic.
Usage
GANDiscriminativeLR(mult_lr = 5)
Arguments
mult_lr |
mult learning rate |
GAN Learner from learners
Description
Create a GAN from 'learn_gen' and 'learn_crit'.
Usage
GANLearner_from_learners(
gen_learn,
crit_learn,
switcher = NULL,
weights_gen = NULL,
gen_first = FALSE,
switch_eval = TRUE,
show_img = TRUE,
clip = NULL,
cbs = NULL,
metrics = NULL,
loss_func = NULL,
opt_func = Adam(),
lr = 0.001,
splitter = trainable_params(),
path = NULL,
model_dir = "models",
wd = NULL,
wd_bn_bias = FALSE,
train_bn = TRUE,
moms = list(0.95, 0.85, 0.95)
)
Arguments
gen_learn |
generator learner |
crit_learn |
discriminator learner |
switcher |
switcher |
weights_gen |
weights generator |
gen_first |
generator first |
switch_eval |
switch evaluation |
show_img |
show image or not |
clip |
clip value |
cbs |
Cbs is one or a list of Callbacks to pass to the Learner. |
metrics |
It is an optional list of metrics, that can be either functions or Metrics. |
loss_func |
loss function |
opt_func |
The function used to create the optimizer |
lr |
learning rate |
splitter |
It is a function that takes self.model and returns a list of parameter groups (or just one parameter group if there are no different parameter groups). |
path |
The folder where to work |
model_dir |
Path and model_dir are used to save and/or load models. |
wd |
It is the default weight decay used when training the model. |
wd_bn_bias |
It controls if weight decay is applied to BatchNorm layers and bias. |
train_bn |
It controls if BatchNorm layers are trained even when they are supposed to be frozen according to the splitter. |
moms |
The default momentums used in Learner$fit_one_cycle. |
Value
None
Wgan
Description
Create a WGAN from 'data', 'generator' and 'critic'.
Usage
GANLearner_wgan(
dls,
generator,
critic,
switcher = NULL,
clip = 0.01,
switch_eval = FALSE,
gen_first = FALSE,
show_img = TRUE,
cbs = NULL,
metrics = NULL,
opt_func = Adam(),
lr = 0.001,
splitter = trainable_params,
path = NULL,
model_dir = "models",
wd = NULL,
wd_bn_bias = FALSE,
train_bn = TRUE,
moms = list(0.95, 0.85, 0.95)
)
Arguments
dls |
dataloader |
generator |
generator |
critic |
critic |
switcher |
switcher |
clip |
clip value |
switch_eval |
switch evaluation |
gen_first |
generator first |
show_img |
show image or not |
cbs |
callbacks |
metrics |
metrics |
opt_func |
optimization function |
lr |
learning rate |
splitter |
splitter |
path |
path |
model_dir |
model directory |
wd |
weight decay |
wd_bn_bias |
weight decay bn bias |
train_bn |
It controls if BatchNorm layers are trained even when they are supposed to be frozen according to the splitter. |
moms |
momentums |
Value
None
Examples
## Not run:
learn = GANLearner_wgan(dls, generator, critic, opt_func = partial(Adam(), mom=0.))
## End(Not run)
GAN Loss
Description
Wrapper around 'crit_loss_func' and 'gen_loss_func'
Usage
GANLoss(gen_loss_func, crit_loss_func, gan_model)
Arguments
gen_loss_func |
generator loss funcion |
crit_loss_func |
discriminator loss function |
gan_model |
GAN model |
Value
None
GAN Module
Description
Wrapper around a 'generator' and a 'critic' to create a GAN.
Usage
GANModule(generator = NULL, critic = NULL, gen_mode = FALSE)
Arguments
generator |
generator |
critic |
critic |
gen_mode |
generator mode or not |
Value
None
GAN Trainer
Description
Handles GAN Training.
Usage
GANTrainer(
switch_eval = FALSE,
clip = NULL,
beta = 0.98,
gen_first = FALSE,
show_img = TRUE
)
Arguments
switch_eval |
switch evaluation |
clip |
clip value |
beta |
beta parameter |
gen_first |
generator first |
show_img |
show image or not |
Value
None
GatherPredsCallback
Description
'Callback' that saves the predictions and targets, optionally 'with_loss'
Usage
GatherPredsCallback(
with_input = FALSE,
with_loss = FALSE,
save_preds = NULL,
save_targs = NULL,
concat_dim = 0
)
Arguments
with_input |
include inputs or not |
with_loss |
include loss or not |
save_preds |
save predictions |
save_targs |
save targets/actuals |
concat_dim |
concatenate dimensions |
Value
None
GradientAccumulation
Description
Accumulate gradients before updating weights
Usage
GradientAccumulation(n_acc = 32)
Arguments
n_acc |
number of acc |
Value
None
GrandparentSplitter
Description
Split 'items' from the grand parent folder names ('train_name' and 'valid_name').
Usage
GrandparentSplitter(train_name = "train", valid_name = "valid")
Arguments
train_name |
train folder name |
valid_name |
validation folder name |
Value
None
HF_ARCHITECTURES
Description
An enumeration.
Usage
HF_ARCHITECTURES()
Value
None
HF_BaseInput
Description
A HF_BaseInput object is returned from the decodes method of HF_BatchTransform as a mean to customize '@typedispatched' functions like DataLoaders.show_batch and Learner.show_results. It represents the "input_ids" of a huggingface sequence as a tensor with a show method that requires a huggingface tokenizer for proper display.
Usage
HF_BaseInput(...)
Arguments
... |
parameters to pass |
Value
None
HF_BaseModelCallback
Description
HF_BaseModelCallback
Usage
HF_BaseModelCallback(...)
Arguments
... |
parameters to pass |
Value
None
HF_BaseModelWrapper
Description
Same as 'nn.Module', but no need for subclasses to call 'super().__init__'
Usage
HF_BaseModelWrapper(
hf_model,
output_hidden_states = FALSE,
output_attentions = FALSE,
...
)
Arguments
hf_model |
model |
|
output hidden states | |
output_attentions |
output attentions |
... |
additional arguments to pass |
Value
None
HF_BeforeBatchTransform
Description
Handles everything you need to assemble a mini-batch of inputs and targets, as well as decode the dictionary produced as a byproduct of the tokenization process in the 'encodes' method.
Usage
HF_BeforeBatchTransform(
hf_arch,
hf_tokenizer,
max_length = NULL,
padding = TRUE,
truncation = TRUE,
is_split_into_words = FALSE,
n_tok_inps = 1,
...
)
Arguments
hf_arch |
architecture |
hf_tokenizer |
tokenizer |
max_length |
maximum length |
padding |
padding or not |
truncation |
truncation or not |
is_split_into_words |
to split into words |
n_tok_inps |
number tok inputs |
... |
additional arguments |
Value
None
HF_CausalLMBeforeBatchTransform
Description
Handles everything you need to assemble a mini-batch of inputs and targets, as well as decode the dictionary produced
Usage
HF_CausalLMBeforeBatchTransform(
hf_arch,
hf_tokenizer,
max_length = NULL,
padding = TRUE,
truncation = TRUE,
is_split_into_words = FALSE,
n_tok_inps = 1,
ignore_token_id = -100,
...
)
Arguments
hf_arch |
architecture |
hf_tokenizer |
tokenizer |
max_length |
maximum length |
padding |
padding or not |
truncation |
truncation or not |
is_split_into_words |
to split into words |
n_tok_inps |
number tok inputs |
ignore_token_id |
ignore token id |
... |
additional arguments |
Details
as a byproduct of the tokenization process in the 'encodes' method.
Value
None
HF_QABatchTransform
Description
Handles everything you need to assemble a mini-batch of inputs and targets, as well as decode the dictionary produced
Usage
HF_QABatchTransform(
hf_arch,
hf_tokenizer,
max_length = NULL,
padding = TRUE,
truncation = TRUE,
is_split_into_words = FALSE,
n_tok_inps = 1,
hf_input_return_type = HF_QuestionAnswerInput(),
...
)
Arguments
hf_arch |
architecture |
hf_tokenizer |
tokenizer |
max_length |
maximum length |
padding |
padding |
truncation |
truncation |
is_split_into_words |
to split into words or not |
n_tok_inps |
number of tok inputs |
hf_input_return_type |
input return type |
... |
additional arguments |
Details
as a byproduct of the tokenization process in the 'encodes' method.
Value
None
HF_QABeforeBatchTransform
Description
Handles everything you need to assemble a mini-batch of inputs and targets, as well as decode the dictionary produced
Usage
HF_QABeforeBatchTransform(
hf_arch,
hf_tokenizer,
max_length = NULL,
padding = TRUE,
truncation = TRUE,
is_split_into_words = FALSE,
n_tok_inps = 1,
...
)
Arguments
hf_arch |
architecture |
hf_tokenizer |
tokenizer |
max_length |
maximum length |
padding |
padding or not |
truncation |
truncation or not |
is_split_into_words |
into split into words or not |
n_tok_inps |
number of tok inputs |
... |
additional arguments |
Details
as a byproduct of the tokenization process in the 'encodes' method.
Value
None
HF_QstAndAnsModelCallback
Description
HF_QstAndAnsModelCallback
Usage
HF_QstAndAnsModelCallback(...)
Arguments
... |
parameters to pass |
Value
None
HF_QuestionAnswerInput
Description
HF_QuestionAnswerInput
Usage
HF_QuestionAnswerInput(...)
Arguments
... |
parameters to apss |
Value
None
HF_SummarizationBeforeBatchTransform
Description
Handles everything you need to assemble a mini-batch of inputs and targets, as well as decode the dictionary produced as a byproduct of the tokenization process in the 'encodes' method.
Usage
HF_SummarizationBeforeBatchTransform(
hf_arch,
hf_tokenizer,
max_length = NULL,
padding = TRUE,
truncation = TRUE,
is_split_into_words = FALSE,
n_tok_inps = 2,
ignore_token_id = -100,
...
)
Arguments
hf_arch |
architecture |
hf_tokenizer |
tokenizer |
max_length |
maximum length |
padding |
padding or not |
truncation |
truncation or not |
is_split_into_words |
to split into words |
n_tok_inps |
number tok inputs |
ignore_token_id |
ignore token id |
... |
additional arguments |
Value
None
HF_SummarizationInput
Description
HF_SummarizationInput
Usage
HF_SummarizationInput()
Value
None
HF_SummarizationModelCallback
Description
Basic class handling tweaks of the training loop by changing a 'Learner' in various events
Usage
HF_SummarizationModelCallback(
rouge_metrics = c("rouge1", "rouge2", "rougeL"),
ignore_token_id = -100,
...
)
Arguments
rouge_metrics |
rouge metrics |
ignore_token_id |
integer, ignore token id |
... |
additional arguments |
Value
None
HF_TASKS_ALL
Description
An enumeration.
Usage
HF_TASKS_ALL()
Value
None
HF_TASKS_AUTO
Description
An enumeration.
Usage
HF_TASKS_AUTO()
Value
None
HF_Text2TextAfterBatchTransform
Description
Delegates ('__call__','decode','setup') to (<code>encodes</code>,<code>decodes</code>,<code>setups</code>) if 'split_idx' matches
Usage
HF_Text2TextAfterBatchTransform(
hf_tokenizer,
input_return_type = HF_BaseInput()
)
Arguments
hf_tokenizer |
tokenizer |
input_return_type |
input return type |
Value
None
HF_Text2TextBlock
Description
A basic wrapper that links defaults transforms for the data block API
Usage
HF_Text2TextBlock(...)
Arguments
... |
parameters to pass |
Value
None
HF_TextBlock
Description
A basic wrapper that links defaults transforms for the data block API
Usage
HF_TextBlock(...)
Arguments
... |
arguments to pass |
Value
None
HF_TokenCategorize
Description
Reversible transform of a list of category string to 'vocab' id
Usage
HF_TokenCategorize(vocab = NULL, ignore_token = NULL, ignore_token_id = NULL)
Arguments
vocab |
vocabulary |
ignore_token |
ignore token |
ignore_token_id |
ignore token id |
Value
None
HF_TokenCategoryBlock
Description
'TransformBlock' for single-label categorical targets
Usage
HF_TokenCategoryBlock(
vocab = NULL,
ignore_token = NULL,
ignore_token_id = NULL
)
Arguments
vocab |
vocabulary |
ignore_token |
ignore token |
ignore_token_id |
ignore token id |
Value
None
HF_TokenClassBeforeBatchTransform
Description
Handles everything you need to assemble a mini-batch of inputs and targets, as well as decode the dictionary produced
Usage
HF_TokenClassBeforeBatchTransform(
hf_arch,
hf_tokenizer,
ignore_token_id = -100,
max_length = NULL,
padding = TRUE,
truncation = TRUE,
is_split_into_words = TRUE,
n_tok_inps = 1,
...
)
Arguments
hf_arch |
architecture |
hf_tokenizer |
tokenizer |
ignore_token_id |
ignore token id |
max_length |
maximum length |
padding |
padding or not |
truncation |
truncation or not |
is_split_into_words |
to split into_words |
n_tok_inps |
number tok inputs |
... |
additional arguments |
Details
as a byproduct of the tokenization process in the 'encodes' method.
Value
None
HF_TokenClassInput
Description
HF_TokenClassInput
Usage
HF_TokenClassInput()
Value
None
HF_TokenTensorCategory
Description
HF_TokenTensorCategory
Usage
HF_TokenTensorCategory()
Value
None
Load_dataset
Description
Load a dataset
Usage
HF_load_dataset(
path,
name = NULL,
data_dir = NULL,
data_files = NULL,
split = NULL,
cache_dir = NULL,
features = NULL,
download_config = NULL,
download_mode = NULL,
ignore_verifications = FALSE,
save_infos = FALSE,
script_version = NULL,
...
)
Arguments
path |
path |
name |
name |
data_dir |
dataset dir |
data_files |
dataset files |
split |
split |
cache_dir |
cache directory |
features |
features |
download_config |
download configuration |
download_mode |
download mode |
ignore_verifications |
ignore verifications or not |
save_infos |
save information or not |
script_version |
script version |
... |
additional arguments |
Details
This method does the following under the hood: 1. Download and import in the library the dataset loading script from “path“ if it's not already cached inside the library. Processing scripts are small python scripts that define the citation, info and format of the dataset, contain the URL to the original data files and the code to load examples from the original data files. You can find some of the scripts here: https://github.com/huggingface/datasets/datasets and easily upload yours to share them using the CLI “datasets-cli“. 2. Run the dataset loading script which will: * Download the dataset file from the original URL (see the script) if it's not already downloaded and cached. * Process and cache the dataset in typed Arrow tables for caching. Arrow table are arbitrarily long, typed tables which can store nested objects and be mapped to numpy/pandas/python standard types. They can be directly access from drive, loaded in RAM or even streamed over the web. 3. Return a dataset build from the requested splits in “split“ (default: all).
Value
data frame
HammingLoss
Description
Hamming loss for single-label classification problems
Hamming loss for single-label classification problems
Usage
HammingLoss(axis = -1, sample_weight = NULL)
HammingLoss(axis = -1, sample_weight = NULL)
Arguments
axis |
axis |
sample_weight |
sample_weight |
Value
Loss object
None
HammingLossMulti
Description
Hamming loss for multi-label classification problems
Usage
HammingLossMulti(
thresh = 0.5,
sigmoid = TRUE,
labels = NULL,
sample_weight = NULL
)
Arguments
thresh |
threshold |
sigmoid |
sigmoid |
labels |
labels |
sample_weight |
sample_weight |
Value
Loss object
Hook
Description
Create a hook on 'm' with 'hook_func'.
Usage
Hook(
m,
hook_func,
is_forward = TRUE,
detach = TRUE,
cpu = FALSE,
gather = FALSE
)
Arguments
m |
m aprameter |
hook_func |
hook function |
is_forward |
is_forward or not |
detach |
detach or not |
cpu |
cpu or not |
gather |
gather or not |
Details
Hooks are functions you can attach to a particular layer in your model and that will be executed in the forward pass (for forward hooks) or backward pass (for backward hooks).
Value
None
HookCallback
Description
'Callback' that can be used to register hooks on 'modules'
'Callback' that can be used to register hooks on 'modules'
Usage
HookCallback(
modules = NULL,
every = NULL,
remove_end = TRUE,
is_forward = TRUE,
detach = TRUE,
cpu = TRUE
)
HookCallback(
modules = NULL,
every = NULL,
remove_end = TRUE,
is_forward = TRUE,
detach = TRUE,
cpu = TRUE
)
Arguments
modules |
modules |
every |
every |
remove_end |
remove_end or not |
is_forward |
is_forward or not |
detach |
detach or not |
cpu |
cpu or not |
Value
None
None
Hooks
Description
Create several hooks on the modules in 'ms' with 'hook_func'.
Usage
Hooks(ms, hook_func, is_forward = TRUE, detach = TRUE, cpu = FALSE)
Arguments
ms |
ms parameter |
hook_func |
hook function |
is_forward |
is_forward or not |
detach |
detach or not |
cpu |
cpu or not |
Value
None
Hue
Description
Apply change in hue of 'max_hue' to batch of images with probability 'p'.
Usage
Hue(max_hue = 0.1, p = 0.75, draw = NULL, batch = FALSE)
Arguments
max_hue |
maximum hue |
p |
probability |
draw |
draw |
batch |
batch |
Value
None
IDMap
Description
Works like a dictionary that automatically assign values for new keys.
Usage
IDMap(initial_names = NULL)
Arguments
initial_names |
initial_names |
Value
None
Image
Description
Image
Usage
Image(...)
Arguments
... |
parameters to pass |
Value
None
ImageBW_create
Description
Open an 'Image' from path 'fn'
Usage
ImageBW_create(fn)
Arguments
fn |
file name |
Value
None
ImageBlock
Description
A 'TransformBlock' for images of 'cls'
Usage
ImageBlock(...)
Arguments
... |
parameters to pass |
Value
block
ImageDataLoaders from csv
Description
Create from 'path/csv_fname' using 'fn_col' and 'label_col'
Usage
ImageDataLoaders_from_csv(
path,
csv_fname = "labels.csv",
header = "infer",
delimiter = NULL,
valid_pct = 0.2,
seed = NULL,
fn_col = 0,
folder = NULL,
suff = "",
label_col = 1,
label_delim = NULL,
y_block = NULL,
valid_col = NULL,
item_tfms = NULL,
batch_tfms = NULL,
bs = 64,
val_bs = NULL,
size = NULL,
shuffle_train = TRUE,
device = NULL,
...
)
Arguments
path |
The folder where to work |
csv_fname |
csv file name |
header |
header |
delimiter |
delimiter |
valid_pct |
validation percentage |
seed |
random seed |
fn_col |
column name |
folder |
folder name |
suff |
suff |
label_col |
label column |
label_delim |
label delimiter |
y_block |
y_block |
valid_col |
validation column |
item_tfms |
One or several transforms applied to the items before batching them |
batch_tfms |
One or several transforms applied to the batches once they are formed |
bs |
batch size |
val_bs |
The batch size for the validation DataLoader (defaults to bs) |
size |
image size |
shuffle_train |
If we shuffle the training DataLoader or not |
device |
device name |
... |
additional parameters to pass |
Value
None
ImageDataLoaders from dblock
Description
Create a dataloaders from a given 'dblock'
Usage
ImageDataLoaders_from_dblock(
dblock,
source,
path = ".",
bs = 64,
val_bs = NULL,
shuffle_train = TRUE,
device = NULL,
...
)
Arguments
dblock |
dblock |
source |
source folder |
path |
The folder where to work |
bs |
batch size |
val_bs |
The batch size for the validation DataLoader (defaults to bs) |
shuffle_train |
If we shuffle the training DataLoader or not |
device |
device name |
... |
additional parameters to pass |
Value
None
ImageDataLoaders from df
Description
Create from 'df' using 'fn_col' and 'label_col'
Usage
ImageDataLoaders_from_df(
df,
path = ".",
valid_pct = 0.2,
seed = NULL,
fn_col = 0,
folder = NULL,
suff = "",
label_col = 1,
label_delim = NULL,
y_block = NULL,
valid_col = NULL,
item_tfms = NULL,
batch_tfms = NULL,
bs = 64,
val_bs = NULL,
shuffle_train = TRUE,
device = NULL,
...
)
Arguments
df |
data frame |
path |
The folder where to work |
valid_pct |
validation percentage |
seed |
random seed |
fn_col |
column name |
folder |
folder name |
suff |
suff |
label_col |
label column |
label_delim |
label separator |
y_block |
y_block |
valid_col |
validation column |
item_tfms |
One or several transforms applied to the items before batching them |
batch_tfms |
One or several transforms applied to the batches once they are formed |
bs |
batch size |
val_bs |
The batch size for the validation DataLoader (defaults to bs) |
shuffle_train |
shuffle_train |
device |
device |
... |
additional parameters to pass |
Value
None
ImageDataLoaders from folder
Description
Create from imagenet style dataset in 'path' with 'train' and 'valid' subfolders (or provide 'valid_pct')
Usage
ImageDataLoaders_from_folder(
path,
train = "train",
valid = "valid",
valid_pct = NULL,
seed = NULL,
vocab = NULL,
item_tfms = NULL,
batch_tfms = NULL,
bs = 64,
val_bs = NULL,
shuffle_train = TRUE,
device = NULL,
size = NULL,
...
)
Arguments
path |
The folder where to work |
train |
train data |
valid |
validation data |
valid_pct |
validion percentage |
seed |
random seed |
vocab |
vocabulary |
item_tfms |
One or several transforms applied to the items before batching them |
batch_tfms |
One or several transforms applied to the batches once they are formed |
bs |
batch size |
val_bs |
The batch size for the validation DataLoader (defaults to bs) |
shuffle_train |
If we shuffle the training DataLoader or not |
device |
device name |
size |
image size |
... |
additional parameters to pass |
ImageDataLoaders from lists
Description
Create from list of 'fnames' and 'labels' in 'path'
Usage
ImageDataLoaders_from_lists(
path,
fnames,
labels,
valid_pct = 0.2,
seed = NULL,
y_block = NULL,
item_tfms = NULL,
batch_tfms = NULL,
bs = 64,
val_bs = NULL,
shuffle_train = TRUE,
device = NULL,
...
)
Arguments
path |
The folder where to work |
fnames |
file names |
labels |
labels |
valid_pct |
validation percentage |
seed |
random seed |
y_block |
y_block |
item_tfms |
One or several transforms applied to the items before batching them |
batch_tfms |
One or several transforms applied to the batches once they are formed |
bs |
batch size |
val_bs |
The batch size for the validation DataLoader (defaults to bs) |
shuffle_train |
If we shuffle the training DataLoader or not |
device |
device name |
... |
additional parameters to pass |
Value
None
ImageDataLoaders from name regex
Description
Create from the name attrs of 'fnames' in 'path's with re expression 'pat'
Usage
ImageDataLoaders_from_name_re(
path,
fnames,
pat,
bs = 64,
val_bs = NULL,
shuffle_train = TRUE,
device = NULL,
item_tfms = NULL,
batch_tfms = NULL,
...
)
Arguments
path |
The folder where to work |
fnames |
folder names |
pat |
an argument that requires regex |
bs |
The batch size |
val_bs |
The batch size for the validation DataLoader (defaults to bs) |
shuffle_train |
If we shuffle the training DataLoader or not |
device |
device name |
item_tfms |
One or several transforms applied to the items before batching them |
batch_tfms |
One or several transforms applied to the batches once they are formed |
... |
additional parameters to pass |
Value
None
Examples
## Not run:
URLs_PETS()
path = 'oxford-iiit-pet'
dls = ImageDataLoaders_from_name_re(
path, fnames, pat='(.+)_\\d+.jpg$',
item_tfms = RandomResizedCrop(460, min_scale=0.75), bs = 10,
batch_tfms = list(aug_transforms(size = 299, max_warp = 0),
Normalize_from_stats( imagenet_stats() )
),
device = 'cuda'
)
## End(Not run)
ImageDataLoaders from path function
Description
Create from list of 'fnames' in 'path's with 'label_func'
Usage
ImageDataLoaders_from_path_func(
path,
fnames,
label_func,
valid_pct = 0.2,
seed = NULL,
item_tfms = NULL,
batch_tfms = NULL,
bs = 64,
val_bs = NULL,
shuffle_train = TRUE,
device = NULL,
...
)
Arguments
path |
The folder where to work |
fnames |
file names |
label_func |
label function |
valid_pct |
The random percentage of the dataset to set aside for validation (with an optional seed) |
seed |
random seed |
item_tfms |
One or several transforms applied to the items before batching them |
batch_tfms |
One or several transforms applied to the batches once they are formed |
bs |
batch size |
val_bs |
The batch size for the validation DataLoader (defaults to bs) |
shuffle_train |
If we shuffle the training DataLoader or not |
device |
device name |
... |
additional parameters to pass |
Value
None
ImageDataLoaders from path re
Description
Create from list of 'fnames' in 'path's with re expression 'pat'
Usage
ImageDataLoaders_from_path_re(
path,
fnames,
pat,
valid_pct = 0.2,
seed = NULL,
item_tfms = NULL,
batch_tfms = NULL,
bs = 64,
val_bs = NULL,
shuffle_train = TRUE,
device = NULL,
...
)
Arguments
path |
The folder where to work |
fnames |
file names |
pat |
an argument that requires regex |
valid_pct |
The random percentage of the dataset to set aside for validation (with an optional seed) |
seed |
random seed |
item_tfms |
One or several transforms applied to the items before batching them |
batch_tfms |
One or several transforms applied to the batches once they are formed |
bs |
batch size |
val_bs |
The batch size for the validation DataLoader (defaults to bs) |
shuffle_train |
If we shuffle the training DataLoader or not |
device |
device name |
... |
additional parameters to pass |
Value
None
Image_create
Description
Open an 'Image' from path 'fn'
Usage
Image_create(fn)
Arguments
fn |
file name |
Value
None
Image_open
Description
Opens and identifies the given image file.
Usage
Image_open(fp, mode = "r")
Arguments
fp |
fp |
mode |
mode |
Value
None
Resize
Description
Returns a resized copy of this image.
Usage
Image_resize(img, size, resample = 3, box = NULL, reducing_gap = NULL)
Arguments
img |
image |
size |
size |
resample |
resample |
box |
box |
reducing_gap |
reducing_gap |
Value
None
InceptionModule
Description
The inception Module from ‘ni' inputs to len(’kss')*'nb_filters'+'bottleneck_size'
Usage
InceptionModule(
ni,
nb_filters = 32,
kss = c(39, 19, 9),
bottleneck_size = 32,
stride = 1
)
Arguments
ni |
number of input channels |
nb_filters |
the number of filters |
kss |
kernel size |
bottleneck_size |
bottleneck size |
stride |
stride |
Value
module
Index Splitter
Description
Split 'items' so that 'val_idx' are in the validation set and the others in the training set
Usage
IndexSplitter(valid_idx)
Arguments
valid_idx |
The indices to use for the validation set (defaults to a random split otherwise) |
Value
None
InstanceNorm
Description
InstanceNorm layer with 'nf' features and 'ndim' initialized depending on 'norm_type'.
Usage
InstanceNorm(
nf,
ndim = 2,
norm_type = 5,
affine = TRUE,
eps = 1e-05,
momentum = 0.1,
track_running_stats = FALSE
)
Arguments
nf |
input shape |
ndim |
dimension number |
norm_type |
normalization type |
affine |
affine |
eps |
epsilon |
momentum |
momentum |
track_running_stats |
track running statistics |
Value
None
IntToFloatTensor
Description
Transform image to float tensor, optionally dividing by 255 (e.g. for images).
Usage
IntToFloatTensor(div = 255, div_mask = 1)
Arguments
div |
divide value |
div_mask |
divide mask |
Value
None
Invisible Tensor
Description
Invisible Tensor
Usage
InvisibleTensor(x)
Arguments
x |
tensor |
Value
None
Jaccard
Description
Jaccard score for single-label classification problems
Usage
Jaccard(
axis = -1,
labels = NULL,
pos_label = 1,
average = "binary",
sample_weight = NULL
)
Arguments
axis |
axis |
labels |
labels |
pos_label |
pos_label |
average |
average |
sample_weight |
sample_weight |
Value
None
JaccardCoeff
Description
Implementation of the Jaccard coefficient that is lighter in RAM
Usage
JaccardCoeff(axis = 1)
Arguments
axis |
axis |
Value
None
JaccardMulti
Description
Jaccard score for multi-label classification problems
Usage
JaccardMulti(
thresh = 0.5,
sigmoid = TRUE,
labels = NULL,
pos_label = 1,
average = "macro",
sample_weight = NULL
)
Arguments
thresh |
thresh |
sigmoid |
sigmoid |
labels |
labels |
pos_label |
pos_label |
average |
average |
sample_weight |
sample_weight |
Value
None
L
Description
Behaves like a list of 'items' but can also index with list of indices or masks
Usage
L(...)
Arguments
... |
arguments to pass |
L1LossFlat
Description
Flattens input and output, same as nn$L1LossFlat
Usage
L1LossFlat(...)
Arguments
... |
parameters to pass |
Value
Loss object
LMDataLoader
Description
A 'DataLoader' suitable for language modeling
Usage
LMDataLoader(
dataset,
lens = NULL,
cache = 2,
bs = 64,
seq_len = 72,
num_workers = 0,
shuffle = FALSE,
verbose = FALSE,
do_setup = TRUE,
pin_memory = FALSE,
timeout = 0L,
batch_size = NULL,
drop_last = FALSE,
indexed = NULL,
n = NULL,
device = NULL
)
Arguments
dataset |
dataset |
lens |
lens |
cache |
cache |
bs |
bs |
seq_len |
seq_len |
num_workers |
num_workers |
shuffle |
shuffle |
verbose |
verbose |
do_setup |
do_setup |
pin_memory |
pin_memory |
timeout |
timeout |
batch_size |
batch_size |
drop_last |
drop_last |
indexed |
indexed |
n |
n |
device |
device |
Value
text loader
LMLearner
Description
Add functionality to 'TextLearner' when dealingwith a language model
Add functionality to 'TextLearner' when dealing with a language model
Usage
LMLearner(
dls,
model,
alpha = 2,
beta = 1,
moms = list(0.8, 0.7, 0.8),
loss_func = NULL,
opt_func = Adam(),
lr = 0.001,
splitter = trainable_params(),
cbs = NULL,
metrics = NULL,
path = NULL,
model_dir = "models",
wd = NULL,
wd_bn_bias = FALSE,
train_bn = TRUE
)
LMLearner(
dls,
model,
alpha = 2,
beta = 1,
moms = list(0.8, 0.7, 0.8),
loss_func = NULL,
opt_func = Adam(),
lr = 0.001,
splitter = trainable_params(),
cbs = NULL,
metrics = NULL,
path = NULL,
model_dir = "models",
wd = NULL,
wd_bn_bias = FALSE,
train_bn = TRUE
)
Arguments
dls |
dls |
model |
model |
alpha |
alpha |
beta |
beta |
moms |
moms |
loss_func |
loss_func |
opt_func |
opt_func |
lr |
lr |
splitter |
splitter |
cbs |
cbs |
metrics |
metrics |
path |
path |
model_dir |
model_dir |
wd |
wd |
wd_bn_bias |
wd_bn_bias |
train_bn |
train_bn |
Value
text loader
None
LMLearner_predict
Description
Return 'text' and the 'n_words' that come after
Usage
LMLearner_predict(
text,
n_words = 1,
no_unk = TRUE,
temperature = 1,
min_p = NULL,
no_bar = FALSE,
decoder = decode_spec_tokens(),
only_last_word = FALSE
)
Arguments
text |
text |
n_words |
n_words |
no_unk |
no_unk |
temperature |
temperature |
min_p |
min_p |
no_bar |
no_bar |
decoder |
decoder |
only_last_word |
only_last_word |
Value
None
LabelSmoothingCrossEntropy
Description
Same as 'nn$Module', but no need for subclasses to call 'super()$__init__'
Usage
LabelSmoothingCrossEntropy(eps = 0.1, reduction = "mean")
Arguments
eps |
epsilon |
reduction |
reduction, defaults to mean |
Value
Loss object
LabelSmoothingCrossEntropyFlat
Description
Same as 'nn$Module', but no need for subclasses to call 'super().__init__'
Usage
LabelSmoothingCrossEntropyFlat(...)
Arguments
... |
parameters to pass |
Value
Loss object
LabeledBBox
Description
Basic type for a list of bounding boxes in an image
Usage
LabeledBBox(...)
Arguments
... |
parameters to pass |
Value
None
Lamb
Description
Lamb
Usage
Lamb(...)
Arguments
... |
parameters to pass |
Value
None
Lambda
Description
An easy way to create a pytorch layer for a simple 'func'
Usage
Lambda(func)
Arguments
func |
function |
Value
None
Larc
Description
Larc
Usage
Larc(...)
Arguments
... |
parameters to pass |
Value
None
Learner
Description
Learner
Usage
Learner(...)
Arguments
... |
parameters to pass |
Value
None
Examples
## Not run:
model = LitModel()
data = Data_Loaders(model$train_dataloader(), model$val_dataloader())$cuda()
learn = Learner(data, model, loss_func = F$cross_entropy, opt_func = Adam,
metrics = accuracy)
## End(Not run)
LightingTfm
Description
Apply 'fs' to the logits
Usage
LightingTfm(fs, ...)
Arguments
fs |
fs |
... |
parameters to pass |
Value
None
LinBnDrop
Description
Module grouping 'BatchNorm1d', 'Dropout' and 'Linear' layers
Usage
LinBnDrop(n_in, n_out, bn = TRUE, p = 0, act = NULL, lin_first = FALSE)
Arguments
n_in |
input shape |
n_out |
output shape |
bn |
bn |
p |
probability |
act |
activation |
lin_first |
linear first |
Value
None
LinearDecoder
Description
To go on top of a RNNCore module and create a Language Model.
Usage
LinearDecoder(n_out, n_hid, output_p = 0.1, tie_encoder = NULL, bias = TRUE)
Arguments
n_out |
n_out |
n_hid |
n_hid |
output_p |
output_p |
tie_encoder |
tie_encoder |
bias |
bias |
Value
None
Lit Model
Description
Lit Model
Usage
LitModel()
Value
model
Lookahead
Description
Lookahead
Usage
Lookahead(...)
Arguments
... |
parameters to pass |
Value
None
LossMetric
Description
Create a metric from 'loss_func.attr' named 'nm'
Usage
LossMetric(attr, nm = NULL)
Arguments
attr |
attr |
nm |
nm |
Value
None
MCDropoutCallback
Description
Turns on dropout during inference, allowing you to call Learner$get_preds multiple times to approximate your model uncertainty using Monte Carlo Dropout. https://arxiv.org/pdf/1506.02142.pdf
Usage
MCDropoutCallback(...)
Arguments
... |
arguments to pass |
Value
None
MSELossFlat
Description
Flattens input and output, same as nn$MSELoss
Usage
MSELossFlat(...)
Arguments
... |
parameters to pass |
Value
Loss object
MaskBlock
Description
A 'TransformBlock' for segmentation masks, potentially with 'codes'
Usage
MaskBlock(codes = NULL)
Arguments
codes |
codes |
Value
block
Mask Freq
Description
Google SpecAugment frequency masking from https://arxiv.org/abs/1904.08779.
Usage
MaskFreq(num_masks = 1, size = 20, start = NULL, val = NULL)
Arguments
num_masks |
number of masks |
size |
size |
start |
starting point |
val |
value |
Value
None
MaskTime
Description
Google SpecAugment time masking from https://arxiv.org/abs/1904.08779.
Usage
MaskTime(num_masks = 1, size = 20, start = NULL, val = NULL)
Arguments
num_masks |
number of masks |
size |
size |
start |
starting point |
val |
value |
Value
None
Mask_create
Description
Delegates ('__call__','decode','setup') to ('encodes','decodes','setups') if 'split_idx' matches
Usage
Mask_create(enc = NULL, dec = NULL, split_idx = NULL, order = NULL)
Arguments
enc |
encoder |
dec |
decoder |
split_idx |
split by index |
order |
order |
Value
None
MatthewsCorrCoef
Description
Matthews correlation coefficient for single-label classification problems
Usage
MatthewsCorrCoef(...)
Arguments
... |
parameters to pass |
Value
None
MatthewsCorrCoefMulti
Description
Matthews correlation coefficient for multi-label classification problems
Usage
MatthewsCorrCoefMulti(thresh = 0.5, sigmoid = TRUE, sample_weight = NULL)
Arguments
thresh |
thresh |
sigmoid |
sigmoid |
sample_weight |
sample_weight |
Value
None
MaxPool
Description
nn.MaxPool layer for 'ndim'
Usage
MaxPool(ks = 2, stride = NULL, padding = 0, ndim = 2, ceil_mode = FALSE)
Arguments
ks |
kernel size |
stride |
the stride of the window. Default value is kernel_size |
padding |
implicit zero padding to be added on both sides |
ndim |
dimension number |
ceil_mode |
when True, will use ceil instead of floor to compute the output shape |
Value
None
MergeLayer
Description
Merge a shortcut with the result of the module by adding them or concatenating them if 'dense=TRUE'.
Usage
MergeLayer(dense = FALSE)
Arguments
dense |
dense |
Value
None
MishJitAutoFn
Description
Records operation history and defines formulas for differentiating ops.
Usage
MishJitAutoFn(...)
Arguments
... |
parameters to pass |
Value
None
Class Mish
Description
Class Mish
Usage
Mish_(...)
Arguments
... |
parameters to pass |
Value
None
MixHandler
Description
A handler class for implementing 'MixUp' style scheduling
Usage
MixHandler(alpha = 0.5)
Arguments
alpha |
alpha |
Value
None
MixUp
Description
Implementation of https://arxiv.org/abs/1710.09412
Usage
MixUp(alpha = 0.4)
Arguments
alpha |
alpha |
Value
None
ModelResetter
Description
Callback that resets the model at each validation/training step
Usage
ModelResetter(...)
Arguments
... |
arguments to pass |
Value
None
Module module
Description
Module module
Usage
Module()
Value
None
NN module
Description
NN module
Usage
Module_test()
Value
None
MultiCategorize
Description
Reversible transform of multi-category strings to 'vocab' id
Usage
MultiCategorize(vocab = NULL, add_na = FALSE)
Arguments
vocab |
vocabulary |
add_na |
add NA |
Value
None
MultiCategoryBlock
Description
'TransformBlock' for multi-label categorical targets
Usage
MultiCategoryBlock(encoded = FALSE, vocab = NULL, add_na = FALSE)
Arguments
encoded |
encoded or not |
vocab |
vocabulary |
add_na |
add NA |
Value
Block object
MultiTargetLoss
Description
Provides the ability to apply different loss functions to multi-modal targets/predictions
Usage
MultiTargetLoss(...)
Arguments
... |
additional arguments |
Value
None
Net
Description
Net model from Migrating_Pytorch
Usage
Net()
Value
model
Examples
## Not run:
Net()
## End(Not run)
NoiseColor module
Description
NoiseColor module
Usage
NoiseColor()
Value
None
NoneReduce
Description
A context manager to evaluate 'loss_func' with none reduce.
Usage
NoneReduce(loss_func)
Arguments
loss_func |
loss function |
Value
None
Normalize
Description
Normalize the continuous variables.
Usage
Normalize(cat_names, cont_names)
Arguments
cat_names |
cat_names |
cont_names |
cont_names |
Value
None
NormalizeTS
Description
Normalize the x variables.
Usage
NormalizeTS(enc = NULL, dec = NULL, split_idx = NULL, order = NULL)
Arguments
enc |
encoder |
dec |
decoder |
split_idx |
split by index |
order |
order |
Value
None
Normalize from stats
Description
Normalize from stats
Usage
Normalize_from_stats(mean, std, dim = 1, ndim = 4, cuda = TRUE)
Arguments
mean |
mean |
std |
standard deviation |
dim |
dimension |
ndim |
number of dimensions |
cuda |
cuda or not |
Value
list
Numericalize
Description
Reversible transform of tokenized texts to numericalized ids
Usage
Numericalize(
vocab = NULL,
min_freq = 3,
max_vocab = 60000,
special_toks = NULL,
pad_tok = NULL
)
Arguments
vocab |
vocab |
min_freq |
min_freq |
max_vocab |
max_vocab |
special_toks |
special_toks |
pad_tok |
pad_tok |
Value
None
OldRandomCrop
Description
Randomly crop an image to 'size'
Usage
OldRandomCrop(size, pad_mode = "zeros", ...)
Arguments
size |
size |
pad_mode |
padding mode |
... |
additional arguments |
Value
None
OpenAudio
Description
Transform that creates AudioTensors from a list of files.
Usage
OpenAudio(items)
Arguments
items |
vector, items |
Value
None
OptimWrapper
Description
OptimWrapper
Usage
OptimWrapper(...)
Arguments
... |
parameters to pass |
Value
None
Optimizer
Description
Optimizer
Usage
Optimizer(...)
Arguments
... |
parameters to pass |
Value
None
ParamScheduler
Description
Schedule hyper-parameters according to 'scheds'
Usage
ParamScheduler(scheds)
Arguments
scheds |
scheds |
Value
None
PartialDL
Description
Select randomly partial quantity of data at each epoch
Usage
PartialDL(
dataset = NULL,
bs = NULL,
partial_n = NULL,
shuffle = FALSE,
num_workers = NULL,
verbose = FALSE,
do_setup = TRUE,
pin_memory = FALSE,
timeout = 0,
batch_size = NULL,
drop_last = FALSE,
indexed = NULL,
n = NULL,
device = NULL,
persistent_workers = FALSE
)
Arguments
dataset |
dataset |
bs |
bs |
partial_n |
partial_n |
shuffle |
shuffle |
num_workers |
num_workers |
verbose |
verbose |
do_setup |
do_setup |
pin_memory |
pin_memory |
timeout |
timeout |
batch_size |
batch_size |
drop_last |
drop_last |
indexed |
indexed |
n |
n |
device |
device |
persistent_workers |
persistent_workers |
Value
None
Partial Lambda
Description
Layer that applies 'partial(func, ...)'
Usage
PartialLambda(func)
Arguments
func |
function |
Value
None
PearsonCorrCoef
Description
Pearson correlation coefficient for regression problem
Usage
PearsonCorrCoef(
dim_argmax = NULL,
activation = "no",
thresh = NULL,
to_np = FALSE,
invert_arg = FALSE,
flatten = TRUE
)
Arguments
dim_argmax |
dim_argmax |
activation |
activation |
thresh |
thresh |
to_np |
to_np |
invert_arg |
invert_arg |
flatten |
flatten |
Value
None
Perplexity
Description
Perplexity
Usage
Perplexity(...)
Arguments
... |
parameters to pass |
Value
None
Pipeline
Description
A pipeline of composed (for encode/decode) transforms, setup with types
Usage
Pipeline(funcs = NULL, split_idx = NULL)
Arguments
funcs |
functions |
split_idx |
split by index |
Value
None
PixelShuffle_ICNR
Description
Upsample by 'scale' from 'ni' filters to 'nf' (default 'ni'), using 'nn.PixelShuffle'.
Usage
PixelShuffle_ICNR(
ni,
nf = NULL,
scale = 2,
blur = FALSE,
norm_type = 3,
act_cls = nn()$ReLU
)
Arguments
ni |
input shape |
nf |
number of features / outputs |
scale |
scale |
blur |
blur |
norm_type |
normalziation type |
act_cls |
activation |
Value
None
PointBlock
Description
A 'TransformBlock' for points in an image
Usage
PointBlock()
Value
None
PointScaler
Description
Scale a tensor representing points
Usage
PointScaler(do_scale = TRUE, y_first = FALSE)
Arguments
do_scale |
do scale |
y_first |
y first |
Value
None
PoolFlatten
Description
Combine 'nn.AdaptiveAvgPool2d' and 'Flatten'.
Usage
PoolFlatten(pool_type = "Avg")
Arguments
pool_type |
pooling type |
Value
None
PooledSelfAttention2d
Description
Pooled self attention layer for 2d.
Usage
PooledSelfAttention2d(n_channels)
Arguments
n_channels |
number of channels |
Value
None
PoolingLinearClassifier
Description
Create a linear classifier with pooling
Usage
PoolingLinearClassifier(dims, ps, bptt, y_range = NULL)
Arguments
dims |
dims |
ps |
ps |
bptt |
bptt |
y_range |
y_range |
Value
None
Precision
Description
Precision for single-label classification problems
Usage
Precision(
axis = -1,
labels = NULL,
pos_label = 1,
average = "binary",
sample_weight = NULL
)
Arguments
axis |
axis |
labels |
labels |
pos_label |
pos_label |
average |
average |
sample_weight |
sample_weight |
Value
None
PrecisionMulti
Description
Precision for multi-label classification problems
Usage
PrecisionMulti(
thresh = 0.5,
sigmoid = TRUE,
labels = NULL,
pos_label = 1,
average = "macro",
sample_weight = NULL
)
Arguments
thresh |
thresh |
sigmoid |
sigmoid |
labels |
labels |
pos_label |
pos_label |
average |
average |
sample_weight |
sample_weight |
Value
None
Preprocess Audio
Description
Creates an audio tensor and run the basic preprocessing transforms on it.
Usage
PreprocessAudio(sample_rate = 16000, force_mono = TRUE, crop_signal_to = NULL)
Arguments
sample_rate |
sample rate |
force_mono |
force mono or not |
crop_signal_to |
int, crop signal |
Details
Used while preprocessing the audios, this is not a 'Transform'.
Value
None
QHAdam
Description
QHAdam
Usage
QHAdam(...)
Arguments
... |
parameters to pass |
Value
None
QRNN
Description
Apply a multiple layer Quasi-Recurrent Neural Network (QRNN) to an input sequence.
Usage
QRNN(
input_size,
hidden_size,
n_layers = 1,
batch_first = TRUE,
dropout = 0,
bidirectional = FALSE,
save_prev_x = FALSE,
zoneout = 0,
window = NULL,
output_gate = TRUE
)
Arguments
input_size |
input_size |
|
hidden_size | |
n_layers |
n_layers |
batch_first |
batch_first |
dropout |
dropout |
bidirectional |
bidirectional |
save_prev_x |
save_prev_x |
zoneout |
zoneout |
window |
window |
output_gate |
output_gate |
Value
None
QRNNLayer
Description
Apply a single layer Quasi-Recurrent Neural Network (QRNN) to an input sequence.
Usage
QRNNLayer(
input_size,
hidden_size = NULL,
save_prev_x = FALSE,
zoneout = 0,
window = 1,
output_gate = TRUE,
batch_first = TRUE,
backward = FALSE
)
Arguments
input_size |
input_size |
|
hidden_size | |
save_prev_x |
save_prev_x |
zoneout |
zoneout |
window |
window |
output_gate |
output_gate |
batch_first |
batch_first |
backward |
backward |
Value
None
R2Score
Description
R2 score between predictions and targets
Usage
R2Score(sample_weight = NULL)
Arguments
sample_weight |
sample_weight |
Value
None
RAdam
Description
RAdam
Usage
RAdam(...)
Arguments
... |
parameters to pass |
Value
None
RMSProp
Description
RMSProp
Usage
RMSProp(...)
Arguments
... |
parameters to pass |
Value
None
RNNDropout
Description
Dropout with probability 'p' that is consistent on the seq_len dimension.
Usage
RNNDropout(p = 0.5)
Arguments
p |
p |
Value
None
RNNRegularizer
Description
'Callback' that adds AR and TAR regularization in RNN training
Usage
RNNRegularizer(alpha = 0, beta = 0)
Arguments
alpha |
alpha |
beta |
beta |
Value
None
RandPair
Description
a random image from domain B, resulting in a random pair of images from domain A and B.
Usage
RandPair(itemsB)
Arguments
itemsB |
a random image from domain B |
Value
None
RandTransform
Description
A transform that before_call its state at each '__call__'
Usage
RandTransform(p = 1, nm = NULL, before_call = NULL, ...)
Arguments
p |
probability |
nm |
nm |
before_call |
before call |
... |
additional arguments to pass |
Value
None
RandomCrop
Description
Randomly crop an image to 'size'
Usage
RandomCrop(size, ...)
Arguments
size |
size |
... |
additional arguments |
Value
None
RandomErasing
Description
Randomly selects a rectangle region in an image and randomizes its pixels.
Usage
RandomErasing(p = 0.5, sl = 0, sh = 0.3, min_aspect = 0.3, max_count = 1)
Arguments
p |
probability |
sl |
sl |
sh |
sh |
min_aspect |
minimum aspect |
max_count |
maximum count |
Value
None
RandomResizedCrop
Description
Picks a random scaled crop of an image and resize it to 'size'
Usage
RandomResizedCrop(
size,
min_scale = 0.08,
ratio = list(0.75, 1.33333333333333),
resamples = list(2, 0),
val_xtra = 0.14
)
Arguments
size |
size |
min_scale |
minimum scale |
ratio |
ratio |
resamples |
resamples |
val_xtra |
validation xtra |
Value
None
RandomResizedCropGPU
Description
Picks a random scaled crop of an image and resize it to 'size'
Usage
RandomResizedCropGPU(
size,
min_scale = 0.08,
ratio = list(0.75, 1.33333333333333),
mode = "bilinear",
valid_scale = 1
)
Arguments
size |
size |
min_scale |
minimum scale |
ratio |
ratio |
mode |
mode |
valid_scale |
validation scale |
Value
None
RandomSplitter
Description
Create function that splits 'items' between train/val with 'valid_pct' randomly.
Usage
RandomSplitter(valid_pct = 0.2, seed = NULL)
Arguments
valid_pct |
validation percenatge split |
seed |
random seed |
Value
None
RatioResize
Description
Resizes the biggest dimension of an image to 'max_sz' maintaining the aspect ratio
Usage
RatioResize(max_sz, resamples = list(2, 0), ...)
Arguments
max_sz |
maximum sz |
resamples |
resamples |
... |
additional arguments |
Value
None
ReadTSBatch
Description
A transform that always take lists as items
Usage
ReadTSBatch(to)
Arguments
to |
output from TSDataTable function |
Value
None
Recall
Description
Recall for single-label classification problems
Usage
Recall(
axis = -1,
labels = NULL,
pos_label = 1,
average = "binary",
sample_weight = NULL
)
Arguments
axis |
axis |
labels |
labels |
pos_label |
pos_label |
average |
average |
sample_weight |
sample_weight |
Value
None
RecallMulti
Description
Recall for multi-label classification problems
Usage
RecallMulti(
thresh = 0.5,
sigmoid = TRUE,
labels = NULL,
pos_label = 1,
average = "macro",
sample_weight = NULL
)
Arguments
thresh |
thresh |
sigmoid |
sigmoid |
labels |
labels |
pos_label |
pos_label |
average |
average |
sample_weight |
sample_weight |
Value
None
ReduceLROnPlateau
Description
ReduceLROnPlateau
Usage
ReduceLROnPlateau(...)
Arguments
... |
parameters to pass |
Value
None
Examples
## Not run:
URLs_MNIST_SAMPLE()
# transformations
tfms = aug_transforms(do_flip = FALSE)
path = 'mnist_sample'
bs = 20
#load into memory
data = ImageDataLoaders_from_folder(path, batch_tfms = tfms, size = 26, bs = bs)
learn = cnn_learner(data, resnet18(), metrics = accuracy, path = getwd())
learn %>% fit_one_cycle(10, 1e-2, cbs = ReduceLROnPlateau(monitor='valid_loss', patience = 1))
## End(Not run)
RegressionBlock
Description
'TransformBlock' for float targets
Usage
RegressionBlock(n_out = NULL)
Arguments
n_out |
number of out features |
Value
Block object
Remove Silence
Description
Split signal at points of silence greater than 2*pad_ms
Usage
RemoveSilence(
remove_type = RemoveType()$Trim$value,
threshold = 20,
pad_ms = 20
)
Arguments
remove_type |
remove type from RemoveType module |
threshold |
threshold point |
pad_ms |
pad milliseconds |
Value
None
RemoveType module
Description
RemoveType module
Usage
RemoveType()
Value
None
ResBlock
Description
Resnet block from 'ni' to 'nh' with 'stride'
Usage
ResBlock(
expansion,
ni,
nf,
stride = 1,
groups = 1,
reduction = NULL,
nh1 = NULL,
nh2 = NULL,
dw = FALSE,
g2 = 1,
sa = FALSE,
sym = FALSE,
norm_type = 1,
act_cls = nn$ReLU,
ndim = 2,
ks = 3,
pool = AvgPool(),
pool_first = TRUE,
padding = NULL,
bias = NULL,
bn_1st = TRUE,
transpose = FALSE,
init = "auto",
xtra = NULL,
bias_std = 0.01,
dilation = 1,
padding_mode = "zeros"
)
Arguments
expansion |
decoder |
ni |
number of linear inputs |
nf |
number of features |
stride |
stride number |
groups |
groups number |
reduction |
reduction |
nh1 |
out channels 1 |
nh2 |
out channels 2 |
dw |
dw paramer |
g2 |
g2 block |
sa |
sa parameter |
sym |
symmetric |
norm_type |
normalization type |
act_cls |
activation |
ndim |
dimension number |
ks |
kernel size |
pool |
pooling type, Average, Max |
pool_first |
pooling first |
padding |
padding |
bias |
bias |
bn_1st |
batch normalization 1st |
transpose |
transpose |
init |
initializer |
xtra |
xtra |
bias_std |
bias standard deviation |
dilation |
dilation number |
padding_mode |
padding mode |
Value
Block object
ResNet
Description
Base class for all neural network modules.
Usage
ResNet(
block,
layers,
num_classes = 1000,
zero_init_residual = FALSE,
groups = 1,
width_per_group = 64,
replace_stride_with_dilation = NULL,
norm_layer = NULL
)
Arguments
block |
the blocks that need to passed to ResNet |
layers |
the layers to pass to ResNet |
num_classes |
the number of classes |
zero_init_residual |
logical, initializer |
groups |
the groups |
width_per_group |
the width per group |
replace_stride_with_dilation |
logical, replace stride with dilation |
norm_layer |
norm_layer |
Resample
Description
Resample using faster polyphase technique and avoiding FFT computation
Usage
Resample(sr_new)
Arguments
sr_new |
input |
Value
None
Resize
Description
A transform that before_call its state at each '__call__'
Usage
Resize(size, method = "crop", pad_mode = "reflection", resamples = list(2, 0))
Arguments
size |
size of image |
method |
method |
pad_mode |
reflection, zeros, border as string parameter |
resamples |
list of integers |
Value
None
ResizeBatch
Description
Reshape x to size, keeping batch dim the same size
Usage
ResizeBatch(...)
Arguments
... |
parameters to pass |
Value
None
Resize Signal
Description
Crops signal to be length specified in ms by duration, padding if needed
Usage
ResizeSignal(duration, pad_mode = AudioPadType()$Zeros)
Arguments
duration |
int, duration |
pad_mode |
padding mode |
Value
None
ResnetBlock
Description
nn()$Module for the ResNet Block
Usage
ResnetBlock(
dim,
pad_mode = "reflection",
norm_layer = NULL,
dropout = 0,
bias = TRUE
)
Arguments
dim |
dimension |
pad_mode |
padding mode |
norm_layer |
normalization layer |
dropout |
dropout rate |
bias |
bias or not |
Value
None
RetinaNet
Description
Implements RetinaNet from https://arxiv.org/abs/1708.02002
Usage
RetinaNet(...)
Arguments
... |
arguments to pass |
Value
model
Examples
## Not run:
encoder = create_body(resnet34(), pretrained = TRUE)
arch = RetinaNet(encoder, get_c(dls), final_bias=-4)
## End(Not run)
RetinaNetFocalLoss
Description
Base class for all neural network modules.
Usage
RetinaNetFocalLoss(...)
Arguments
... |
parameters to pass |
Details
Your models should also subclass this class. Modules can also contain other Modules, allowing to nest them in a tree structure. You can assign the submodules as regular attributes:: import torch.nn as nn import torch.nn.functional as F class Model(nn.Module): def __init__(self): super(Model, self).__init__() self.conv1 = nn.Conv2d(1, 20, 5) self.conv2 = nn.Conv2d(20, 20, 5) def forward(self, x): x = F.relu(self.conv1(x)) return F.relu(self.conv2(x)) Submodules assigned in this way will be registered, and will have their parameters converted too when you call :meth:'to', etc.
Value
None
RocAuc
Description
Area Under the Receiver Operating Characteristic Curve for single-label multiclass classification problems
Usage
RocAuc(
axis = -1,
average = "macro",
sample_weight = NULL,
max_fpr = NULL,
multi_class = "ovr"
)
Arguments
axis |
axis |
average |
average |
sample_weight |
sample_weight |
max_fpr |
max_fpr |
multi_class |
multi_class |
Value
None
RocAucBinary
Description
Area Under the Receiver Operating Characteristic Curve for single-label binary classification problems
Usage
RocAucBinary(
axis = -1,
average = "macro",
sample_weight = NULL,
max_fpr = NULL,
multi_class = "raise"
)
Arguments
axis |
axis |
average |
average |
sample_weight |
sample_weight |
max_fpr |
max_fpr |
multi_class |
multi_class |
Value
None
Examples
## Not run:
model = dls %>% tabular_learner(layers=c(200,100,100,200),
config = tabular_config(embed_p = 0.3, use_bn = FALSE),
metrics = list(accuracy, RocAucBinary(),
Precision(), Recall(),
F1Score()))
## End(Not run)
RocAucMulti
Description
Area Under the Receiver Operating Characteristic Curve for multi-label binary classification problems
Usage
RocAucMulti(
sigmoid = TRUE,
average = "macro",
sample_weight = NULL,
max_fpr = NULL
)
Arguments
sigmoid |
sigmoid |
average |
average |
sample_weight |
sample_weight |
max_fpr |
max_fpr |
Value
None
Rotate
Description
Apply a random rotation of at most 'max_deg' with probability 'p' to a batch of images
Usage
Rotate(
max_deg = 10,
p = 0.5,
draw = NULL,
size = NULL,
mode = "bilinear",
pad_mode = "reflection",
align_corners = TRUE,
batch = FALSE
)
Arguments
max_deg |
maximum degrees |
p |
probability |
draw |
draw |
size |
size of image |
mode |
mode |
pad_mode |
reflection, zeros, border as string parameter |
align_corners |
align corners or not |
batch |
batch or not |
Value
None
SEBlock
Description
SEBlock
Usage
SEBlock(expansion, ni, nf, groups = 1, reduction = 16, stride = 1)
Arguments
expansion |
decoder |
ni |
number of inputs |
nf |
number of features |
groups |
number of groups |
reduction |
number of reduction |
stride |
number of strides |
Value
Block object
SEModule
Description
SEModule
Usage
SEModule(ch, reduction, act_cls = nn()$ReLU)
Arguments
ch |
ch |
reduction |
reduction |
act_cls |
activation |
Value
None
SEResNeXtBlock
Description
SEResNeXtBlock
Usage
SEResNeXtBlock(
expansion,
ni,
nf,
groups = 32,
reduction = 16,
stride = 1,
base_width = 4
)
Arguments
expansion |
decoder |
ni |
number of linear inputs |
nf |
number of features |
groups |
groups number |
reduction |
reduction number |
stride |
stride number |
base_width |
int, base width |
Value
Block object
SGD
Description
SGD
Usage
SGD(...)
Arguments
... |
parameters to pass |
Value
None
SGRoll
Description
Shifts spectrogram along x-axis wrapping around to other side
Usage
SGRoll(max_shift_pct = 0.5, direction = 0)
Arguments
max_shift_pct |
maximum shift percentage |
direction |
direction |
Value
None
Saturation
Description
Apply change in saturation of 'max_lighting' to batch of images with probability 'p'.
Usage
Saturation(max_lighting = 0.2, p = 0.75, draw = NULL, batch = FALSE)
Arguments
max_lighting |
maximum lighting |
p |
probability |
draw |
draw |
batch |
batch |
Value
None
SaveModelCallback
Description
SaveModelCallback
Usage
SaveModelCallback(...)
Arguments
... |
parameters to pass |
Value
None
SchedCos
Description
Cosine schedule function from 'start' to 'end'
Usage
SchedCos(start, end)
Arguments
start |
start |
end |
end |
Value
None
SchedExp
Description
Exponential schedule function from 'start' to 'end'
Usage
SchedExp(start, end)
Arguments
start |
start |
end |
end |
Value
None
SchedLin
Description
Linear schedule function from 'start' to 'end'
Usage
SchedLin(start, end)
Arguments
start |
start |
end |
end |
Value
None
SchedNo
Description
Constant schedule function with 'start' value
Usage
SchedNo(start, end)
Arguments
start |
start |
end |
end |
Value
None
SchedPoly
Description
Polynomial schedule (of 'power') function from 'start' to 'end'
Usage
SchedPoly(start, end, power)
Arguments
start |
start |
end |
end |
power |
power |
Value
None
SegmentationDataLoaders_from_label_func
Description
Create from list of 'fnames' in 'path's with 'label_func'.
Usage
SegmentationDataLoaders_from_label_func(
path,
fnames,
label_func,
valid_pct = 0.2,
seed = NULL,
codes = NULL,
item_tfms = NULL,
batch_tfms = NULL,
bs = 64,
val_bs = NULL,
shuffle_train = TRUE,
device = NULL
)
Arguments
path |
path |
fnames |
file names |
label_func |
label function |
valid_pct |
validation percentage |
seed |
seed |
codes |
codes |
item_tfms |
item transformations |
batch_tfms |
batch transformations |
bs |
batch size |
val_bs |
validation batch size |
shuffle_train |
shuffle train |
device |
device name |
Value
None
SelfAttention
Description
Self attention layer for 'n_channels'.
Usage
SelfAttention(n_channels)
Arguments
n_channels |
number of channels |
Value
None
SentenceEncoder
Description
Create an encoder over 'module' that can process a full sentence.
Usage
SentenceEncoder(bptt, module, pad_idx = 1, max_len = NULL)
Arguments
bptt |
bptt |
module |
module |
pad_idx |
pad_idx |
max_len |
max_len |
Value
None
SentencePieceTokenizer
Description
SentencePiece tokenizer for 'lang'
Usage
SentencePieceTokenizer(
lang = "en",
special_toks = NULL,
sp_model = NULL,
vocab_sz = NULL,
max_vocab_sz = 30000,
model_type = "unigram",
char_coverage = NULL,
cache_dir = "tmp"
)
Arguments
lang |
lang |
special_toks |
special_toks |
sp_model |
sp_model |
vocab_sz |
vocab_sz |
max_vocab_sz |
max_vocab_sz |
model_type |
model_type |
char_coverage |
char_coverage |
cache_dir |
cache_dir |
Value
None
SeparableBlock
Description
SeparableBlock
Usage
SeparableBlock(expansion, ni, nf, reduction = 16, stride = 1, base_width = 4)
Arguments
expansion |
decoder |
ni |
number of inputs |
nf |
number of features |
reduction |
number of reduction |
stride |
number of stride |
base_width |
base width |
Value
Block object
SequentialEx
Description
SequentialEx
Usage
SequentialEx(...)
Arguments
... |
parameters to pass |
Value
None
Sequential RNN
Description
Sequential RNN
Usage
SequentialRNN(...)
Arguments
... |
parameters to pass |
Value
layer
ShapInterpretation
Description
Base interpereter to use the 'SHAP' interpretation library
Usage
ShapInterpretation(
learn,
test_data = NULL,
link = "identity",
l1_reg = "auto",
n_samples = 128
)
Arguments
learn |
learner/model |
test_data |
should be either a Pandas dataframe or a TabularDataLoader. If not, 100 random rows of the training data will be used instead. |
link |
link can either be "identity" or "logit". A generalized linear model link to connect the feature importance values to the model output. Since the feature importance values, phi, sum up to the model output, it often makes sense to connect them to the ouput with a link function where link(outout) = sum(phi). If the model output is a probability then the LogitLink link function makes the feature importance values have log-odds units. |
l1_reg |
can be an integer value representing the number of features, "auto", "aic", "bic", or a float value. The l1 regularization to use for feature selection (the estimation procedure is based on a debiased lasso). The auto option currently uses "aic" when less that 20 space is enumerated, otherwise it uses no regularization. |
n_samples |
can either be "auto" or an integer value. This is the number of times to re-evaluate the model when explaining each predictions. More samples leads to lower variance estimations of the SHAP values |
Value
None
ShortEpochCallback
Description
Fit just 'pct' of an epoch, then stop
Usage
ShortEpochCallback(pct = 0.01, short_valid = TRUE)
Arguments
pct |
percentage |
short_valid |
short_valid or not |
Value
None
Shortcut
Description
Merge a shortcut with the result of the module by adding them. Adds Conv, BN and ReLU
Usage
Shortcut(ni, nf, act_fn = nn$ReLU(inplace = TRUE))
Arguments
ni |
number of input channels |
nf |
number of features |
act_fn |
activation |
Value
None
ShowCycleGANImgsCallback
Description
Update the progress bar with input and prediction images
Usage
ShowCycleGANImgsCallback(imgA = FALSE, imgB = TRUE, show_img_interval = 10)
Arguments
imgA |
img from A domain |
imgB |
img from B domain |
show_img_interval |
show image interval |
Value
None
ShowGraphCallback
Description
ShowGraphCallback
Usage
ShowGraphCallback(...)
Arguments
... |
parameters to pass |
Value
None
SigmoidRange
Description
Sigmoid module with range '(low, high)'
Usage
SigmoidRange(low, high)
Arguments
low |
low value |
high |
high value |
Value
None
Signal Cutout
Description
Randomly zeros some portion of the signal
Usage
SignalCutout(p = 0.5, max_cut_pct = 0.15)
Arguments
p |
probability |
max_cut_pct |
max cut percentage |
Value
None
Signal Loss
Description
Randomly loses some portion of the signal
Usage
SignalLoss(p = 0.5, max_loss_pct = 0.15)
Arguments
p |
probability |
max_loss_pct |
max loss percentage |
Value
None
Signal Shifter
Description
Randomly shifts the audio signal by 'max_pct'
Usage
SignalShifter(
p = 0.5,
max_pct = 0.2,
max_time = NULL,
direction = 0,
roll = FALSE
)
Arguments
p |
probability |
max_pct |
max percentage |
max_time |
maximum time |
direction |
direction |
roll |
roll or not |
Details
direction must be -1(left) 0(bidirectional) or 1(right).
Value
None
SimpleCNN
Description
Create a simple CNN with 'filters'.
Usage
SimpleCNN(filters, kernel_szs = NULL, strides = NULL, bn = TRUE)
Arguments
filters |
filters number |
kernel_szs |
kernel size |
strides |
strides |
bn |
batch normalization |
Value
None
SimpleSelfAttention
Description
Same as 'nn()$Module', but no need for subclasses to call 'super()$__init__'
Usage
SimpleSelfAttention(n_in, ks = 1, sym = FALSE)
Arguments
n_in |
inputs |
ks |
kernel size |
sym |
sym |
Value
None
SortedDL
Description
A 'DataLoader' that goes throught the item in the order given by 'sort_func'
Usage
SortedDL(
dataset,
sort_func = NULL,
res = NULL,
bs = 64,
shuffle = FALSE,
num_workers = NULL,
verbose = FALSE,
do_setup = TRUE,
pin_memory = FALSE,
timeout = 0,
batch_size = NULL,
drop_last = FALSE,
indexed = NULL,
n = NULL,
device = NULL
)
Arguments
dataset |
dataset |
sort_func |
sort_func |
res |
res |
bs |
bs |
shuffle |
shuffle |
num_workers |
num_workers |
verbose |
verbose |
do_setup |
do_setup |
pin_memory |
pin_memory |
timeout |
timeout |
batch_size |
batch_size |
drop_last |
drop_last |
indexed |
indexed |
n |
n |
device |
device |
Value
None
SpacyTokenizer
Description
Spacy tokenizer for 'lang'
Usage
SpacyTokenizer(lang = "en", special_toks = NULL, buf_sz = 5000)
Arguments
lang |
language |
special_toks |
special tokenizers |
buf_sz |
buffer size |
Value
none
SpearmanCorrCoef
Description
Spearman correlation coefficient for regression problem
Usage
SpearmanCorrCoef(
dim_argmax = NULL,
axis = 0,
nan_policy = "propagate",
activation = "no",
thresh = NULL,
to_np = FALSE,
invert_arg = FALSE,
flatten = TRUE
)
Arguments
dim_argmax |
dim_argmax |
axis |
axis |
nan_policy |
nan_policy |
activation |
activation |
thresh |
thresh |
to_np |
to_np |
invert_arg |
invert_arg |
flatten |
flatten |
Value
None
Spectrogram Transformer
Description
Creates a factory for creating AudioToSpec
Usage
SpectrogramTransformer(mel = TRUE, to_db = TRUE)
Arguments
mel |
mel-spectrogram or not |
to_db |
to decibels |
Details
transforms with different parameters
Value
None
SqueezeNet
Description
Base class for all neural network modules.
Usage
SqueezeNet(version = "1_0", num_classes = 1000)
Arguments
version |
version of SqueezeNet |
num_classes |
the number of classes |
Details
Your models should also subclass this class. Modules can also contain other Modules, allowing to nest them in a tree structure. You can assign the submodules as regular attributes:: import torch.nn as nn import torch.nn.functional as F class Model(nn.Module): def __init__(self): super(Model, self).__init__() self.conv1 = nn.Conv2d(1, 20, 5) self.conv2 = nn.Conv2d(20, 20, 5) def forward(self, x): x = F.relu(self.conv1(x)) return F.relu(self.conv2(x)) Submodules assigned in this way will be registered, and will have their parameters converted too when you call :meth:'to', etc.
Value
model
Swish
Description
Same as nn()$Module, but no need for subclasses to call super()$__init__
Usage
Swish_(...)
Arguments
... |
parameters to pass |
Value
None
TSBlock
Description
A TimeSeries Block to process one timeseries
Usage
TSBlock(...)
Arguments
... |
parameters to pass |
Value
None
TSDataLoaders_from_dfs
Description
Create a DataLoader from a df_train and df_valid
Usage
TSDataLoaders_from_dfs(
df_train,
df_valid,
path = ".",
x_cols = NULL,
label_col = NULL,
y_block = NULL,
item_tfms = NULL,
batch_tfms = NULL,
bs = 64,
val_bs = NULL,
shuffle_train = TRUE,
device = NULL
)
Arguments
df_train |
train data |
df_valid |
validation data |
path |
path (optional) |
x_cols |
predictors |
label_col |
label/output column |
y_block |
y_block |
item_tfms |
item transformations |
batch_tfms |
batch transformations |
bs |
batch size |
val_bs |
validation batch size |
shuffle_train |
shuffle train data |
device |
device name |
Value
None
TSDataTable
Description
A 'DataFrame' wrapper that knows which cols are x/y, and returns rows in '__getitem__'
Usage
TSDataTable(
df,
procs = NULL,
x_names = NULL,
y_names = NULL,
block_y = NULL,
splits = NULL,
do_setup = TRUE,
device = NULL,
inplace = FALSE
)
Arguments
df |
A DataFrame of your data |
procs |
list of preprocess functions |
x_names |
predictors names |
y_names |
the names of the dependent variables |
block_y |
the TransformBlock to use for the target |
splits |
How to split your data |
do_setup |
A parameter for if Tabular will run the data through the procs upon initialization |
device |
device name |
inplace |
If True, Tabular will not keep a separate copy of your original DataFrame in memory |
Value
None
TSeries
Description
Basic Time series wrapper
Usage
TSeries(...)
Arguments
... |
parameters to pass |
Value
None
TSeries_create
Description
TSeries_create
Usage
TSeries_create(x, ...)
Arguments
x |
tensor |
... |
additional parameters |
Value
tensor
Examples
## Not run:
res = TSeries_create(as.array(runif(100)))
res %>% show(title = 'R array') %>% plot(dpi = 200)
## End(Not run)
TabularDataTable
Description
A 'Tabular' object with transforms
Usage
TabularDataTable(
df,
procs = NULL,
cat_names = NULL,
cont_names = NULL,
y_names = NULL,
y_block = NULL,
splits = NULL,
do_setup = TRUE,
device = NULL,
inplace = FALSE,
reduce_memory = TRUE,
...
)
Arguments
df |
A DataFrame of your data |
procs |
list of preprocess functions |
cat_names |
the names of the categorical variables |
cont_names |
the names of the continuous variables |
y_names |
the names of the dependent variables |
y_block |
the TransformBlock to use for the target |
splits |
How to split your data |
do_setup |
A parameter for if Tabular will run the data through the procs upon initialization |
device |
cuda or cpu |
inplace |
If True, Tabular will not keep a separate copy of your original DataFrame in memory |
reduce_memory |
fastai will attempt to reduce the overall memory usage |
... |
additional parameters to pass |
Value
None
TabularModel
Description
Basic model for tabular data.
Usage
TabularModel(
emb_szs,
n_cont,
out_sz,
layers,
ps = NULL,
embed_p = 0,
y_range = NULL,
use_bn = TRUE,
bn_final = FALSE,
bn_cont = TRUE,
act_cls = nn()$ReLU(inplace = TRUE)
)
Arguments
emb_szs |
embedding size |
n_cont |
number of cont |
out_sz |
output size |
layers |
layers |
ps |
ps |
embed_p |
embed proportion |
y_range |
y range |
use_bn |
use batch normalization |
bn_final |
batch normalization final |
bn_cont |
batch normalization cont |
act_cls |
activation |
Value
None
TabularTS
Description
A 'DataFrame' wrapper that knows which cols are x/y, and returns rows in '__getitem__'
Usage
TabularTS(
df,
procs = NULL,
x_names = NULL,
y_names = NULL,
block_y = NULL,
splits = NULL,
do_setup = TRUE,
device = NULL,
inplace = FALSE
)
Arguments
df |
A DataFrame of your data |
procs |
list of preprocess functions |
x_names |
predictors names |
y_names |
the names of the dependent variables |
block_y |
the TransformBlock to use for the target |
splits |
How to split your data |
do_setup |
A parameter for if Tabular will run the data through the procs upon initialization |
device |
device name |
inplace |
If True, Tabular will not keep a separate copy of your original DataFrame in memory |
Value
None
TabularTSDataloader
Description
Transformed 'DataLoader'
Usage
TabularTSDataloader(
dataset,
bs = 16,
shuffle = FALSE,
after_batch = NULL,
num_workers = 0,
verbose = FALSE,
do_setup = TRUE,
pin_memory = FALSE,
timeout = 0,
batch_size = NULL,
drop_last = FALSE,
indexed = NULL,
n = NULL,
device = NULL
)
Arguments
dataset |
data set |
bs |
batch size |
shuffle |
shuffle or not |
after_batch |
after batch |
num_workers |
the number of workers |
verbose |
verbose |
do_setup |
A parameter for if Tabular will run the data through the procs upon initialization |
pin_memory |
pin memory or not |
timeout |
timeout |
batch_size |
batch size |
drop_last |
drop last |
indexed |
indexed |
n |
n |
device |
device name |
Value
None
TensorBBox
Description
Basic type for a tensor of bounding boxes in an image
Usage
TensorBBox(x)
Arguments
x |
tensor |
Value
None
TensorBBox_create
Description
TensorBBox_create
Usage
TensorBBox_create(x, img_size = NULL)
Arguments
x |
tensor |
img_size |
image size |
Value
None
TensorImage
Description
TensorImage
Usage
TensorImage(x)
Arguments
x |
tensor |
Value
None
TensorImageBW
Description
TensorImageBW
Usage
TensorImageBW(x)
Arguments
x |
tensor |
Value
None
TensorMultiCategory
Description
TensorMultiCategory
Usage
TensorMultiCategory(x)
Arguments
x |
tensor |
Value
None
TensorPoint
Description
Basic type for points in an image
Usage
TensorPoint(x)
Arguments
x |
tensor |
Value
None
TensorPoint_create
Description
Delegates ('__call__','decode','setup') to ('encodes','decodes','setups') if 'split_idx' matches
Usage
TensorPoint_create(...)
Arguments
... |
arguments to pass |
Value
None
TerminateOnNaNCallback
Description
TerminateOnNaNCallback
Usage
TerminateOnNaNCallback(...)
Arguments
... |
parameters to pass |
Value
None
TextBlock
Description
A 'TransformBlock' for texts
Usage
TextBlock(
tok_tfm,
vocab = NULL,
is_lm = FALSE,
seq_len = 72,
backwards = FALSE,
min_freq = 3,
max_vocab = 60000,
special_toks = NULL,
pad_tok = NULL
)
Arguments
tok_tfm |
tok_tfm |
vocab |
vocab |
is_lm |
is_lm |
seq_len |
seq_len |
backwards |
backwards |
min_freq |
min_freq |
max_vocab |
max_vocab |
special_toks |
special_toks |
pad_tok |
pad_tok |
Value
block object
TextBlock_from_df
Description
Build a 'TextBlock' from a dataframe using 'text_cols'
Usage
TextBlock_from_df(
text_cols,
vocab = NULL,
is_lm = FALSE,
seq_len = 72,
backwards = FALSE,
min_freq = 3,
max_vocab = 60000,
tok = NULL,
rules = NULL,
sep = " ",
n_workers = 6,
mark_fields = NULL,
tok_text_col = "text"
)
Arguments
text_cols |
text columns |
vocab |
vocabulary |
is_lm |
is_lm |
seq_len |
sequence length |
backwards |
backwards |
min_freq |
minimum frequency |
max_vocab |
max vocabulary |
tok |
tokenizer |
rules |
rules |
sep |
separator |
n_workers |
number workers |
mark_fields |
mark_fields |
tok_text_col |
result column name |
Value
None
TextBlock_from_folder
Description
Build a 'TextBlock' from a 'path'
Usage
TextBlock_from_folder(
path,
vocab = NULL,
is_lm = FALSE,
seq_len = 72,
backwards = FALSE,
min_freq = 3,
max_vocab = 60000,
tok = NULL,
rules = NULL,
extensions = NULL,
folders = NULL,
output_dir = NULL,
skip_if_exists = TRUE,
output_names = NULL,
n_workers = 6,
encoding = "utf8"
)
Arguments
path |
path |
vocab |
vocabualry |
is_lm |
is_lm |
seq_len |
sequence length |
backwards |
backwards |
min_freq |
minimum frequency |
max_vocab |
max vocabulary |
tok |
tokenizer |
rules |
rules |
extensions |
extensions |
folders |
folders |
output_dir |
output_dir |
skip_if_exists |
skip_if_exists |
output_names |
output_names |
n_workers |
number of workers |
encoding |
encoding |
Value
None
TextDataLoaders_from_csv
Description
Create from 'csv' file in 'path/csv_fname'
Usage
TextDataLoaders_from_csv(
path,
csv_fname = "labels.csv",
header = "infer",
delimiter = NULL,
valid_pct = 0.2,
seed = NULL,
text_col = 0,
label_col = 1,
label_delim = NULL,
y_block = NULL,
text_vocab = NULL,
is_lm = FALSE,
valid_col = NULL,
tok_tfm = NULL,
seq_len = 72,
backwards = FALSE,
bs = 64,
val_bs = NULL,
shuffle_train = TRUE,
device = NULL
)
Arguments
path |
path |
csv_fname |
csv file name |
header |
header |
delimiter |
delimiter |
valid_pct |
valid_ation percentage |
seed |
random seed |
text_col |
text column |
label_col |
label column |
label_delim |
label separator |
y_block |
y_block |
text_vocab |
text vocabulary |
is_lm |
is_lm |
valid_col |
valid column |
tok_tfm |
tok_tfm |
seq_len |
seq_len |
backwards |
backwards |
bs |
batch size |
val_bs |
validation batch size |
shuffle_train |
shuffle train data |
device |
device |
Value
text loader
TextDataLoaders_from_df
Description
Create from 'df' in 'path' with 'valid_pct' '
Usage
TextDataLoaders_from_df(
df,
path = ".",
valid_pct = 0.2,
seed = NULL,
text_col = 0,
label_col = 1,
label_delim = NULL,
y_block = NULL,
text_vocab = NULL,
is_lm = FALSE,
valid_col = NULL,
tok_tfm = NULL,
seq_len = 72,
backwards = FALSE,
bs = 64,
val_bs = NULL,
shuffle_train = TRUE,
device = NULL
)
Arguments
df |
df |
path |
path |
valid_pct |
validation percentage |
seed |
seed |
text_col |
text_col |
label_col |
label_col |
label_delim |
label_delim |
y_block |
y_block |
text_vocab |
text_vocab |
is_lm |
is_lm |
valid_col |
valid_col |
tok_tfm |
tok_tfm |
seq_len |
seq_len |
backwards |
backwards |
bs |
batch size |
val_bs |
validation batch size, if not specified then val_bs is the same as bs. |
shuffle_train |
shuffle_train |
device |
device |
Value
text loader
TextDataLoaders_from_folder
Description
Create from imagenet style dataset in 'path' with 'train' and 'valid' subfolders (or provide 'valid_pct')
Usage
TextDataLoaders_from_folder(
path,
train = "train",
valid = "valid",
valid_pct = NULL,
seed = NULL,
vocab = NULL,
text_vocab = NULL,
is_lm = FALSE,
tok_tfm = NULL,
seq_len = 72,
backwards = FALSE,
bs = 64,
val_bs = NULL,
shuffle_train = TRUE,
device = NULL
)
Arguments
path |
path |
train |
train data |
valid |
validation data |
valid_pct |
validation percentage |
seed |
random seed |
vocab |
vocabulary |
text_vocab |
text_vocab |
is_lm |
is_lm |
tok_tfm |
tok_tfm |
seq_len |
seq_len |
backwards |
backwards |
bs |
batch size |
val_bs |
validation batch size |
shuffle_train |
shuffle train data |
device |
device |
Value
text loader
TextLearner
Description
Basic class for a 'Learner' in NLP.
Usage
TextLearner(
dls,
model,
alpha = 2,
beta = 1,
moms = list(0.8, 0.7, 0.8),
loss_func = NULL,
opt_func = Adam(),
lr = 0.001,
splitter = trainable_params(),
cbs = NULL,
metrics = NULL,
path = NULL,
model_dir = "models",
wd = NULL,
wd_bn_bias = FALSE,
train_bn = TRUE
)
Arguments
dls |
dls |
model |
model |
alpha |
alpha |
beta |
beta |
moms |
moms |
loss_func |
loss_func |
opt_func |
opt_func |
lr |
lr |
splitter |
splitter |
cbs |
cbs |
metrics |
metrics |
path |
path |
model_dir |
model_dir |
wd |
wd |
wd_bn_bias |
wd_bn_bias |
train_bn |
train_bn |
Value
None
Load_encoder
Description
Load the encoder ‘file' from the model directory, optionally ensuring it’s on 'device'
Usage
TextLearner_load_encoder(file, device = NULL)
Arguments
file |
file |
device |
device |
Value
None
Load_pretrained
Description
Load a pretrained model and adapt it to the data vocabulary.
Usage
TextLearner_load_pretrained(wgts_fname, vocab_fname, model = NULL)
Arguments
wgts_fname |
wgts_fname |
vocab_fname |
vocab_fname |
model |
model |
Value
None
Save_encoder
Description
Save the encoder to 'file' in the model directory
Usage
TextLearner_save_encoder(file)
Arguments
file |
file |
Value
None
TfmResize
Description
Temporary fix to allow image resizing transform
Usage
TfmResize(size, interp_mode = "bilinear")
Arguments
size |
size |
interp_mode |
interpolation mode |
Value
None
TfmdDL
Description
Transformed 'DataLoader'
Usage
TfmdDL(
dataset,
bs = 64,
shuffle = FALSE,
num_workers = NULL,
verbose = FALSE,
do_setup = TRUE,
pin_memory = FALSE,
timeout = 0,
batch_size = NULL,
drop_last = FALSE,
indexed = NULL,
n = NULL,
device = NULL,
after_batch = NULL,
...
)
Arguments
dataset |
dataset |
bs |
batch size |
shuffle |
shuffle |
num_workers |
number of workers |
verbose |
verbose |
do_setup |
do setup |
pin_memory |
pin memory |
timeout |
timeout |
batch_size |
batch size |
drop_last |
drop last |
indexed |
indexed |
n |
int, n |
device |
device |
after_batch |
after_batch |
... |
additional arguments to pass |
Value
None
TfmdLists
Description
A 'Pipeline' of 'tfms' applied to a collection of 'items'
Usage
TfmdLists(...)
Arguments
... |
parameters to pass |
ToTensor
Description
Convert item to appropriate tensor class
Usage
ToTensor(enc = NULL, dec = NULL, split_idx = NULL, order = NULL)
Arguments
enc |
encoder |
dec |
decoder |
split_idx |
int, split by index |
order |
order |
Value
None
TokenizeWithRules
Description
A wrapper around 'tok' which applies 'rules', then tokenizes, then applies 'post_rules'
Usage
TokenizeWithRules(tok, rules = NULL, post_rules = NULL)
Arguments
tok |
tokenizer |
rules |
rules |
post_rules |
post_rules |
Value
None
Tokenizer
Description
Provides a consistent 'Transform' interface to tokenizers operating on 'DataFrame's and folders
Usage
Tokenizer(
tok,
rules = NULL,
counter = NULL,
lengths = NULL,
mode = NULL,
sep = " "
)
Arguments
tok |
tokenizer |
rules |
rules |
counter |
counter |
lengths |
lengths |
mode |
mode |
sep |
separator |
Value
None
Tokenizer_from_df
Description
Tokenizer_from_df
Usage
Tokenizer_from_df(
text_cols,
tok = NULL,
rules = NULL,
sep = " ",
n_workers = 6,
mark_fields = NULL,
tok_text_col = "text"
)
Arguments
text_cols |
text columns |
tok |
tokenizer |
rules |
special rules |
sep |
separator |
n_workers |
number of workers |
mark_fields |
mark fields |
tok_text_col |
output column name |
Value
None
TrackerCallback
Description
A 'Callback' that keeps track of the best value in 'monitor'.
Usage
TrackerCallback(monitor = "valid_loss", comp = NULL, min_delta = 0)
Arguments
monitor |
monitor the loss |
comp |
comp |
min_delta |
minimum delta |
Value
None
TrainEvalCallback
Description
TrainEvalCallback
Usage
TrainEvalCallback(...)
Arguments
... |
parameters to pass |
Value
None
Transform
Description
Delegates ('__call__','decode','setup') to ('encodes','decodes','setups') if 'split_idx' matches
Usage
Transform(enc = NULL, dec = NULL, split_idx = NULL, order = NULL)
Arguments
enc |
encoder |
dec |
decoder |
split_idx |
split by index |
order |
order |
Value
None
TransformBlock
Description
A basic wrapper that links defaults transforms for the data block API
Usage
TransformBlock(
type_tfms = NULL,
item_tfms = NULL,
batch_tfms = NULL,
dl_type = NULL,
dls_kwargs = NULL
)
Arguments
type_tfms |
transformation type |
item_tfms |
item transformation type |
batch_tfms |
one or several transforms applied to the batches once they are formed |
dl_type |
DL application |
dls_kwargs |
additional arguments |
Value
block
TransformersDropOutput
Description
TransformersDropOutput
Usage
TransformersDropOutput()
Value
None
TransformersTokenizer
Description
TransformersTokenizer
Usage
TransformersTokenizer(tokenizer)
Arguments
tokenizer |
tokenizer object |
Value
None
ADULT_SAMPLE dataset
Description
download ADULT_SAMPLE dataset
Usage
URLs_ADULT_SAMPLE(filename = "ADULT_SAMPLE", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
Examples
## Not run:
URLs_ADULT_SAMPLE()
## End(Not run)
AG_NEWS dataset
Description
download AG_NEWS dataset
Usage
URLs_AG_NEWS(filename = "AG_NEWS", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
Examples
## Not run:
URLs_AG_NEWS()
## End(Not run)
AMAZON_REVIEWSAMAZON_REVIEWS dataset
Description
download AMAZON_REVIEWSAMAZON_REVIEWS dataset
Usage
URLs_AMAZON_REVIEWSAMAZON_REVIEWS(
filename = "AMAZON_REVIEWSAMAZON_REVIEWS",
untar = TRUE
)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
AMAZON_REVIEWS_POLARITY dataset
Description
download AMAZON_REVIEWS_POLARITY dataset
Usage
URLs_AMAZON_REVIEWS_POLARITY(
filename = "AMAZON_REVIEWS_POLARITY",
untar = TRUE
)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
BIWI_HEAD_POSE dataset
Description
download BIWI_HEAD_POSE dataset
Usage
URLs_BIWI_HEAD_POSE(filename = "BIWI_HEAD_POSE", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
CALTECH_101 dataset
Description
download CALTECH_101 dataset
Usage
URLs_CALTECH_101(filename = "CALTECH_101", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
CAMVID dataset
Description
download CAMVID dataset
Usage
URLs_CAMVID(filename = "CAMVID", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
CAMVID_TINY dataset
Description
download CAMVID_TINY dataset
Usage
URLs_CAMVID_TINY(filename = "CAMVID_TINY", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
CARS dataset
Description
download CARS dataset
Usage
URLs_CARS(filename = "CARS", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
CIFAR dataset
Description
download CIFAR dataset
Usage
URLs_CIFAR(filename = "CIFAR", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
CIFAR_100 dataset
Description
download CIFAR_100 dataset
Usage
URLs_CIFAR_100(filename = "CIFAR_100", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
COCO_TINY dataset
Description
download COCO_TINY dataset
Usage
URLs_COCO_TINY(filename = "COCO_TINY", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
CUB_200_2011 dataset
Description
download CUB_200_2011 dataset
Usage
URLs_CUB_200_2011(filename = "CUB_200_2011", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
DBPEDIA dataset
Description
download DBPEDIA dataset
Usage
URLs_DBPEDIA(filename = "DBPEDIA", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
DOGS dataset
Description
download DOGS dataset
Usage
URLs_DOGS(filename = "DOGS", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
FLOWERS dataset
Description
download FLOWERS dataset
Usage
URLs_FLOWERS(filename = "FLOWERS", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
FOOD dataset
Description
download FOOD dataset
Usage
URLs_FOOD(filename = "FOOD", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
HORSE_2_ZEBRA dataset
Description
download HORSE_2_ZEBRA dataset
Usage
URLs_HORSE_2_ZEBRA(filename = "horse2zebra", unzip = TRUE)
Arguments
filename |
the name of the file |
unzip |
logical, whether to unzip the '.zip' file |
Value
None
HUMAN_NUMBERS dataset
Description
download HUMAN_NUMBERS dataset
Usage
URLs_HUMAN_NUMBERS(filename = "HUMAN_NUMBERS", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
IMAGENETTE dataset
Description
download IMAGENETTE dataset
Usage
URLs_IMAGENETTE(filename = "IMAGENETTE", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
IMAGENETTE_160 dataset
Description
download IMAGENETTE_160 dataset
Usage
URLs_IMAGENETTE_160(filename = "IMAGENETTE_160", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
IMAGENETTE_320 dataset
Description
download IMAGENETTE_320 dataset
Usage
URLs_IMAGENETTE_320(filename = "IMAGENETTE_320", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
IMAGEWOOF dataset
Description
download IMAGEWOOF dataset
Usage
URLs_IMAGEWOOF(filename = "IMAGEWOOF", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
IMAGEWOOF_160 dataset
Description
download IMAGEWOOF_160 dataset
Usage
URLs_IMAGEWOOF_160(filename = "IMAGEWOOF_160", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
IMAGEWOOF_320 dataset
Description
download IMAGEWOOF_320 dataset
Usage
URLs_IMAGEWOOF_320(filename = "IMAGEWOOF_320", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
IMDB dataset
Description
download IMDB dataset
Usage
URLs_IMDB(filename = "IMDB", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
IMDB_SAMPLE dataset
Description
download IMDB_SAMPLE dataset
Usage
URLs_IMDB_SAMPLE(filename = "IMDB_SAMPLE", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
LSUN_BEDROOMS dataset
Description
download LSUN_BEDROOMS dataset
Usage
URLs_LSUN_BEDROOMS(filename = "LSUN_BEDROOMS", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
ML_SAMPLE dataset
Description
download ML_SAMPLE dataset
Usage
URLs_ML_SAMPLE(filename = "ML_SAMPLE", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
MNIST dataset
Description
download MNIST dataset
Usage
URLs_MNIST(filename = "MNIST", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
MNIST_SAMPLE dataset
Description
download MNIST_SAMPLE dataset
Usage
URLs_MNIST_SAMPLE(filename = "MNIST_SAMPLE", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
MNIST_TINY dataset
Description
download MNIST_TINY dataset
Usage
URLs_MNIST_TINY(filename = "MNIST_TINY", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
MNIST_VAR_SIZE_TINY dataset
Description
download MNIST_VAR_SIZE_TINY dataset
Usage
URLs_MNIST_VAR_SIZE_TINY(filename = "MNIST_VAR_SIZE_TINY", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
MOVIE_LENS_ML_100k dataset
Description
download MOVIE_LENS_ML_100k dataset
Usage
URLs_MOVIE_LENS_ML_100k(filename = "ml-100k", unzip = TRUE)
Arguments
filename |
the name of the file |
unzip |
logical, whether to unzip the '.zip' file |
Value
None
MT_ENG_FRA dataset
Description
download MT_ENG_FRA dataset
Usage
URLs_MT_ENG_FRA(filename = "MT_ENG_FRA", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
OPENAI_TRANSFORMER dataset
Description
download OPENAI_TRANSFORMER dataset
Usage
URLs_OPENAI_TRANSFORMER(filename = "OPENAI_TRANSFORMER", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
PASCAL_2007 dataset
Description
download PASCAL_2007 dataset
Usage
URLs_PASCAL_2007(filename = "PASCAL_2007", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
PASCAL_2012 dataset
Description
download PASCAL_2012 dataset
Usage
URLs_PASCAL_2012(filename = "PASCAL_2012", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
PETS dataset
Description
download PETS dataset
Usage
URLs_PETS(filename = "PETS", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
PLANET_SAMPLE dataset
Description
download PLANET_SAMPLE dataset
Usage
URLs_PLANET_SAMPLE(filename = "PLANET_SAMPLE", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
PLANET_TINY dataset
Description
download PLANET_TINY dataset
Usage
URLs_PLANET_TINY(filename = "PLANET_TINY", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
S3_COCO dataset
Description
download S3_COCO dataset
Usage
URLs_S3_COCO(filename = "S3_COCO", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
S3_IMAGE dataset
Description
download S3_IMAGE dataset
Usage
URLs_S3_IMAGE(filename = "S3_IMAGE", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
S3_IMAGELOC dataset
Description
download S3_IMAGELOC dataset
Usage
URLs_S3_IMAGELOC(filename = "S3_IMAGELOC", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
S3_MODEL dataset
Description
download S3_MODEL dataset
Usage
URLs_S3_MODEL(filename = "S3_MODEL", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
S3_NLP dataset
Description
download S3_NLP dataset
Usage
URLs_S3_NLP(filename = "S3_NLP", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
SIIM_SMALL
Description
download YELP_REVIEWS_POLARITY dataset
Usage
URLs_SIIM_SMALL(filename = "SIIM_SMALL", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
SKIN_LESION dataset
Description
download SKIN_LESION dataset
Usage
URLs_SKIN_LESION(filename = "SKIN_LESION", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
SOGOU_NEWS dataset
Description
download SOGOU_NEWS dataset
Usage
URLs_SOGOU_NEWS(filename = "SOGOU_NEWS", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
SPEAKERS10 dataset
Description
download SPEAKERS10 dataset
Usage
URLs_SPEAKERS10(filename = "SPEAKERS10", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
Examples
## Not run:
URLs_SPEAKERS10()
## End(Not run)
SPEECHCOMMANDS dataset
Description
download SPEECHCOMMANDS dataset
Usage
URLs_SPEECHCOMMANDS(filename = "SPEECHCOMMANDS", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
Examples
## Not run:
URLs_SPEECHCOMMANDS()
## End(Not run)
WIKITEXT dataset
Description
download WIKITEXT dataset
Usage
URLs_WIKITEXT(filename = "WIKITEXT", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
WIKITEXT_TINY dataset
Description
download WIKITEXT_TINY dataset
Usage
URLs_WIKITEXT_TINY(filename = "WIKITEXT_TINY", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
WT103_BWD dataset
Description
download WT103_BWD dataset
Usage
URLs_WT103_BWD(filename = "WT103_BWD", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
WT103_FWD dataset
Description
download WT103_FWD dataset
Usage
URLs_WT103_FWD(filename = "WT103_FWD", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
YAHOO_ANSWERS dataset
Description
download YAHOO_ANSWERS dataset
Usage
URLs_YAHOO_ANSWERS(filename = "YAHOO_ANSWERS", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
YELP_REVIEWS dataset
Description
download YELP_REVIEWS dataset
Usage
URLs_YELP_REVIEWS(filename = "YELP_REVIEWS", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
YELP_REVIEWS_POLARITY dataset
Description
download YELP_REVIEWS_POLARITY dataset
Usage
URLs_YELP_REVIEWS_POLARITY(filename = "YELP_REVIEWS_POLARITY", untar = TRUE)
Arguments
filename |
the name of the file |
untar |
logical, whether to untar the '.tgz' file |
Value
None
UnetBlock
Description
A quasi-UNet block, using 'PixelShuffle_ICNR upsampling'.
Usage
UnetBlock(
up_in_c,
x_in_c,
hook,
final_div = TRUE,
blur = FALSE,
act_cls = nn()$ReLU,
self_attention = FALSE,
init = nn()$init$kaiming_normal_,
norm_type = NULL,
ks = 3,
stride = 1,
padding = NULL,
bias = NULL,
ndim = 2,
bn_1st = TRUE,
transpose = FALSE,
xtra = NULL,
bias_std = 0.01,
dilation = 1,
groups = 1,
padding_mode = "zeros"
)
Arguments
up_in_c |
up_in_c parameter |
x_in_c |
x_in_c parameter |
hook |
The hook is set to this intermediate layer to store the output needed for this block. |
final_div |
final div |
blur |
blur is used to avoid checkerboard artifacts at each layer. |
act_cls |
activation |
self_attention |
self_attention determines if we use a self-attention layer |
init |
initializer |
norm_type |
normalization type |
ks |
kernel size |
stride |
stride |
padding |
padding mode |
bias |
bias |
ndim |
number of dimensions |
bn_1st |
batch normalization 1st |
transpose |
transpose |
xtra |
xtra |
bias_std |
bias standard deviation |
dilation |
dilation |
groups |
groups |
padding_mode |
The mode of padding |
Value
None
Voice
Description
Voice
Usage
Voice(
sample_rate = 16000,
n_fft = 1024,
win_length = NULL,
hop_length = 128,
f_min = 50,
f_max = 8000,
pad = 0,
n_mels = 128,
window_fn = torch()$hann_window,
power = 2,
normalized = FALSE,
wkwargs = NULL,
mel = TRUE,
to_db = TRUE
)
Arguments
sample_rate |
sample rate |
n_fft |
number of fast fourier transforms |
win_length |
windowing length |
hop_length |
hopping length |
f_min |
minimum frequency |
f_max |
maximum frequency |
pad |
padding mode |
n_mels |
number of mel-spectrograms |
window_fn |
window function |
power |
power |
normalized |
normalized or not |
wkwargs |
additional arguments |
mel |
mel-spectrogram or not |
to_db |
to decibels |
Value
None
WandbCallback
Description
Saves model topology, losses & metrics
Usage
WandbCallback(
log = "gradients",
log_preds = TRUE,
log_model = TRUE,
log_dataset = FALSE,
dataset_name = NULL,
valid_dl = NULL,
n_preds = 36,
seed = 12345,
reorder = TRUE
)
Arguments
log |
"gradients" (default), "parameters", "all" or None. Losses & metrics are always logged. |
log_preds |
whether we want to log prediction samples (default to True). |
log_model |
whether we want to log our model (default to True). This also requires SaveModelCallback. |
log_dataset |
Options: - False (default) - True will log folder referenced by learn.dls.path. - a path can be defined explicitly to reference which folder to log. Note: subfolder "models" is always ignored. |
dataset_name |
name of logged dataset (default to folder name). |
valid_dl |
DataLoaders containing items used for prediction samples (default to random items from learn.dls.valid. |
n_preds |
number of logged predictions (default to 36). |
seed |
used for defining random samples. |
reorder |
reorder or not |
Value
None
Warp
Description
Apply perspective warping with 'magnitude' and 'p' on a batch of matrices
Usage
Warp(
magnitude = 0.2,
p = 0.5,
draw_x = NULL,
draw_y = NULL,
size = NULL,
mode = "bilinear",
pad_mode = "reflection",
batch = FALSE,
align_corners = TRUE
)
Arguments
magnitude |
magnitude |
p |
probability |
draw_x |
draw x |
draw_y |
draw y |
size |
size |
mode |
mode |
pad_mode |
padding mode |
batch |
batch |
align_corners |
align corners |
Value
None
WeightDropout
Description
A module that wraps another layer in which some weights will be replaced by 0 during training.
Usage
WeightDropout(module, weight_p, layer_names = "weight_hh_l0")
Arguments
module |
module |
weight_p |
weight_p |
layer_names |
layer_names |
Value
None
WeightedDL
Description
Transformed 'DataLoader'
Usage
WeightedDL(
dataset = NULL,
bs = NULL,
wgts = NULL,
shuffle = FALSE,
num_workers = NULL,
verbose = FALSE,
do_setup = TRUE,
pin_memory = FALSE,
timeout = 0,
batch_size = NULL,
drop_last = FALSE,
indexed = NULL,
n = NULL,
device = NULL,
persistent_workers = FALSE
)
Arguments
dataset |
dataset |
bs |
bs |
wgts |
weights |
shuffle |
shuffle |
num_workers |
number of workers |
verbose |
verbose |
do_setup |
do_setup |
pin_memory |
pin_memory |
timeout |
timeout |
batch_size |
batch_size |
drop_last |
drop_last |
indexed |
indexed |
n |
n |
device |
device |
persistent_workers |
persistent_workers |
Value
None
XResNet
Description
A sequential container.
Usage
XResNet(block, expansion, layers, c_in = 3, c_out = 1000, ...)
Arguments
block |
the blocks to pass to XResNet |
expansion |
argument for inputs and filters |
layers |
the layers to pass to XResNet |
c_in |
number of inputs |
c_out |
number of outputs |
... |
additional arguments |
Zoom
Description
Apply a random zoom of at most 'max_zoom' with probability 'p' to a batch of images
Usage
Zoom_(
min_zoom = 1,
max_zoom = 1.1,
p = 0.5,
draw = NULL,
draw_x = NULL,
draw_y = NULL,
size = NULL,
mode = "bilinear",
pad_mode = "reflection",
batch = FALSE,
align_corners = TRUE
)
Arguments
min_zoom |
minimum zoom |
max_zoom |
maximum zoom |
p |
probability |
draw |
draw |
draw_x |
draw x |
draw_y |
draw y |
size |
size |
mode |
mode |
pad_mode |
pad mode |
batch |
batch |
align_corners |
align corners or not |
Value
None
Pow
Description
Pow
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
a ^ b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Abs
Description
Abs
Usage
## S3 method for class 'torch.Tensor'
abs(x)
Arguments
x |
tensor |
Value
tensor
Abs
Description
Abs
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
abs(x)
Arguments
x |
tensor, e.g.: tensor(-1:-10) |
Value
tensor
Accuracy
Description
Compute accuracy with 'targ' when 'pred' is bs * n_classes
Usage
accuracy(inp, targ, axis = -1)
Arguments
inp |
predictions |
targ |
targets |
axis |
axis |
Value
None
Accuracy_multi
Description
Compute accuracy when 'inp' and 'targ' are the same size.
Usage
accuracy_multi(inp, targ, thresh = 0.5, sigmoid = TRUE)
Arguments
inp |
predictions |
targ |
targets |
thresh |
threshold point |
sigmoid |
sigmoid |
Value
None
Accuracy threshold expand
Description
Compute accuracy after expanding 'y_true' to the size of 'y_pred'.
Usage
accuracy_thresh_expand(y_pred, y_true, thresh = 0.5, sigmoid = TRUE)
Arguments
y_pred |
predictions |
y_true |
actuals |
thresh |
threshold point |
sigmoid |
sigmoid function |
Value
None
Adam_step
Description
Step for Adam with 'lr' on 'p'
Usage
adam_step(p, lr, mom, step, sqr_mom, grad_avg, sqr_avg, eps, ...)
Arguments
p |
p |
lr |
learning rate |
mom |
momentum |
step |
step |
sqr_mom |
sqr momentum |
grad_avg |
grad average |
sqr_avg |
sqr average |
eps |
epsilon |
... |
additional arguments to pass |
Value
None
Adaptive_pool
Description
Adaptive_pool
Usage
adaptive_pool(pool_type)
Arguments
pool_type |
pooling type |
Value
Nonee
Add
Description
Add
Sinh
Usage
## S3 method for class 'torch.Tensor'
a + b
## S3 method for class 'torch.Tensor'
sinh(x)
Arguments
a |
tensor |
b |
tensor |
x |
tensor |
Value
tensor
tensor
Add cyclic datepart
Description
Helper function that adds trigonometric date/time features to a date in the column 'field_name' of 'df'.
Usage
add_cyclic_datepart(
df,
field_name,
prefix = NULL,
drop = TRUE,
time = FALSE,
add_linear = FALSE
)
Arguments
df |
df |
field_name |
field_name |
prefix |
prefix |
drop |
drop |
time |
time |
add_linear |
add_linear |
Value
data frame
Add datepart
Description
Helper function that adds columns relevant to a date in the column 'field_name' of 'df'.
Usage
add_datepart(df, field_name, prefix = NULL, drop = TRUE, time = FALSE)
Arguments
df |
df |
field_name |
field_name |
prefix |
prefix |
drop |
drop |
time |
time |
Value
data frame
Aaffine_coord
Description
Aaffine_coord
Usage
affine_coord(
x,
mat = NULL,
coord_tfm = NULL,
sz = NULL,
mode = "bilinear",
pad_mode = "reflection",
align_corners = TRUE,
...
)
Arguments
x |
tensor |
mat |
mat |
coord_tfm |
coordinate tfm |
sz |
sz |
mode |
mode |
pad_mode |
padding mode |
align_corners |
align corners |
... |
additional arguments |
Value
None
Affline mat
Description
Affline mat
Usage
affine_mat(...)
Arguments
... |
parameters to pass |
Value
None
Alexnet
Description
AlexNet model architecture
Usage
alexnet(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
"One weird trick..." <https://arxiv.org/abs/1404.5997>
Value
model
Examples
## Not run:
alexnet(pretrained = FALSE, progress = TRUE)
## End(Not run)
Apply_perspective
Description
Apply perspective tranfom on 'coords' with 'coeffs'
Usage
apply_perspective(coords, coeffs)
Arguments
coords |
coordinates |
coeffs |
coefficient |
Value
None
As_array
Description
As_array
Usage
as_array(tensor)
Arguments
tensor |
tensor object |
Value
array
Aspect
Description
Aspect
Usage
aspect(img)
Arguments
img |
image |
Value
None
Audio_extensions
Description
get all allowed audio extensions
Usage
audio_extensions()
Value
vector
Augmentation
Description
Utility func to easily create a list of flip, rotate, zoom, warp, lighting transforms.
Usage
aug_transforms(
mult = 1,
do_flip = TRUE,
flip_vert = FALSE,
max_rotate = 10,
min_zoom = 1,
max_zoom = 1.1,
max_lighting = 0.2,
max_warp = 0.2,
p_affine = 0.75,
p_lighting = 0.75,
xtra_tfms = NULL,
size = NULL,
mode = "bilinear",
pad_mode = "reflection",
align_corners = TRUE,
batch = FALSE,
min_scale = 1
)
Arguments
mult |
ratio |
do_flip |
to do flip |
flip_vert |
flip vertical or not |
max_rotate |
maximum rotation |
min_zoom |
minimum zoom |
max_zoom |
maximum zoom |
max_lighting |
maximum lighting |
max_warp |
maximum warp |
p_affine |
probability affine |
p_lighting |
probability lighting |
xtra_tfms |
extra transformations |
size |
size of image |
mode |
mode |
pad_mode |
padding mode |
align_corners |
align_corners |
batch |
batch size |
min_scale |
minimum scale |
Value
None
Examples
## Not run:
URLs_PETS()
path = 'oxford-iiit-pet'
path_img = 'oxford-iiit-pet/images'
fnames = get_image_files(path_img)
dls = ImageDataLoaders_from_name_re(
path, fnames, pat='(.+)_.jpg$',
item_tfms=Resize(size = 460), bs = 10,
batch_tfms=list(aug_transforms(size = 224, min_scale = 0.75),
Normalize_from_stats( imagenet_stats() )
)
)
## End(Not run)
Average_grad
Description
Keeps track of the avg grads of 'p' in 'state' with 'mom'.
Usage
average_grad(p, mom, dampening = FALSE, grad_avg = NULL, ...)
Arguments
p |
p |
mom |
momentum |
dampening |
dampening |
grad_avg |
grad average |
... |
additional args to pass |
Value
None
Average_sqr_grad
Description
Average_sqr_grad
Usage
average_sqr_grad(p, sqr_mom, dampening = TRUE, sqr_avg = NULL, ...)
Arguments
p |
p |
sqr_mom |
sqr momentum |
dampening |
dampening |
sqr_avg |
sqr average |
... |
additional args to pass |
Value
None
Awd_lstm_clas_split
Description
Split a RNN 'model' in groups for differential learning rates.
Usage
awd_lstm_clas_split(model)
Arguments
model |
model |
Value
None
Awd_lstm_lm_split
Description
Split a RNN 'model' in groups for differential learning rates.
Usage
awd_lstm_lm_split(model)
Arguments
model |
model |
Value
None
Basic critic
Description
A basic critic for images 'n_channels' x 'in_size' x 'in_size'.
Usage
basic_critic(in_size, n_channels, ...)
Arguments
in_size |
input size |
n_channels |
The number of channels |
... |
additional parameters to pass |
Value
None
Examples
## Not run:
critic = basic_critic(in_size = 64, n_channels = 3, n_extra_layers = 1,
act_cls = partial(nn()$LeakyReLU, negative_slope = 0.2))
## End(Not run)
Basic generator
Description
A basic generator from 'in_sz' to images 'n_channels' x 'out_size' x 'out_size'.
Usage
basic_generator(out_size, n_channels, ...)
Arguments
out_size |
out_size |
n_channels |
n_channels |
... |
additional params to pass |
Value
generator object
Examples
## Not run:
generator = basic_generator(out_size = 64, n_channels = 3, n_extra_layers = 1)
## End(Not run)
Bb_pad
Description
Function that collect 'samples' of labelled bboxes and adds padding with 'pad_idx'.
Usage
bb_pad(samples, pad_idx = 0)
Arguments
samples |
samples |
pad_idx |
pad index |
Value
None
Hugging Face module
Description
Hugging Face module
Blurr module
Usage
blurr()
blurr()
Value
None
None
Bs_find
Description
Launch a mock training to find a good batch size to minimize training time.
Usage
bs_find(
object,
lr,
num_it = NULL,
n_batch = 5,
simulate_multi_gpus = TRUE,
show_plot = TRUE
)
Arguments
object |
model/learner |
lr |
learning rate |
num_it |
number of iterations |
n_batch |
number of batches |
simulate_multi_gpus |
simulate on multi gpus or not |
show_plot |
show plot or not |
Details
However, it may not be a good batch size to minimize the validation loss. A good batch size is where the Simple Noise Scale converge ignoring the small growing trend with the number of iterations if exists. The optimal batch size is about an order the magnitud where Simple Noise scale converge. Typically, the optimal batch size in image classification problems will be 2-3 times lower where
Bs finder
Description
Bs finder
Usage
bs_finder()
Value
None
Builtins module
Description
Builtins module
Usage
bt()
Value
None
Calculate_rouge
Description
Calculate_rouge
Usage
calculate_rouge(
predicted_txts,
reference_txts,
rouge_keys = c("rouge1", "rouge2", "rougeL"),
use_stemmer = TRUE
)
Arguments
predicted_txts |
predicted texts |
reference_txts |
reference texts |
rouge_keys |
rouge keys |
use_stemmer |
use stemmer or not |
Value
None
Catalyst module
Description
Catalyst module
Usage
catalyst()
Value
None
Catalyst model
Description
Catalyst model
Usage
catalyst_model()
Value
model
Ceil
Description
Ceil
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
ceiling(x)
Arguments
x |
tensor |
Value
tensor
Ceil
Description
Ceil
Usage
## S3 method for class 'torch.Tensor'
ceiling(x)
Arguments
x |
tensor |
Value
tensor
Children_and_parameters
Description
Return the children of 'm' and its direct parameters not registered in modules.
Usage
children_and_parameters(m)
Arguments
m |
parameters |
Value
None
Clean_raw_keys
Description
Clean_raw_keys
Usage
clean_raw_keys(wgts)
Arguments
wgts |
wgts |
Value
None
Clip_remove_empty
Description
Clip bounding boxes with image border and label background the empty ones
Usage
clip_remove_empty(bbox, label)
Arguments
bbox |
bbox |
label |
label |
Value
None
Cm module
Description
Cm module
Usage
cm()
Value
None
Cnn config
Description
Convenience function to easily create a config for 'create_cnn_model'
Usage
cnn_config(
cut = NULL,
pretrained = TRUE,
n_in = 3,
init = nn()$init$kaiming_normal_,
custom_head = NULL,
concat_pool = TRUE,
lin_ftrs = NULL,
ps = 0.5,
bn_final = FALSE,
lin_first = FALSE,
y_range = NULL
)
Arguments
cut |
cut |
pretrained |
pre-trained or not |
n_in |
input shape |
init |
initializer |
custom_head |
custom head |
concat_pool |
concatenate pooling |
lin_ftrs |
linear filters |
ps |
parameter server |
bn_final |
batch normalization final |
lin_first |
linear first |
y_range |
y_range |
Value
None
Cnn_learner
Description
Build a convnet style learner from 'dls' and 'arch'
Usage
cnn_learner(
dls,
arch,
loss_func = NULL,
pretrained = TRUE,
cut = NULL,
splitter = NULL,
y_range = NULL,
config = NULL,
n_out = NULL,
normalize = TRUE,
opt_func = Adam(),
lr = 0.001,
cbs = NULL,
metrics = NULL,
path = NULL,
model_dir = "models",
wd = NULL,
wd_bn_bias = FALSE,
train_bn = TRUE,
moms = list(0.95, 0.85, 0.95)
)
Arguments
dls |
data loader object |
arch |
a model architecture |
loss_func |
loss function |
pretrained |
pre-trained or not |
cut |
cut |
splitter |
It is a function that takes self.model and returns a list of parameter groups (or just one parameter group if there are no different parameter groups). |
y_range |
y_range |
config |
configuration |
n_out |
the number of out |
normalize |
normalize |
opt_func |
The function used to create the optimizer |
lr |
learning rate |
cbs |
Cbs is one or a list of Callbacks to pass to the Learner. |
metrics |
It is an optional list of metrics, that can be either functions or Metrics. |
path |
The folder where to work |
model_dir |
Path and model_dir are used to save and/or load models. |
wd |
It is the default weight decay used when training the model. |
wd_bn_bias |
It controls if weight decay is applied to BatchNorm layers and bias. |
train_bn |
It controls if BatchNorm layers are trained even when they are supposed to be frozen according to the splitter. |
moms |
The default momentums used in Learner.fit_one_cycle. |
Value
learner object
Examples
## Not run:
URLs_MNIST_SAMPLE()
# transformations
tfms = aug_transforms(do_flip = FALSE)
path = 'mnist_sample'
bs = 20
#load into memory
data = ImageDataLoaders_from_folder(path, batch_tfms = tfms, size = 26, bs = bs)
learn = cnn_learner(data, resnet18(), metrics = accuracy, path = getwd())
## End(Not run)
Collab module
Description
Collab module
Usage
collab()
Value
None
Collab_learner
Description
Create a Learner for collaborative filtering on 'dls'.
Usage
collab_learner(
dls,
n_factors = 50,
use_nn = FALSE,
emb_szs = NULL,
layers = NULL,
config = NULL,
y_range = NULL,
loss_func = NULL,
opt_func = Adam(),
lr = 0.001,
splitter = trainable_params(),
cbs = NULL,
metrics = NULL,
path = NULL,
model_dir = "models",
wd = NULL,
wd_bn_bias = FALSE,
train_bn = TRUE,
moms = list(0.95, 0.85, 0.95)
)
Arguments
dls |
a data loader object |
n_factors |
The number of factors |
use_nn |
use_nn |
emb_szs |
embedding size |
layers |
list of layers |
config |
configuration |
y_range |
y_range |
loss_func |
It can be any loss function you like. It needs to be one of fastai's if you want to use Learn.predict or Learn.get_preds, or you will have to implement special methods (see more details after the BaseLoss documentation). |
opt_func |
The function used to create the optimizer |
lr |
learning rate |
splitter |
It is a function that takes self.model and returns a list of parameter groups (or just one parameter group if there are no different parameter groups). |
cbs |
Cbs is one or a list of Callbacks to pass to the Learner. |
metrics |
It is an optional list of metrics, that can be either functions or Metrics. |
path |
The folder where to work |
model_dir |
Path and model_dir are used to save and/or load models. |
wd |
It is the default weight decay used when training the model. |
wd_bn_bias |
It controls if weight decay is applied to BatchNorm layers and bias. |
train_bn |
It controls if BatchNorm layers are trained even when they are supposed to be frozen according to the splitter. |
moms |
The default momentums used in Learner.fit_one_cycle. |
Value
learner object
Examples
## Not run:
URLs_MOVIE_LENS_ML_100k()
c(user,item,title) %<-% list('userId','movieId','title')
ratings = fread('ml-100k/u.data', col.names = c(user,item,'rating','timestamp'))
movies = fread('ml-100k/u.item', col.names = c(item, 'title', 'date', 'N', 'url',
paste('g',1:19,sep = '')))
rating_movie = ratings[movies[, .SD, .SDcols=c(item,title)], on = item]
dls = CollabDataLoaders_from_df(rating_movie, seed = 42, valid_pct = 0.1, bs = 64,
item_name=title, path='ml-100k')
learn = collab_learner(dls, n_factors = 40, y_range=c(0, 5.5))
learn %>% fit_one_cycle(1, 5e-3, wd = 1e-1)
## End(Not run)
Colors module
Description
Colors module
Usage
colors()
Value
None
Combined_flat_anneal
Description
Create a schedule with constant learning rate 'start_lr' for 'pct' proportion of the training, and a 'curve_type' learning rate (till 'end_lr') for remaining portion of training.
Usage
combined_flat_anneal(pct, start_lr, end_lr = 0, curve_type = "linear")
Arguments
pct |
Proportion of training with a constant learning rate. |
start_lr |
Desired starting learning rate, used for beginnning pct of training. |
end_lr |
Desired end learning rate, training will conclude at this learning rate. |
curve_type |
Curve type for learning rate annealing. Options are 'linear', 'cosine', and 'exponential'. |
Competition download file
Description
download a competition file to a designated location, or use
Usage
competition_download_file(
competition,
file_name,
path = NULL,
force = FALSE,
quiet = FALSE
)
Arguments
competition |
the name of the competition |
file_name |
the configuration file name |
path |
a path to download the file to |
force |
force the download if the file already exists (default FALSE) |
quiet |
suppress verbose output (default is FALSE) |
Value
None
Examples
## Not run:
com_nm = 'titanic'
titanic_files = competition_list_files(com_nm)
titanic_files = lapply(1:length(titanic_files),
function(x) as.character(titanic_files[[x]]))
str(titanic_files)
if(!dir.exists(com_nm)) {
dir.create(com_nm)
}
# download via api
competition_download_files(competition = com_nm, path = com_nm, unzip = TRUE)
## End(Not run)
Competition download files
Description
Competition download files
Usage
competition_download_files(
competition,
path = NULL,
force = FALSE,
quiet = FALSE,
unzip = FALSE
)
Arguments
competition |
the name of the competition |
path |
a path to download the file to |
force |
force the download if the file already exists (default FALSE) |
quiet |
suppress verbose output (default is TRUE) |
unzip |
unzip downloaded files |
Value
None
Competition leaderboard download
Description
Download competition leaderboards
Usage
competition_leaderboard_download(competition, path, quiet = TRUE)
Arguments
competition |
the name of the competition |
path |
a path to download the file to |
quiet |
suppress verbose output (default is TRUE) |
Value
data frame
Competition list files
Description
list files for competition
Usage
competition_list_files(competition)
Arguments
competition |
the name of the competition |
Value
list of files
Examples
## Not run:
com_nm = 'titanic'
titanic_files = competition_list_files(com_nm)
## End(Not run)
Competition submit
Description
Competition submit
Usage
competition_submit(file_name, message, competition, quiet = FALSE)
Arguments
file_name |
the competition metadata file |
message |
the submission description |
competition |
the competition name |
quiet |
suppress verbose output (default is FALSE) |
Value
None
Competitions list
Description
Competitions list
Usage
competitions_list(
group = NULL,
category = NULL,
sort_by = NULL,
page = 1,
search = NULL
)
Arguments
group |
group to filter result to |
category |
category to filter result to |
sort_by |
how to sort the result, see valid_competition_sort_by for options |
page |
the page to return (default is 1) |
search |
a search term to use (default is empty string) |
Value
list of competitions
ConvT_norm_relu
Description
ConvT_norm_relu
Usage
convT_norm_relu(ch_in, ch_out, norm_layer, ks = 3, stride = 2, bias = TRUE)
Arguments
ch_in |
input |
ch_out |
output |
norm_layer |
normalziation layer |
ks |
kernel size |
stride |
stride size |
bias |
bias true or not |
Value
None
Conv_norm_lr
Description
Conv_norm_lr
Usage
conv_norm_lr(
ch_in,
ch_out,
norm_layer = NULL,
ks = 3,
bias = TRUE,
pad = 1,
stride = 1,
activ = TRUE,
slope = 0.2,
init = nn()$init$normal_,
init_gain = 0.02
)
Arguments
ch_in |
input |
ch_out |
output |
norm_layer |
normalziation layer |
ks |
kernel size |
bias |
bias |
pad |
pad |
stride |
stride |
activ |
activation |
slope |
slope |
init |
inititializer |
init_gain |
initializer gain |
Value
None
Cos
Description
Cos
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
cos(x)
Arguments
x |
tensor |
Value
tensor
Cos
Description
Cos
Usage
## S3 method for class 'torch.Tensor'
cos(x)
Arguments
x |
tensor |
Value
tensor
Cosh
Description
Cosh
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
cosh(x)
Arguments
x |
tensor |
Value
tensor
Cosh
Description
Cosh
Usage
## S3 method for class 'torch.Tensor'
cosh(x)
Arguments
x |
tensor |
Value
tensor
Crappify module
Description
Crappify module
Usage
crap()
Value
None
Crappifier
Description
Crappifier
Usage
crappifier(path_lr, path_hr)
Arguments
path_lr |
path from (origin) |
path_hr |
path to (destination) |
Value
None
Examples
## Not run:
items = get_image_files(path_hr)
parallel(crappifier(path_lr, path_hr), items)
## End(Not run)
Create_body
Description
Cut off the body of a typically pretrained 'arch' as determined by 'cut'
Usage
create_body(...)
Arguments
... |
parameters to pass |
Value
None
Examples
## Not run:
encoder = create_body(resnet34(), pretrained = TRUE)
## End(Not run)
Create_cnn_model
Description
Create custom convnet architecture using 'arch', 'n_in' and 'n_out'
Usage
create_cnn_model(
arch,
n_out,
cut = NULL,
pretrained = TRUE,
n_in = 3,
init = nn()$init$kaiming_normal_,
custom_head = NULL,
concat_pool = TRUE,
lin_ftrs = NULL,
ps = 0.5,
bn_final = FALSE,
lin_first = FALSE,
y_range = NULL
)
Arguments
arch |
a model architecture |
n_out |
number of outs |
cut |
cut |
pretrained |
pretrained model or not |
n_in |
input shape |
init |
initializer |
custom_head |
custom head |
concat_pool |
concatenate pooling |
lin_ftrs |
linear fiters |
ps |
parameter server |
bn_final |
batch normalization final |
lin_first |
linear first |
y_range |
y_range |
Value
None
Create_fcn
Description
A bunch of convolutions stacked together.
Usage
create_fcn(ni, nout, ks = 9, conv_sizes = c(128, 256, 128), stride = 1)
Arguments
ni |
number of input channels |
nout |
output shape |
ks |
kernel size |
conv_sizes |
convolution sizes |
stride |
stride |
Value
model
Create_head
Description
Model head that takes 'nf' features, runs through 'lin_ftrs', and out 'n_out' classes.
Usage
create_head(
nf,
n_out,
lin_ftrs = NULL,
ps = 0.5,
concat_pool = TRUE,
bn_final = FALSE,
lin_first = FALSE,
y_range = NULL
)
Arguments
nf |
number of features |
n_out |
number of out features |
lin_ftrs |
linear features |
ps |
parameter server |
concat_pool |
concatenate pooling |
bn_final |
batch normalization final |
lin_first |
linear first |
y_range |
y_range |
Value
None
Create_inception
Description
Creates an InceptionTime arch from 'ni' channels to 'nout' outputs.
Usage
create_inception(
ni,
nout,
kss = c(39, 19, 9),
depth = 6,
bottleneck_size = 32,
nb_filters = 32,
head = TRUE
)
Arguments
ni |
number of input channels |
nout |
number of outputs, should be equal to the number of classes for classification tasks. |
kss |
kernel sizes for the inception Block. |
depth |
depth |
bottleneck_size |
The number of channels on the convolution bottleneck. |
nb_filters |
Channels on the convolution of each kernel. |
head |
TRUE if we want a head attached. |
Value
model
Create_mlp
Description
A simple model builder to create a bunch of BatchNorm1d, Dropout and Linear layers, with “'act_fn“' activations.
Usage
create_mlp(ni, nout, linear_sizes = c(500, 500, 500))
Arguments
ni |
number of input channels |
nout |
output shape |
linear_sizes |
linear output sizes |
Value
model
Create_resnet
Description
Basic 11 Layer - 1D resnet builder
Usage
create_resnet(
ni,
nout,
kss = c(9, 5, 3),
conv_sizes = c(64, 128, 128),
stride = 1
)
Arguments
ni |
number of input channels |
nout |
output shape |
kss |
kernel size |
conv_sizes |
convolution sizes |
stride |
stride |
Value
model
Create_unet_model
Description
Create custom unet architecture
Usage
create_unet_model(
arch,
n_out,
img_size,
pretrained = TRUE,
cut = NULL,
n_in = 3,
blur = FALSE,
blur_final = TRUE,
self_attention = FALSE,
y_range = NULL,
last_cross = TRUE,
bottle = FALSE,
act_cls = nn()$ReLU,
init = nn()$init$kaiming_normal_,
norm_type = NULL
)
Arguments
arch |
architecture |
n_out |
number of out features |
img_size |
imgage shape |
pretrained |
pretrained or not |
cut |
cut |
n_in |
number of input |
blur |
blur is used to avoid checkerboard artifacts at each layer. |
blur_final |
blur final is specific to the last layer. |
self_attention |
self_attention determines if we use a self attention layer at the third block before the end. |
y_range |
If y_range is passed, the last activations go through a sigmoid rescaled to that range. |
last_cross |
last_cross |
bottle |
bottle |
act_cls |
activation |
init |
initialzier |
norm_type |
normalization type |
Value
None
Loss NN module
Description
Loss NN module
Usage
custom_loss()
Value
None
Cutout_gaussian
Description
Replace all 'areas' in 'x' with N(0,1) noise
Usage
cutout_gaussian(x, areas)
Arguments
x |
tensor |
areas |
areas |
Value
None
Cycle_learner
Description
Initialize and return a 'Learner' object with the data in 'dls', CycleGAN model 'm', optimizer function 'opt_func', metrics 'metrics',
Usage
cycle_learner(
dls,
m,
opt_func = Adam(),
show_imgs = TRUE,
imgA = TRUE,
imgB = TRUE,
show_img_interval = 10,
...
)
Arguments
dls |
dataloader |
m |
CycleGAN model |
opt_func |
optimizer |
show_imgs |
show images |
imgA |
image a (from) |
imgB |
image B (to) |
show_img_interval |
show images interval rafe |
... |
additional arguments |
Details
and callbacks 'cbs'. Additionally, if 'show_imgs' is TRUE, it will show intermediate predictions during training. It will show domain B-to-A predictions if 'imgA' is TRUE and/or domain A-to-B predictions if 'imgB' is TRUE. Additionally, it will show images every 'show_img_interval' epochs. ' Other 'Learner' arguments can be passed as well.
Value
None
Dataloaders from dls object
Description
Create a 'DataLoaders' object from 'source'
Usage
dataloaders(object, ...)
Arguments
object |
model |
... |
additional parameters to pass |
Examples
## Not run:
dls = TabularDataTable(df, procs, cat_names, cont_names,
y_names = dep_var, splits = list(tr_idx, ts_idx) ) %>%
dataloaders(bs = 50)
## End(Not run)
Read dicom
Description
Open a 'DICOM' file
Usage
dcmread(fn, force = FALSE)
Arguments
fn |
file name |
force |
logical, force |
Value
dicom object
Examples
## Not run:
img = dcmread('hemorrhage.dcm')
## End(Not run)
Debias
Description
Debias
Usage
debias(mom, damp, step)
Arguments
mom |
mom |
damp |
damp |
step |
step |
Value
None
Decision_plot
Description
Visualizes a model's decisions using cumulative SHAP values.
Usage
decision_plot(object, class_id = 0, row_idx = -1, dpi = 200, ...)
Arguments
object |
ShapInterpretation object |
class_id |
is used to indicate the class of interest for a classification model. It can either be an int or str representation for a class of choice. Each colored line in the plot represents the model's prediction for a single observation. |
row_idx |
If no index is passed in to use from the data, it will default to the first ten samples on the test set. Note:plotting too many samples at once can make the plot illegible. |
dpi |
dots per inch |
... |
additional arguments |
Value
None
Decode_spec_tokens
Description
Decode the special tokens in 'tokens'
Usage
decode_spec_tokens(tokens)
Arguments
tokens |
tokens |
Value
None
Default_split
Description
Default split of a model between body and head
Usage
default_split(m)
Arguments
m |
parameters |
Value
None
Denormalize_imagenet
Description
Denormalize_imagenet
Usage
denormalize_imagenet(img)
Arguments
img |
img |
Value
None
Densenet121
Description
Densenet121
Usage
densenet121(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>
Value
model
Densenet161
Description
Densenet161
Usage
densenet161(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>
Value
model
Densenet169
Description
Densenet169
Usage
densenet169(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>
Value
model
Densenet201
Description
Densenet201
Usage
densenet201(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
"Densely Connected Convolutional Networks" <https://arxiv.org/pdf/1608.06993.pdf>
Value
model
Dependence_plot
Description
Plots the value of a variable on the x-axis and the SHAP value of the same variable on the y-axis. Accepts a class_id and variable_name.
Usage
dependence_plot(object, variable_name = "", class_id = 0, dpi = 200, ...)
Arguments
object |
ShapInterpretation object |
variable_name |
the name of the column |
class_id |
is used to indicate the class of interest for a classification model. It can either be an int or str representation for a class of choice. This plot shows how the model depends on the given variable. Vertical dispersion of the datapoints represent interaction effects. Gray ticks along the y-axis are datapoints where the variable's values were NaN. |
dpi |
dots per inch |
... |
additional arguments |
Value
None
Detuplify_pg
Description
Detuplify_pg
Usage
detuplify_pg(d)
Arguments
d |
d |
Value
None
Dicom_windows module
Description
Dicom_windows module
Usage
dicom_windows()
Value
None
Dihedral_mat
Description
Return a random dihedral matrix
Usage
dihedral_mat(x, p = 0.5, draw = NULL, batch = FALSE)
Arguments
x |
tensor |
p |
probability |
draw |
draw |
batch |
batch |
Value
None
Dim
Description
Dim
Usage
## S3 method for class 'torch.Tensor'
dim(x)
Arguments
x |
tensor |
Value
tensor
Dim
Description
Dim
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
dim(x)
Arguments
x |
tensor |
Value
tensor
Discriminator
Description
Discriminator
Usage
discriminator(
ch_in,
n_ftrs = 64,
n_layers = 3,
norm_layer = NULL,
sigmoid = FALSE
)
Arguments
ch_in |
input |
n_ftrs |
number of filters |
n_layers |
number of layers |
norm_layer |
normalization layer |
sigmoid |
apply sigmoid function or not |
Div
Description
Div
Usage
## S3 method for class 'torch.Tensor'
a / b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Dropout_mask
Description
Return a dropout mask of the same type as 'x', size 'sz', with probability 'p' to cancel an element.
Usage
dropout_mask(x, sz, p)
Arguments
x |
x |
sz |
sz |
p |
p |
Value
None
Dummy_eval
Description
Evaluate 'm' on a dummy input of a certain 'size'
Usage
dummy_eval(m, size = list(64, 64))
Arguments
m |
m parameter |
size |
size |
Value
None
Efficientdet infer dataloader
Description
A 'DataLoader' with a custom 'collate_fn' that batches items as required for inferring the model.
Usage
efficientdet_infer_dl(dataset, batch_tfms = NULL, ...)
Arguments
dataset |
Possibly a 'Dataset' object, but more generally, any 'Sequence' that returns records. |
batch_tfms |
Transforms to be applied at the batch level. **dataloader_kwargs: Keyword arguments that will be internally passed to a Pytorch 'DataLoader'. The parameter 'collate_fn' is already defined internally and cannot be passed here. |
... |
additional arguments |
Value
None
MaskRCNN learner
Description
Fastai 'Learner' adapted for MaskRCNN.
Usage
efficientdet_learner(dls, model, cbs = NULL, ...)
Arguments
dls |
'Sequence' of 'DataLoaders' passed to the 'Learner'. The first one will be used for training and the second for validation. |
model |
The model to train. |
cbs |
Optional 'Sequence' of callbacks. |
... |
learner_kwargs: Keyword arguments that will be internally passed to 'Learner'. |
Value
model
Eficientdet model
Description
Creates the efficientdet model specified by 'model_name'.
Usage
efficientdet_model(model_name, num_classes, img_size, pretrained = TRUE)
Arguments
model_name |
Specifies the model to create. For pretrained models, check [this](https://github.com/rwightman/efficientdet-pytorch#models) table. |
num_classes |
Number of classes of your dataset (including background). |
img_size |
Image size that will be fed to the model. Must be squared and divisible by 128. |
pretrained |
If TRUE, use a pretrained backbone (on COCO). |
Value
model
Efficientdet predict dataloader
Description
Efficientdet predict dataloader
Usage
efficientdet_predict_dl(model, infer_dl, show_pbar = TRUE)
Arguments
model |
model |
infer_dl |
infer_dl |
show_pbar |
show_pbar |
Value
None
Efficientdet train dataloader
Description
A 'DataLoader' with a custom 'collate_fn' that batches items as required for training the model.
Usage
efficientdet_train_dl(dataset, batch_tfms = NULL, ...)
Arguments
dataset |
Possibly a 'Dataset' object, but more generally, any 'Sequence' that returns records. |
batch_tfms |
Transforms to be applied at the batch level. |
... |
dataloader_kwargs: Keyword arguments that will be internally passed to a Pytorch 'DataLoader'. The parameter 'collate_fn' is already defined internally and cannot be passed here. |
Value
None
Efficientdet valid dataloader
Description
A 'DataLoader' with a custom 'collate_fn' that batches items as required for training the model.
Usage
efficientdet_valid_dl(dataset, batch_tfms = NULL, ...)
Arguments
dataset |
Possibly a 'Dataset' object, but more generally, any 'Sequence' that returns records. |
batch_tfms |
Transforms to be applied at the batch level. |
... |
dataloader_kwargs: Keyword arguments that will be internally passed to a Pytorch 'DataLoader'. The parameter 'collate_fn' is already defined internally and cannot be passed here. |
Value
None
Emb_sz_rule
Description
Rule of thumb to pick embedding size corresponding to 'n_cat'
Usage
emb_sz_rule(n_cat)
Arguments
n_cat |
n_cat |
Value
None
Error rate
Description
1 - 'accuracy'
Usage
error_rate(inp, targ, axis = -1)
Arguments
inp |
The predictions of the model |
targ |
The corresponding labels |
axis |
Axis |
Value
tensor
Examples
## Not run:
learn = cnn_learner(dls, resnet34(), metrics = error_rate)
## End(Not run)
Exp
Description
Exp
Usage
## S3 method for class 'torch.Tensor'
exp(x)
Arguments
x |
tensor |
Value
tensor
Exp
Description
Exp
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
exp(x)
Arguments
x |
tensor |
Value
tensor
Exp_rmspe
Description
Root mean square percentage error of the exponential of predictions and targets
Usage
exp_rmspe(preds, targs)
Arguments
preds |
predicitons |
targs |
targets |
Value
None
Expm1
Description
Expm1
Usage
## S3 method for class 'torch.Tensor'
expm1(x)
Arguments
x |
tensor |
Value
tensor
Expm1
Description
Expm1
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
expm1(x)
Arguments
x |
tensor |
Value
tensor
Export_generator
Description
Export_generator
Usage
export_generator(
learn,
generator_name = "generator",
path = ".",
convert_to = "B"
)
Arguments
learn |
learner/model |
generator_name |
generator name |
path |
path (save dir) |
convert_to |
convert to |
Value
None
View
Description
Reshape x to size
Usage
fView(...)
Arguments
... |
parameters to pass |
Value
None
Fa_collate
Description
Fa_collate
Usage
fa_collate(t)
Arguments
t |
text |
Value
None
Da_convert
Description
Da_convert
Usage
fa_convert(t)
Arguments
t |
text |
Value
None
Fastai version
Description
Fastai version
Usage
fastai_version()
Value
None
Fastaudio module
Description
Fastaudio module
Usage
fastaudio()
Value
None
Faster RCNN infer dataloader
Description
A 'DataLoader' with a custom 'collate_fn' that batches items as required for inferring the model.
Usage
faster_rcnn_infer_dl(dataset, batch_tfms = NULL, ...)
Arguments
dataset |
Possibly a 'Dataset' object, but more generally, any 'Sequence' that returns records. |
batch_tfms |
Transforms to be applied at the batch level. **dataloader_kwargs: Keyword arguments that will be internally passed to a Pytorch 'DataLoader'. The parameter 'collate_fn' is already defined internally and cannot be passed here. |
... |
additional arguments |
Value
None
Faster RSNN learner
Description
Fastai 'Learner' adapted for Faster RCNN.
Usage
faster_rcnn_learner(dls, model, cbs = NULL, ...)
Arguments
dls |
'Sequence' of 'DataLoaders' passed to the 'Learner'. The first one will be used for training and the second for validation. |
model |
The model to train. |
cbs |
Optional 'Sequence' of callbacks. |
... |
learner_kwargs: Keyword arguments that will be internally passed to 'Learner'. |
Value
model
Faster RSNN model
Description
FasterRCNN model implemented by torchvision.
Usage
faster_rcnn_model(
num_classes,
backbone = NULL,
remove_internal_transforms = TRUE,
pretrained = TRUE
)
Arguments
num_classes |
Number of classes. |
backbone |
Backbone model to use. Defaults to a resnet50_fpn model. |
remove_internal_transforms |
The torchvision model internally applies transforms like resizing and normalization, but we already do this at the ‘Dataset' level, so it’s safe to remove those internal transforms. |
pretrained |
Argument passed to 'fastercnn_resnet50_fpn' if 'backbone is NULL'. By default it is set to TRUE: this is generally used when training a new model (transfer learning). 'pretrained = FALSE' is used during inference (prediction) for cases where the users have their own pretrained weights. **faster_rcnn_kwargs: Keyword arguments that internally are going to be passed to 'torchvision.models.detection.faster_rcnn.FastRCNN'. |
Value
model
Faster RCNN predict dataloader
Description
Faster RCNN predict dataloader
Usage
faster_rcnn_predict_dl(model, infer_dl, show_pbar = TRUE)
Arguments
model |
model |
infer_dl |
infer_dl |
show_pbar |
show_pbar |
Value
None
Faster RSNN train dataloader
Description
A 'DataLoader' with a custom 'collate_fn' that batches items as required for training the model.
Usage
faster_rcnn_train_dl(dataset, batch_tfms = NULL, ...)
Arguments
dataset |
Possibly a 'Dataset' object, but more generally, any 'Sequence' that returns records. |
batch_tfms |
Transforms to be applied at the batch level. |
... |
dataloader_kwargs: Keyword arguments that will be internally passed to a Pytorch 'DataLoader'. The parameter 'collate_fn' is already defined internally and cannot be passed here. |
Value
None
Faster RSNN valid dataloader
Description
A 'DataLoader' with a custom 'collate_fn' that batches items as required for training the model.
Usage
faster_rcnn_valid_dl(dataset, batch_tfms = NULL, ...)
Arguments
dataset |
Possibly a 'Dataset' object, but more generally, any 'Sequence' that returns records. |
batch_tfms |
Transforms to be applied at the batch level. |
... |
dataloader_kwargs: Keyword arguments that will be internally passed to a Pytorch 'DataLoader'. The parameter 'collate_fn' is already defined internally and cannot be passed here. |
Value
None
Wandb module
Description
Wandb module
Usage
fastinf()
Value
None
Find_coeffs
Description
Find coefficients for warp tfm from 'p1' to 'p2'
Usage
find_coeffs(p1, p2)
Arguments
p1 |
coefficient p1 |
p2 |
coefficient p2 |
Value
None
Fine_tune
Description
Fine tune with 'freeze' for 'freeze_epochs' then with 'unfreeze' from 'epochs' using discriminative LR
Usage
fine_tune(
object,
epochs,
base_lr = 0.002,
freeze_epochs = 1,
lr_mult = 100,
pct_start = 0.3,
div = 5,
...
)
Arguments
object |
learner/model |
epochs |
epoch number |
base_lr |
base learning rate |
freeze_epochs |
freeze epochs number |
lr_mult |
learning rate multiply |
pct_start |
start percentage |
div |
divide |
... |
additional arguments |
Value
None
Fit
Description
Fit the model on this learner with 'lr' learning rate, 'wd' weight decay for 'epochs' with 'callbacks' as cbs argument.
Usage
## S3 method for class 'fastai.learner.Learner'
fit(object, ...)
Arguments
object |
a learner object |
... |
parameters to pass |
Value
train history
Fit
Description
Fit the model on this learner with 'lr' learning rate, 'wd' weight decay for 'epochs' with 'callbacks'.
Usage
## S3 method for class 'fastai.tabular.learner.TabularLearner'
fit(object, ...)
Arguments
object |
model |
... |
additional arguments |
Value
data frame
Fit
Description
Fit the model on this learner with 'lr' learning rate, 'wd' weight decay for 'epochs' with 'callbacks'.
Usage
## S3 method for class 'fastai.vision.gan.GANLearner'
fit(object, ...)
Arguments
object |
model |
... |
additonal parameters to pass |
Value
train history
Examples
## Not run:
learn %>% fit(1, 2e-4, wd = 0)
## End(Not run)
Fit_flat_cos
Description
Fit_flat_cos
Usage
fit_flat_cos(
object,
n_epoch,
lr = NULL,
div_final = 1e+05,
pct_start = 0.75,
wd = NULL,
cbs = NULL,
reset_opt = FALSE
)
Arguments
object |
learner/model |
n_epoch |
number of epochs |
lr |
learning rate |
div_final |
divide final value |
pct_start |
start percentage |
wd |
weight decay |
cbs |
callbacks |
reset_opt |
reset optimizer |
Value
None
Fit_flat_lin
Description
Fit 'self.model' for 'n_epoch' at flat 'start_lr' before 'curve_type' annealing to 'end_lr' with weight decay of 'wd' and callbacks 'cbs'.
Usage
fit_flat_lin(
object,
n_epochs = 100,
n_epochs_decay = 100,
start_lr = NULL,
end_lr = 0,
curve_type = "linear",
wd = NULL,
cbs = NULL,
reset_opt = FALSE
)
Arguments
object |
model / learner |
n_epochs |
number of epochs |
n_epochs_decay |
number of epochs with decay |
start_lr |
Desired starting learning rate, used for beginning pct of training. |
end_lr |
Desired end learning rate, training will conclude at this learning rate. |
curve_type |
Curve type for learning rate annealing. Options are 'linear', 'cosine', and 'exponential'. |
wd |
weight decay |
cbs |
callbacks |
reset_opt |
reset optimizer |
Value
None
Fit one cycle
Description
Fit one cycle
Usage
fit_one_cycle(object, ...)
Arguments
object |
model |
... |
parameters to pass, e.g. lr, n_epoch, wd, and etc. |
Value
None
Fit_sgdr
Description
Fit_sgdr
Usage
fit_sgdr(
object,
n_cycles,
cycle_len,
lr_max = NULL,
cycle_mult = 2,
cbs = NULL,
reset_opt = FALSE,
wd = NULL
)
Arguments
object |
learner/model |
n_cycles |
number of cycles |
cycle_len |
length of cycle |
lr_max |
maximum learning rate |
cycle_mult |
cycle mult |
cbs |
callbacks |
reset_opt |
reset optimizer |
wd |
weight decay |
Value
None
Fix fit
Description
Fix fit
Usage
fix_fit(disable_graph = FALSE)
Arguments
disable_graph |
to remove dynamic plot, by default is FALSE |
Value
None
Fix_html
Description
Various messy things we've seen in documents
Usage
fix_html(x)
Arguments
x |
text |
Value
string
Flatten check
Description
Check that 'out' and 'targ' have the same number of elements and flatten them.
Usage
flatten_check(inp, targ)
Arguments
inp |
predictions |
targ |
targets |
Value
tensor
Flatten_model
Description
Return the list of all submodules and parameters of 'm'
Usage
flatten_model(m)
Arguments
m |
parameters |
Value
None
Flip_mat
Description
Return a random flip matrix
Usage
flip_mat(x, p = 0.5, draw = NULL, batch = FALSE)
Arguments
x |
tensor |
p |
probability |
draw |
draw |
batch |
batch |
Value
None
Tensor to float
Description
Tensor to float
Usage
float(tensor)
Arguments
tensor |
tensor |
Value
tensor
Floor
Description
Floor
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
floor(x)
Arguments
x |
tensor |
Value
tensor
Floor
Description
Floor
Usage
## S3 method for class 'torch.Tensor'
floor(x)
Arguments
x |
tensor |
Value
tensor
Floor divide
Description
Floor divide
Usage
## S3 method for class 'torch.Tensor'
x %/% y
Arguments
x |
tensor |
y |
tensor |
Value
tensor
Floor mod
Description
Floor mod
Usage
## S3 method for class 'torch.Tensor'
x %% y
Arguments
x |
tensor |
y |
tensor |
Value
tensor
Module
Description
Module
Usage
fmodule(...)
Arguments
... |
parameters to pass |
Details
Decorator to create an nn()$Module using f as forward method
Value
None
Force_plot
Description
Visualizes the SHAP values with an added force layout. Accepts a class_id which is used to indicate the class of interest for a classification model.
Usage
force_plot(object, class_id = 0, ...)
Arguments
object |
ShapInterpretation object |
class_id |
Accepts a class_id which is used to indicate the class of interest for a classification model. It can either be an int or str representation for a class of choice. |
... |
additional arguments |
Value
None
Foreground accuracy
Description
Computes non-background accuracy for multiclass segmentation
Usage
foreground_acc(inp, targ, bkg_idx = 0, axis = 1)
Arguments
inp |
predictions |
targ |
targets |
bkg_idx |
bkg_idx |
axis |
axis |
Value
None
Forget_mult_CPU
Description
ForgetMult gate applied to 'x' and 'f' on the CPU.
Usage
forget_mult_CPU(x, f, first_h = NULL, batch_first = TRUE, backward = FALSE)
Arguments
x |
x |
f |
f |
first_h |
first_h |
batch_first |
batch_first |
backward |
backward |
Value
None
Freeze a model
Description
Freeze a model
Usage
freeze(object, ...)
Arguments
object |
A model |
... |
Additional parameters |
Value
None
Examples
## Not run:
learnR %>% freeze()
## End(Not run)
Gan critic
Description
Critic to train a 'GAN'.
Usage
gan_critic(n_channels = 3, nf = 128, n_blocks = 3, p = 0.15)
Arguments
n_channels |
number of channels |
nf |
number of features |
n_blocks |
number of blocks |
p |
probability |
Value
GAN object
GAN loss from function
Description
Define loss functions for a GAN from 'loss_gen' and 'loss_crit'.
Usage
gan_loss_from_func(loss_gen, loss_crit, weights_gen = NULL)
Arguments
loss_gen |
generator loss |
loss_crit |
discriminator loss |
weights_gen |
weight generator |
Value
None
Gauss_blur2d
Description
Apply gaussian_blur2d kornia filter
Usage
gauss_blur2d(x, s)
Arguments
x |
image |
s |
effect |
Value
None
Generate noise
Description
Generate noise
Usage
generate_noise(fn, size = 100)
Arguments
fn |
path |
size |
the size |
Value
None
Examples
## Not run:
generate_noise()
## End(Not run)
Get_annotations
Description
Open a COCO style json in 'fname' and returns the lists of filenames (with maybe 'prefix') and labelled bboxes.
Usage
get_annotations(fname, prefix = NULL)
Arguments
fname |
folder name |
prefix |
prefix |
Value
None
Get_audio_files
Description
Get audio files in 'path' recursively, only in 'folders', if specified.
Usage
get_audio_files(path, recurse = TRUE, folders = NULL)
Arguments
path |
path |
recurse |
recursive or not |
folders |
vector, folders |
Value
None
Get bias
Description
Bias for item or user (based on 'is_item') for all in 'arr'
Usage
get_bias(object, arr, is_item = TRUE, convert = TRUE)
Arguments
object |
extract bias |
arr |
R data frame |
is_item |
logical, is item |
convert |
to R matrix |
Value
tensor
Examples
## Not run:
movie_bias = learn %>% get_bias(top_movies, is_item = TRUE)
## End(Not run)
Get_c
Description
Get_c
Usage
get_c(dls)
Arguments
dls |
dataloader object |
Value
number of layers
Examples
## Not run:
get_c(dls)
## End(Not run)
Extract confusion matrix
Description
Extract confusion matrix
Usage
get_confusion_matrix(object)
Arguments
object |
model |
Value
matrix
Examples
## Not run:
model %>% get_confusion_matrix()
## End(Not run)
Get data loaders
Description
Get data loaders
Usage
get_data_loaders(train_batch_size, val_batch_size)
Arguments
train_batch_size |
train dataset batch size |
val_batch_size |
validation dataset batch size |
Value
None
Get image matrix
Description
Get image matrix
Usage
get_dcm_matrix(img, type = "raw", scan = "", size = 50, convert = TRUE)
Arguments
img |
dicom file |
type |
img transformation |
scan |
apply uniform or gaussian blur effects |
size |
size of image |
convert |
to R matrix or keep tensor |
Value
tensor
Examples
## Not run:
img = dcmread('hemorrhage.dcm')
img %>% get_dcm_matrix(type = 'raw')
## End(Not run)
get_dicom_files
Description
Get dicom files in 'path' recursively, only in 'folders', if specified.
Usage
get_dicom_files(path, recurse = TRUE, folders = NULL)
Arguments
path |
path to files |
recurse |
recursive or not |
folders |
folder names |
Value
lsit of files
Examples
## Not run:
items = get_dicom_files("siim_small/train/")
## End(Not run)
Get dls
Description
Given image files from two domains ('pathA', 'pathB'), create 'DataLoaders' object.
Usage
get_dls(
pathA,
pathB,
num_A = NULL,
num_B = NULL,
load_size = 512,
crop_size = 256,
bs = 4,
num_workers = 2
)
Arguments
pathA |
path A (from domain) |
pathB |
path B (to domain) |
num_A |
subset of A data |
num_B |
subset of B data |
load_size |
load size |
crop_size |
crop size |
bs |
bathc size |
num_workers |
number of workers |
Details
Loading and randomly cropped sizes of 'load_size' and 'crop_size' are set to defaults of 512 and 256. Batch size is specified by 'bs' (default=4).
Value
None
Get_emb_sz
Description
Get default embedding size from 'TabularPreprocessor' 'proc' or the ones in 'sz_dict'
Usage
get_emb_sz(to, sz_dict = NULL)
Arguments
to |
to |
sz_dict |
dictionary size |
Value
None
Get_files
Description
Get all the files in 'path' with optional 'extensions', optionally with 'recurse', only in 'folders', if specified.
Usage
get_files(
path,
extensions = NULL,
recurse = TRUE,
folders = NULL,
followlinks = TRUE
)
Arguments
path |
path |
extensions |
extensions |
recurse |
recurse |
folders |
folders |
followlinks |
followlinks |
Value
list
Get_grid
Description
Return a grid of 'n' axes, 'rows' by 'cols'
Usage
get_grid(
n,
nrows = NULL,
ncols = NULL,
add_vert = 0,
figsize = NULL,
double = FALSE,
title = NULL,
return_fig = FALSE,
imsize = 3
)
Arguments
n |
n |
nrows |
number of rows |
ncols |
number of columns |
add_vert |
add vertical |
figsize |
figure size |
double |
double |
title |
title |
return_fig |
return figure or not |
imsize |
image size |
Value
None
Get_hf_objects
Description
Returns the architecture (str), config (obj), tokenizer (obj), and model (obj) given at minimum a
Usage
get_hf_objects(...)
Arguments
... |
parameters to pass |
Details
'pre-trained model name or path'. Specify a 'task' to ensure the right "AutoModelFor<task>" is used to create the model. Optionally, you can pass a config (obj), tokenizer (class), and/or model (class) (along with any related kwargs for each) to get as specific as you want w/r/t what huggingface objects are returned.
Value
None
Get image files
Description
Get image files in 'path' recursively, only in 'folders', if specified.
Usage
get_image_files(path, recurse = TRUE, folders = NULL)
Arguments
path |
The folder where to work |
recurse |
recursive path |
folders |
folder names |
Value
None
Examples
## Not run:
URLs_PETS()
path = 'oxford-iiit-pet'
path_img = 'oxford-iiit-pet/images'
fnames = get_image_files(path_img)
## End(Not run)
Get_language_model
Description
Create a language model from 'arch' and its 'config'.
Usage
get_language_model(arch, vocab_sz, config = NULL, drop_mult = 1)
Arguments
arch |
arch |
vocab_sz |
vocab_sz |
config |
config |
drop_mult |
drop_mult |
Value
model
Get_preds_cyclegan
Description
A prediction function that takes the Learner object 'learn' with the trained model, the 'test_path' folder with the images to perform
Usage
get_preds_cyclegan(
learn,
test_path,
pred_path,
bs = 4,
num_workers = 4,
suffix = "tif"
)
Arguments
learn |
learner/model |
test_path |
testdat path |
pred_path |
predict data path |
bs |
batch size |
num_workers |
number of workers |
suffix |
suffix |
Details
batch inference on, and the output folder 'pred_path' where the predictions will be saved, with a batch size 'bs', 'num_workers', and suffix of the prediction images ‘suffix' (default=’png').
Get_text_classifier
Description
Create a text classifier from 'arch' and its 'config', maybe 'pretrained'
Usage
get_text_classifier(
arch,
vocab_sz,
n_class,
seq_len = 72,
config = NULL,
drop_mult = 1,
lin_ftrs = NULL,
ps = NULL,
pad_idx = 1,
max_len = 1440,
y_range = NULL
)
Arguments
arch |
arch |
vocab_sz |
vocab_sz |
n_class |
n_class |
seq_len |
seq_len |
config |
config |
drop_mult |
drop_mult |
lin_ftrs |
lin_ftrs |
ps |
ps |
pad_idx |
pad_idx |
max_len |
max_len |
y_range |
y_range |
Value
None
Get_text_files
Description
Get text files in 'path' recursively, only in 'folders', if specified.
Usage
get_text_files(path, recurse = TRUE, folders = NULL)
Arguments
path |
path |
recurse |
recurse |
folders |
folders |
Value
None
Get weights
Description
Weight for item or user (based on 'is_item') for all in 'arr'
Usage
get_weights(object, arr, is_item = TRUE, convert = FALSE)
Arguments
object |
extract weights |
arr |
R data frame |
is_item |
logical, is item |
convert |
to R matrix |
Value
tensor
Examples
## Not run:
movie_w = learn %>% get_weights(top_movies, is_item = TRUE, convert = TRUE)
## End(Not run)
Grayscale
Description
Tensor to grayscale tensor. Uses the ITU-R 601-2 luma transform.
Usage
grayscale(x)
Arguments
x |
tensor |
Value
None
Greater
Description
Greater
Usage
## S3 method for class 'torch.Tensor'
a > b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Greater or equal
Description
Greater or equal
Usage
## S3 method for class 'torch.Tensor'
a >= b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Has_params
Description
Check if 'm' has at least one parameter
Usage
has_params(m)
Arguments
m |
m parameter |
Value
None
Has_pool_type
Description
Return 'TRUE' if 'm' is a pooling layer or has one in its children
Usage
has_pool_type(m)
Arguments
m |
parameters |
Value
None
BLURR_MODEL_HELPER
Description
BLURR_MODEL_HELPER
Usage
helper()
Value
None
Hf_splitter
Description
Splits the huggingface model based on various model architecture conventions
Usage
hf_splitter(m)
Arguments
m |
parameters |
Value
None
Hook_output
Description
Return a 'Hook' that stores activations of 'module' in 'self$stored'
Usage
hook_output(module, detach = TRUE, cpu = FALSE, grad = FALSE)
Arguments
module |
module |
detach |
detach or not |
cpu |
cpu or not |
grad |
grad or not |
Value
None
Hook_outputs
Description
Return 'Hooks' that store activations of all 'modules' in 'self.stored'
Usage
hook_outputs(modules, detach = TRUE, cpu = FALSE, grad = FALSE)
Arguments
modules |
modules |
detach |
detach or not |
cpu |
cpu or not |
grad |
grad or not |
Value
None
Hsv2rgb
Description
Converts a HSV image to an RGB image.
Usage
hsv2rgb(img)
Arguments
img |
image object |
Value
None
Transformer module
Description
Transformer module
Usage
hug()
Value
None
Icevision module
Description
Icevision module
Usage
icevision()
Value
None
Adapter
Description
Adapter that enables the use of albumentations transforms.
Usage
icevision_Adapter(tfms)
Arguments
tfms |
'Sequence' of albumentation transforms. |
Value
None
BasicIAATransform
Description
BasicIAATransform
Usage
icevision_BasicIAATransform(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
BasicTransform
Description
BasicTransform
Usage
icevision_BasicTransform(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
Blur
Description
Blur the input image using a random-sized kernel.
Usage
icevision_Blur(blur_limit = 7, always_apply = FALSE, p = 0.5)
Arguments
blur_limit |
blur_limit |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
CLAHE
Description
Apply Contrast Limited Adaptive Histogram Equalization to the input image.
Usage
icevision_CLAHE(
clip_limit = 4,
tile_grid_size = list(8, 8),
always_apply = FALSE,
p = 0.5
)
Arguments
clip_limit |
clip_limit |
tile_grid_size |
tile_grid_size |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8
ChannelDropout
Description
Randomly Drop Channels in the input Image.
Usage
icevision_ChannelDropout(
channel_drop_range = list(1, 1),
fill_value = 0,
always_apply = FALSE,
p = 0.5
)
Arguments
channel_drop_range |
channel_drop_range |
fill_value |
fill_value |
always_apply |
always_apply |
p |
p |
Targets
image
Image types
uint8, uint16, unit32, float32
ChannelShuffle
Description
Randomly rearrange channels of the input RGB image.
Usage
icevision_ChannelShuffle(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
ClassMap
Description
Utility class for mapping between class name and id.
Usage
icevision_ClassMap(classes, background = 0)
Arguments
classes |
classes |
background |
background |
Value
Python dictionary
CoarseDropout
Description
CoarseDropout of the rectangular regions in the image.
Usage
icevision_CoarseDropout(
max_holes = 8,
max_height = 8,
max_width = 8,
min_holes = NULL,
min_height = NULL,
min_width = NULL,
fill_value = 0,
mask_fill_value = NULL,
always_apply = FALSE,
p = 0.5
)
Arguments
max_holes |
max_holes |
max_height |
max_height |
max_width |
max_width |
min_holes |
min_holes |
min_height |
min_height |
min_width |
min_width |
fill_value |
fill_value |
mask_fill_value |
mask_fill_value |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask
Image types
uint8, float32
Reference
| https://arxiv.org/abs/1708.04552 | https://github.com/uoguelph-mlrg/Cutout/blob/master/util/cutout.py | https://github.com/aleju/imgaug/blob/master/imgaug/augmenters/arithmetic.py
ColorJitter
Description
Randomly changes the brightness, contrast, and saturation of an image. Compared to ColorJitter from torchvision,
Usage
icevision_ColorJitter(
brightness = 0.2,
contrast = 0.2,
saturation = 0.2,
hue = 0.2,
always_apply = FALSE,
p = 0.5
)
Arguments
brightness |
brightness |
contrast |
contrast |
saturation |
saturation |
hue |
hue |
always_apply |
always_apply |
p |
p |
Details
this transform gives a little bit different results because Pillow (used in torchvision) and OpenCV (used in Albumentations) transform an image to HSV format by different formulas. Another difference - Pillow uses uint8 overflow, but we use value saturation.
Value
None
Compose
Description
Compose transforms and handle all transformations regrading bounding boxes
Usage
icevision_Compose(
transforms,
bbox_params = NULL,
keypoint_params = NULL,
additional_targets = NULL,
p = 1
)
Arguments
transforms |
transforms |
bbox_params |
bbox_params |
keypoint_params |
keypoint_params |
additional_targets |
additional_targets |
p |
p |
Value
None
Crop
Description
Crop region from image.
Usage
icevision_Crop(
x_min = 0,
y_min = 0,
x_max = 1024,
y_max = 1024,
always_apply = FALSE,
p = 1
)
Arguments
x_min |
x_min |
y_min |
y_min |
x_max |
x_max |
y_max |
y_max |
always_apply |
always_apply |
p |
p |
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
CropNonEmptyMaskIfExists
Description
Crop area with mask if mask is non-empty, else make random crop.
Usage
icevision_CropNonEmptyMaskIfExists(
height,
width,
ignore_values = NULL,
ignore_channels = NULL,
always_apply = FALSE,
p = 1
)
Arguments
height |
height |
width |
width |
ignore_values |
ignore_values |
ignore_channels |
ignore_channels |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
Cutout
Description
CoarseDropout of the square regions in the image.
Usage
icevision_Cutout(
num_holes = 8,
max_h_size = 8,
max_w_size = 8,
fill_value = 0,
always_apply = FALSE,
p = 0.5
)
Arguments
num_holes |
num_holes |
max_h_size |
max_h_size |
max_w_size |
max_w_size |
fill_value |
fill_value |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
Reference
| https://arxiv.org/abs/1708.04552 | https://github.com/uoguelph-mlrg/Cutout/blob/master/util/cutout.py | https://github.com/aleju/imgaug/blob/master/imgaug/augmenters/arithmetic.py
Dataset
Description
Container for a list of records and transforms.
Usage
icevision_Dataset(records, tfm = NULL)
Arguments
records |
A list of records. |
tfm |
Transforms to be applied to each item. |
Details
Steps each time an item is requested (normally via directly indexing the 'Dataset'): Grab a record from the internal list of records. Prepare the record (open the image, open the mask, add metadata). Apply transforms to the record.
Value
None
Icevision Dataset from images
Description
Creates a 'Dataset' from a list of images.
Usage
icevision_Dataset_from_images(images, tfm = NULL, ...)
Arguments
images |
'Sequence' of images in memory (numpy arrays). |
tfm |
Transforms to be applied to each item. |
... |
additional arguments |
Value
None
Downscale
Description
Decreases image quality by downscaling and upscaling back.
Usage
icevision_Downscale(
scale_min = 0.25,
scale_max = 0.25,
interpolation = 0,
always_apply = FALSE,
p = 0.5
)
Arguments
scale_min |
scale_min |
scale_max |
scale_max |
interpolation |
cv2 interpolation method. cv2.INTER_NEAREST by default |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
DualIAATransform
Description
Transform for segmentation task.
Usage
icevision_DualIAATransform(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
DualTransform
Description
Transform for segmentation task.
Usage
icevision_DualTransform(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
ElasticTransform
Description
Elastic deformation of images as described in [Simard2003]_ (with modifications).
Usage
icevision_ElasticTransform(
alpha = 1,
sigma = 50,
alpha_affine = 50,
interpolation = 1,
border_mode = 4,
value = NULL,
mask_value = NULL,
always_apply = FALSE,
approximate = FALSE,
p = 0.5
)
Arguments
alpha |
alpha |
sigma |
sigma |
alpha_affine |
alpha_affine |
interpolation |
interpolation |
border_mode |
border_mode |
value |
value |
mask_value |
mask_value |
always_apply |
always_apply |
approximate |
approximate |
p |
p |
Details
Based on https://gist.github.com/erniejunior/601cdf56d2b424757de5 .. [Simard2003] Simard, Steinkraus and Platt, "Best Practices for Convolutional Neural Networks applied to Visual Document Analysis", in Proc. of the International Conference on Document Analysis and Recognition, 2003.
Value
None
Targets
image, mask
Image types
uint8, float32
Equalize
Description
Equalize the image histogram.
Usage
icevision_Equalize(mode = "cv", by_channels = TRUE, mask = NULL, ...)
Arguments
mode |
mode |
by_channels |
by_channels |
mask |
mask |
... |
additional arguments |
Value
None
Targets
image
Image types
uint8
FDA
Description
Fourier Domain Adaptation from https://github.com/YanchaoYang/FDA
Usage
icevision_FDA(
reference_images,
beta_limit = 0.1,
read_fn = icevision_read_rgb_image(),
always_apply = FALSE,
p = 0.5
)
Arguments
reference_images |
reference_images |
beta_limit |
beta_limit |
read_fn |
read_fn |
always_apply |
always_apply |
p |
p |
Details
Simple "style transfer".
Value
None
Fourier Domain Adaptation from https
//github.com/YanchaoYang/FDA: Simple "style transfer".
Targets
image
Image types
uint8, float32
Reference
https://github.com/YanchaoYang/FDA https://openaccess.thecvf.com/content_CVPR_2020/papers/Yang_FDA_Fourier_Domain_Adaptation_for_Semantic_Segmentation_CVPR_2020_paper.pdf
Example
>>> import numpy as np >>> import albumentations as A >>> image = np.random.randint(0, 256, [100, 100, 3], dtype=np.uint8) >>> target_image = np.random.randint(0, 256, [100, 100, 3], dtype=np.uint8) >>> aug = A.Compose([A.FDA([target_image], p=1, read_fn=lambda x: x)]) >>> result = aug(image=image)
FancyPCA
Description
Augment RGB image using FancyPCA from Krizhevsky's paper
Usage
icevision_FancyPCA(alpha = 0.1, always_apply = FALSE, p = 0.5)
Arguments
alpha |
alpha |
always_apply |
always_apply |
p |
p |
Details
"ImageNet Classification with Deep Convolutional Neural Networks"
Value
None
Targets
image
Image types
3-channel uint8 images only
Credit
http://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf https://deshanadesai.github.io/notes/Fancy-PCA-with-Scikit-Image https://pixelatedbrian.github.io/2018-04-29-fancy_pca/
FixedSplitter
Description
Split 'ids' based on predefined splits.
Usage
icevision_FixedSplitter(splits)
Arguments
splits |
The predefined splits. |
Value
None
Flip
Description
Flip the input either horizontally, vertically or both horizontally and vertically.
Usage
icevision_Flip(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
FromFloat
Description
Take an input array where all values should lie in the range [0, 1.0], multiply them by 'max_value' and then
Usage
icevision_FromFloat(
dtype = "uint16",
max_value = NULL,
always_apply = FALSE,
p = 1
)
Arguments
dtype |
dtype |
max_value |
max_value |
always_apply |
always_apply |
p |
p |
Details
cast the resulted value to a type specified by 'dtype'. If 'max_value' is NULL the transform will try to infer the maximum value for the data type from the 'dtype' argument. This is the inverse transform for :class:'~albumentations.augmentations.transforms.ToFloat'.
Value
None
Targets
image
Image types
float32
GaussNoise
Description
Apply gaussian noise to the input image.
Usage
icevision_GaussNoise(
var_limit = list(10, 50),
mean = 0,
always_apply = FALSE,
p = 0.5
)
Arguments
var_limit |
var_limit |
mean |
mean |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
GaussianBlur
Description
Blur the input image using a Gaussian filter with a random kernel size.
Usage
icevision_GaussianBlur(
blur_limit = list(3, 7),
sigma_limit = 0,
always_apply = FALSE,
p = 0.5
)
Arguments
blur_limit |
blur_limit |
sigma_limit |
sigma_limit |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
GlassBlur
Description
Apply glass noise to the input image.
Usage
icevision_GlassBlur(
sigma = 0.7,
max_delta = 4,
iterations = 2,
always_apply = FALSE,
mode = "fast",
p = 0.5
)
Arguments
sigma |
sigma |
max_delta |
max_delta |
iterations |
iterations |
always_apply |
always_apply |
mode |
mode |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
Reference
| https://arxiv.org/abs/1903.12261 | https://github.com/hendrycks/robustness/blob/master/ImageNet-C/create_c/make_imagenet_c.py
GridDistortion
Description
Args:
Usage
icevision_GridDistortion(
num_steps = 5,
distort_limit = 0.3,
interpolation = 1,
border_mode = 4,
value = NULL,
mask_value = NULL,
always_apply = FALSE,
p = 0.5
)
Arguments
num_steps |
num_steps |
distort_limit |
distort_limit |
interpolation |
interpolation |
border_mode |
border_mode |
value |
value |
mask_value |
mask_value |
always_apply |
always_apply |
p |
p |
Details
num_steps (int): count of grid cells on each side. distort_limit (float, (float, float)): If distort_limit is a single float, the range will be (-distort_limit, distort_limit). Default: (-0.03, 0.03). interpolation (OpenCV flag): flag that is used to specify the interpolation algorithm. Should be one of: cv2.INTER_NEAREST, cv2.INTER_LINEAR, cv2.INTER_CUBIC, cv2.INTER_AREA, cv2.INTER_LANCZOS4. Default: cv2.INTER_LINEAR. border_mode (OpenCV flag): flag that is used to specify the pixel extrapolation method. Should be one of: cv2.BORDER_CONSTANT, cv2.BORDER_REPLICATE, cv2.BORDER_REFLECT, cv2.BORDER_WRAP, cv2.BORDER_REFLECT_101. Default: cv2.BORDER_REFLECT_101 value (int, float, list of ints, list of float): padding value if border_mode is cv2.BORDER_CONSTANT. mask_value (int, float, list of ints, list of float): padding value if border_mode is cv2.BORDER_CONSTANT applied for masks. Targets: image, mask Image types: uint8, float32
Value
None
Targets
image, mask
Image types
uint8, float32
GridDropout
Description
GridDropout, drops out rectangular regions of an image and the corresponding mask in a grid fashion.
Usage
icevision_GridDropout(
ratio = 0.5,
unit_size_min = NULL,
unit_size_max = NULL,
holes_number_x = NULL,
holes_number_y = NULL,
shift_x = 0,
shift_y = 0,
random_offset = FALSE,
fill_value = 0,
mask_fill_value = NULL,
always_apply = FALSE,
p = 0.5
)
Arguments
ratio |
ratio |
unit_size_min |
unit_size_min |
unit_size_max |
unit_size_max |
holes_number_x |
holes_number_x |
holes_number_y |
holes_number_y |
shift_x |
shift_x |
shift_y |
shift_y |
random_offset |
random_offset |
fill_value |
fill_value |
mask_fill_value |
mask_fill_value |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask
Image types
uint8, float32
References
https://arxiv.org/abs/2001.04086
HistogramMatching
Description
Apply histogram matching. It manipulates the pixels of an input image so that its histogram matches
Usage
icevision_HistogramMatching(
reference_images,
blend_ratio = list(0.5, 1),
read_fn = icevision_read_rgb_image(),
always_apply = FALSE,
p = 0.5
)
Arguments
reference_images |
reference_images |
blend_ratio |
blend_ratio |
read_fn |
read_fn |
always_apply |
always_apply |
p |
p |
Details
the histogram of the reference image. If the images have multiple channels, the matching is done independently for each channel, as long as the number of channels is equal in the input image and the reference. Histogram matching can be used as a lightweight normalisation for image processing, such as feature matching, especially in circumstances where the images have been taken from different sources or in different conditions (i.e. lighting). See: https://scikit-image.org/docs/dev/auto_examples/color_exposure/plot_histogram_matching.html
Value
None
See
https://scikit-image.org/docs/dev/auto_examples/color_exposure/plot_histogram_matching.html
Targets
image
Image types
uint8, uint16, float32
HorizontalFlip
Description
Flip the input horizontally around the y-axis.
Usage
icevision_HorizontalFlip(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
HueSaturationValue
Description
Randomly change hue, saturation and value of the input image.
Usage
icevision_HueSaturationValue(
hue_shift_limit = 20,
sat_shift_limit = 30,
val_shift_limit = 20,
always_apply = FALSE,
p = 0.5
)
Arguments
hue_shift_limit |
hue_shift_limit |
sat_shift_limit |
sat_shift_limit |
val_shift_limit |
val_shift_limit |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
IAAAdditiveGaussianNoise
Description
Add gaussian noise to the input image.
Usage
icevision_IAAAdditiveGaussianNoise(
loc = 0,
scale = list(2.55, 12.75),
per_channel = FALSE,
always_apply = FALSE,
p = 0.5
)
Arguments
loc |
loc |
scale |
scale |
per_channel |
per_channel |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
IAAAffine
Description
Place a regular grid of points on the input and randomly move the neighbourhood of these point around
Usage
icevision_IAAAffine(
scale = 1,
translate_percent = NULL,
translate_px = NULL,
rotate = 0,
shear = 0,
order = 1,
cval = 0,
mode = "reflect",
always_apply = FALSE,
p = 0.5
)
Arguments
scale |
scale |
translate_percent |
translate_percent |
translate_px |
translate_px |
rotate |
rotate |
shear |
shear |
order |
order |
cval |
cval |
mode |
mode |
always_apply |
always_apply |
p |
p |
Details
via affine transformations. Note: This class introduce interpolation artifacts to mask if it has values other than (0;1)
Value
None
None
Targets
image, mask
IAACropAndPad
Description
Transform for segmentation task.
Usage
icevision_IAACropAndPad(
px = NULL,
percent = NULL,
pad_mode = "constant",
pad_cval = 0,
keep_size = TRUE,
always_apply = FALSE,
p = 1
)
Arguments
px |
px |
percent |
percent |
pad_mode |
pad_mode |
pad_cval |
pad_cval |
keep_size |
keep_size |
always_apply |
always_apply |
p |
p |
IAAEmboss
Description
Emboss the input image and overlays the result with the original image.
Usage
icevision_IAAEmboss(
alpha = list(0.2, 0.5),
strength = list(0.2, 0.7),
always_apply = FALSE,
p = 0.5
)
Arguments
alpha |
alpha |
strength |
strength |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
IAAFliplr
Description
Transform for segmentation task.
Usage
icevision_IAAFliplr(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
IAAFlipud
Description
Transform for segmentation task.
Usage
icevision_IAAFlipud(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
IAAPerspective
Description
Perform a random four point perspective transform of the input.
Usage
icevision_IAAPerspective(
scale = list(0.05, 0.1),
keep_size = TRUE,
always_apply = FALSE,
p = 0.5
)
Arguments
scale |
scale |
keep_size |
keep_size |
always_apply |
always_apply |
p |
p |
Details
Note: This class introduce interpolation artifacts to mask if it has values other than (0;1)
Value
None
Targets
image, mask
IAAPiecewiseAffine
Description
Place a regular grid of points on the input and randomly move the neighbourhood of these point around
Usage
icevision_IAAPiecewiseAffine(
scale = list(0.03, 0.05),
nb_rows = 4,
nb_cols = 4,
order = 1,
cval = 0,
mode = "constant",
always_apply = FALSE,
p = 0.5
)
Arguments
scale |
scale |
nb_rows |
nb_rows |
nb_cols |
nb_cols |
order |
order |
cval |
cval |
mode |
mode |
always_apply |
always_apply |
p |
p |
Details
via affine transformations. Note: This class introduce interpolation artifacts to mask if it has values other than (0;1)
Value
None
Targets
image, mask
IAASharpen
Description
Sharpen the input image and overlays the result with the original image.
Usage
icevision_IAASharpen(
alpha = list(0.2, 0.5),
lightness = list(0.5, 1),
always_apply = FALSE,
p = 0.5
)
Arguments
alpha |
alpha |
lightness |
lightness |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
IAASuperpixels
Description
Completely or partially transform the input image to its superpixel representation. Uses skimage's version
Usage
icevision_IAASuperpixels(
p_replace = 0.1,
n_segments = 100,
always_apply = FALSE,
p = 0.5
)
Arguments
p_replace |
p_replace |
n_segments |
n_segments |
always_apply |
always_apply |
p |
p |
Details
of the SLIC algorithm. May be slow.
Value
None
Targets
image
ISONoise
Description
Apply camera sensor noise.
Usage
icevision_ISONoise(
color_shift = list(0.01, 0.05),
intensity = list(0.1, 0.5),
always_apply = FALSE,
p = 0.5
)
Arguments
color_shift |
color_shift |
intensity |
intensity |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8
ImageCompression
Description
Decrease Jpeg, WebP compression of an image.
Usage
icevision_ImageCompression(
quality_lower = 99,
quality_upper = 100,
compression_type = 0,
always_apply = FALSE,
p = 0.5
)
Arguments
quality_lower |
quality_lower |
quality_upper |
quality_upper |
compression_type |
compression_type |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
ImageOnlyIAATransform
Description
Transform applied to image only.
Usage
icevision_ImageOnlyIAATransform(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
ImageOnlyTransform
Description
Transform applied to image only.
Usage
icevision_ImageOnlyTransform(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
InvertImg
Description
Invert the input image by subtracting pixel values from 255.
Usage
icevision_InvertImg(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8
JpegCompression
Description
Decrease Jpeg compression of an image.
Usage
icevision_JpegCompression(
quality_lower = 99,
quality_upper = 100,
always_apply = FALSE,
p = 0.5
)
Arguments
quality_lower |
quality_lower |
quality_upper |
quality_upper |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
LongestMaxSize
Description
Rescale an image so that maximum side is equal to max_size, keeping the aspect ratio of the initial image.
Usage
icevision_LongestMaxSize(
max_size = 1024,
interpolation = 1,
always_apply = FALSE,
p = 1
)
Arguments
max_size |
max_size |
interpolation |
interpolation |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
MaskDropout
Description
Image & mask augmentation that zero out mask and image regions corresponding
Usage
icevision_MaskDropout(
max_objects = 1,
image_fill_value = 0,
mask_fill_value = 0,
always_apply = FALSE,
p = 0.5
)
Arguments
max_objects |
max_objects |
image_fill_value |
image_fill_value |
mask_fill_value |
mask_fill_value |
always_apply |
always_apply |
p |
p |
Details
to randomly chosen object instance from mask. Mask must be single-channel image, zero values treated as background. Image can be any number of channels. Inspired by https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/114254
Value
None
MedianBlur
Description
Blur the input image using a median filter with a random aperture linear size.
Usage
icevision_MedianBlur(blur_limit = 7, always_apply = FALSE, p = 0.5)
Arguments
blur_limit |
blur_limit |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
MotionBlur
Description
Apply motion blur to the input image using a random-sized kernel.
Usage
icevision_MotionBlur(blur_limit = 7, always_apply = FALSE, p = 0.5)
Arguments
blur_limit |
blur_limit |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
MultiplicativeNoise
Description
Multiply image to random number or array of numbers.
Usage
icevision_MultiplicativeNoise(
multiplier = list(0.9, 1.1),
per_channel = FALSE,
elementwise = FALSE,
always_apply = FALSE,
p = 0.5
)
Arguments
multiplier |
multiplier |
per_channel |
per_channel |
elementwise |
elementwise |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
Any
Normalize
Description
Divide pixel values by 255 = 2**8 - 1, subtract mean per channel and divide by std per channel.
Usage
icevision_Normalize(
mean = list(0.485, 0.456, 0.406),
std = list(0.229, 0.224, 0.225),
max_pixel_value = 255,
always_apply = FALSE,
p = 1
)
Arguments
mean |
mean |
std |
std |
max_pixel_value |
max_pixel_value |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
OpticalDistortion
Description
OpticalDistortion
Usage
icevision_OpticalDistortion(
distort_limit = 0.05,
shift_limit = 0.05,
interpolation = 1,
border_mode = 4,
value = NULL,
mask_value = NULL,
always_apply = FALSE,
p = 0.5
)
Arguments
distort_limit |
distort_limit |
shift_limit |
shift_limit |
interpolation |
interpolation |
border_mode |
border_mode |
value |
value |
mask_value |
mask_value |
always_apply |
always_apply |
p |
p |
Details
distort_limit (float, (float, float)): If distort_limit is a single float, the range will be (-distort_limit, distort_limit). Default: (-0.05, 0.05). shift_limit (float, (float, float))): If shift_limit is a single float, the range will be (-shift_limit, shift_limit). Default: (-0.05, 0.05). interpolation (OpenCV flag): flag that is used to specify the interpolation algorithm. Should be one of: cv2.INTER_NEAREST, cv2.INTER_LINEAR, cv2.INTER_CUBIC, cv2.INTER_AREA, cv2.INTER_LANCZOS4. Default: cv2.INTER_LINEAR. border_mode (OpenCV flag): flag that is used to specify the pixel extrapolation method. Should be one of: cv2.BORDER_CONSTANT, cv2.BORDER_REPLICATE, cv2.BORDER_REFLECT, cv2.BORDER_WRAP, cv2.BORDER_REFLECT_101. Default: cv2.BORDER_REFLECT_101 value (int, float, list of ints, list of float): padding value if border_mode is cv2.BORDER_CONSTANT. mask_value (int, float, list of ints, list of float): padding value if border_mode is cv2.BORDER_CONSTANT applied for masks. Targets: image, mask Image types: uint8, float32
Value
None
Targets
image, mask
Image types
uint8, float32
PadIfNeeded
Description
Pad side of the image / max if side is less than desired number.
Usage
icevision_PadIfNeeded(
min_height = 1024,
min_width = 1024,
pad_height_divisor = NULL,
pad_width_divisor = NULL,
border_mode = 4,
value = NULL,
mask_value = NULL,
always_apply = FALSE,
p = 1
)
Arguments
min_height |
min_height |
min_width |
min_width |
pad_height_divisor |
pad_height_divisor |
pad_width_divisor |
pad_width_divisor |
border_mode |
border_mode |
value |
value |
mask_value |
mask_value |
always_apply |
always_apply |
p |
p |
Targets
image, mask, bbox, keypoints
Image types
uint8, float32
Posterize
Description
Reduce the number of bits for each color channel.
Usage
icevision_Posterize(num_bits = 4, always_apply = FALSE, p = 0.5)
Arguments
num_bits |
num_bits |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8
RGBShift
Description
Randomly shift values for each channel of the input RGB image.
Randomly shift values for each channel of the input RGB image.
Usage
icevision_RGBShift(
r_shift_limit = 20,
g_shift_limit = 20,
b_shift_limit = 20,
always_apply = FALSE,
p = 0.5
)
icevision_RGBShift(
r_shift_limit = 20,
g_shift_limit = 20,
b_shift_limit = 20,
always_apply = FALSE,
p = 0.5
)
Arguments
r_shift_limit |
r_shift_limit |
g_shift_limit |
g_shift_limit |
b_shift_limit |
b_shift_limit |
always_apply |
always_apply |
p |
p |
Value
None
None
Targets
image
image
Image types
uint8, float32
uint8, float32
RandomBrightnessContrast
Description
Randomly change brightness and contrast of the input image.
Usage
icevision_RandomBrightnessContrast(
brightness_limit = 0.2,
contrast_limit = 0.2,
brightness_by_max = TRUE,
always_apply = FALSE,
p = 0.5
)
Arguments
brightness_limit |
brightness_limit |
contrast_limit |
contrast_limit |
brightness_by_max |
brightness_by_max |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
RandomContrast
Description
Randomly change contrast of the input image.
Usage
icevision_RandomContrast(limit = 0.2, always_apply = FALSE, p = 0.5)
Arguments
limit |
limit |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
RandomCrop
Description
Crop a random part of the input.
Usage
icevision_RandomCrop(height, width, always_apply = FALSE, p = 1)
Arguments
height |
height |
width |
width |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
RandomCropNearBBox
Description
Crop bbox from image with random shift by x,y coordinates
Usage
icevision_RandomCropNearBBox(max_part_shift = 0.3, always_apply = FALSE, p = 1)
Arguments
max_part_shift |
max_part_shift |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
RandomFog
Description
Simulates fog for the image
Usage
icevision_RandomFog(
fog_coef_lower = 0.3,
fog_coef_upper = 1,
alpha_coef = 0.08,
always_apply = FALSE,
p = 0.5
)
Arguments
fog_coef_lower |
fog_coef_lower |
fog_coef_upper |
fog_coef_upper |
alpha_coef |
alpha_coef |
always_apply |
always_apply |
p |
p |
Details
From https://github.com/UjjwalSaxena/Automold–Road-Augmentation-Library
Value
None
Targets
image
Image types
uint8, float32
RandomGamma
Description
RandomGamma
Usage
icevision_RandomGamma(
gamma_limit = list(80, 120),
eps = NULL,
always_apply = FALSE,
p = 0.5
)
Arguments
gamma_limit |
gamma_limit |
eps |
Deprecated. |
always_apply |
always_apply |
p |
p |
Details
gamma_limit (float or (float, float)): If gamma_limit is a single float value, the range will be (-gamma_limit, gamma_limit). Default: (80, 120). eps: Deprecated. Targets: image Image types: uint8, float32
Value
None
Targets
image
Image types
uint8, float32
RandomGridShuffle
Description
Random shuffle grid's cells on image.
Usage
icevision_RandomGridShuffle(grid = list(3, 3), always_apply = FALSE, p = 0.5)
Arguments
grid |
grid |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask
Image types
uint8, float32
RandomRain
Description
Adds rain effects.
Usage
icevision_RandomRain(
slant_lower = -10,
slant_upper = 10,
drop_length = 20,
drop_width = 1,
drop_color = list(200, 200, 200),
blur_value = 7,
brightness_coefficient = 0.7,
rain_type = NULL,
always_apply = FALSE,
p = 0.5
)
Arguments
slant_lower |
should be in range [-20, 20]. |
slant_upper |
should be in range [-20, 20]. |
drop_length |
should be in range [0, 100]. |
drop_width |
should be in range [1, 5]. drop_color (list of (r, g, b)): rain lines color. blur_value (int): rainy view are blurry brightness_coefficient (float): rainy days are usually shady. Should be in range [0, 1]. |
drop_color |
drop_color |
blur_value |
blur_value |
brightness_coefficient |
brightness_coefficient |
rain_type |
One of [NULL, "drizzle", "heavy", "torrestial"] |
always_apply |
always_apply |
p |
p |
Details
From https://github.com/UjjwalSaxena/Automold–Road-Augmentation-Library
Value
None
Targets
image
Image types
uint8, float32
RandomResizedCrop
Description
Torchvision's variant of crop a random part of the input and rescale it to some size.
Usage
icevision_RandomResizedCrop(
height,
width,
scale = list(0.08, 1),
ratio = list(0.75, 1.33333333333333),
interpolation = 1,
always_apply = FALSE,
p = 1
)
Arguments
height |
height |
width |
width |
scale |
scale |
ratio |
ratio |
interpolation |
interpolation |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
RandomRotate90
Description
Randomly rotate the input by 90 degrees zero or more times.
Usage
icevision_RandomRotate90(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
RandomScale
Description
Randomly resize the input. Output image size is different from the input image size.
Usage
icevision_RandomScale(
scale_limit = 0.1,
interpolation = 1L,
always_apply = FALSE,
p = 0.5
)
Arguments
scale_limit |
scale_limit |
interpolation |
interpolation |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
RandomShadow
Description
Simulates shadows for the image
Usage
icevision_RandomShadow(
shadow_roi = list(0, 0.5, 1, 1),
num_shadows_lower = 1,
num_shadows_upper = 2,
shadow_dimension = 5,
always_apply = FALSE,
p = 0.5
)
Arguments
shadow_roi |
shadow_roi |
num_shadows_lower |
num_shadows_lower |
num_shadows_upper |
num_shadows_upper |
shadow_dimension |
shadow_dimension |
always_apply |
always_apply |
p |
p |
Details
From https://github.com/UjjwalSaxena/Automold–Road-Augmentation-Library
Value
None
Targets
image
Image types
uint8, float32
RandomSizedBBoxSafeCrop
Description
Crop a random part of the input and rescale it to some size without loss of bboxes.
Crop a random part of the input and rescale it to some size without loss of bboxes.
Usage
icevision_RandomSizedBBoxSafeCrop(
height,
width,
erosion_rate = 0,
interpolation = 1,
always_apply = FALSE,
p = 1
)
icevision_RandomSizedBBoxSafeCrop(
height,
width,
erosion_rate = 0,
interpolation = 1,
always_apply = FALSE,
p = 1
)
Arguments
height |
height |
width |
width |
erosion_rate |
erosion_rate |
interpolation |
interpolation |
always_apply |
always_apply |
p |
p |
Value
None
None
Targets
image, mask, bboxes
image, mask, bboxes
Image types
uint8, float32
uint8, float32
RandomSizedCrop
Description
Crop a random part of the input and rescale it to some size.
Usage
icevision_RandomSizedCrop(
min_max_height,
height,
width,
w2h_ratio = 1,
interpolation = 1,
always_apply = FALSE,
p = 1
)
Arguments
min_max_height |
min_max_height |
height |
height |
width |
width |
w2h_ratio |
w2h_ratio |
interpolation |
interpolation |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
RandomSnow
Description
Bleach out some pixel values simulating snow.
Usage
icevision_RandomSnow(
snow_point_lower = 0.1,
snow_point_upper = 0.3,
brightness_coeff = 2.5,
always_apply = FALSE,
p = 0.5
)
Arguments
snow_point_lower |
snow_point_lower |
snow_point_upper |
snow_point_upper |
brightness_coeff |
brightness_coeff |
always_apply |
always_apply |
p |
p |
Details
From https://github.com/UjjwalSaxena/Automold–Road-Augmentation-Library
Value
None
Targets
image
Image types
uint8, float32
RandomSplitter
Description
Randomly splits items.
Usage
icevision_RandomSplitter(probs, seed = NULL)
Arguments
probs |
'Sequence' of probabilities that must sum to one. The length of the 'Sequence' is the number of groups to to split the items into. |
seed |
Internal seed used for shuffling the items. Define this if you need reproducible results. |
Value
None
RandomSunFlare
Description
Simulates Sun Flare for the image
Usage
icevision_RandomSunFlare(
flare_roi = list(0, 0, 1, 0.5),
angle_lower = 0,
angle_upper = 1,
num_flare_circles_lower = 6,
num_flare_circles_upper = 10,
src_radius = 400,
src_color = list(255, 255, 255),
always_apply = FALSE,
p = 0.5
)
Arguments
flare_roi |
flare_roi |
angle_lower |
angle_lower |
angle_upper |
angle_upper |
num_flare_circles_lower |
num_flare_circles_lower |
num_flare_circles_upper |
num_flare_circles_upper |
src_radius |
src_radius |
src_color |
src_color |
always_apply |
always_apply |
p |
p |
Details
From https://github.com/UjjwalSaxena/Automold–Road-Augmentation-Library
Value
None
Targets
image
Image types
uint8, float32
Resize
Description
Resize the input to the given height and width.
Usage
icevision_Resize(height, width, interpolation = 1, always_apply = FALSE, p = 1)
Arguments
height |
height |
width |
width |
interpolation |
interpolation |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
Rotate
Description
Rotate the input by an angle selected randomly from the uniform distribution.
Usage
icevision_Rotate(
limit = 90,
interpolation = 1,
border_mode = 4,
value = NULL,
mask_value = NULL,
always_apply = FALSE,
p = 0.5
)
Arguments
limit |
limit |
interpolation |
interpolation |
border_mode |
border_mode |
value |
value |
mask_value |
mask_value |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
ShiftScaleRotate
Description
Randomly apply affine transforms: translate, scale and rotate the input.
Randomly apply affine transforms: translate, scale and rotate the input.
Usage
icevision_ShiftScaleRotate(
shift_limit = 0.0625,
scale_limit = 0.1,
rotate_limit = 45,
interpolation = 1,
border_mode = 4,
value = NULL,
mask_value = NULL,
shift_limit_x = NULL,
shift_limit_y = NULL,
always_apply = FALSE,
p = 0.5
)
icevision_ShiftScaleRotate(
shift_limit = 0.0625,
scale_limit = 0.1,
rotate_limit = 45,
interpolation = 1,
border_mode = 4,
value = NULL,
mask_value = NULL,
shift_limit_x = NULL,
shift_limit_y = NULL,
always_apply = FALSE,
p = 0.5
)
Arguments
shift_limit |
shift_limit |
scale_limit |
scale_limit |
rotate_limit |
rotate_limit |
interpolation |
interpolation |
border_mode |
border_mode |
value |
value |
mask_value |
mask_value |
shift_limit_x |
shift_limit_x |
shift_limit_y |
shift_limit_y |
always_apply |
always_apply |
p |
p |
Value
None
None
Targets
image, mask, keypoints
image, mask, keypoints
Image types
uint8, float32
uint8, float32
SingleSplitSplitter
Description
SingleSplitSplitter
Usage
icevision_SingleSplitSplitter(...)
Arguments
... |
arguments to pass |
Value
all items in a single group, without shuffling.
SmallestMaxSize
Description
Rescale an image so that minimum side is equal to max_size, keeping the aspect ratio of the initial image.
Usage
icevision_SmallestMaxSize(
max_size = 1024,
interpolation = 1,
always_apply = FALSE,
p = 1
)
Arguments
max_size |
max_size |
interpolation |
interpolation |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
Solarize
Description
Invert all pixel values above a threshold.
Usage
icevision_Solarize(threshold = 128, always_apply = FALSE, p = 0.5)
Arguments
threshold |
threshold |
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
any
ToFloat
Description
Divide pixel values by 'max_value' to get a float32 output array where all values lie in the range [0, 1.0].
Usage
icevision_ToFloat(max_value = NULL, always_apply = FALSE, p = 1)
Arguments
max_value |
max_value |
always_apply |
always_apply |
p |
p |
Details
If 'max_value' is NULL the transform will try to infer the maximum value by inspecting the data type of the input image. See Also: :class:'~albumentations.augmentations.transforms.FromFloat'
Value
None
See Also
:class:'~albumentations.augmentations.transforms.FromFloat'
Targets
image
Image types
any type
ToGray
Description
Convert the input RGB image to grayscale. If the mean pixel value for the resulting image is greater than 127, invert the resulting grayscale image.
Usage
icevision_ToGray(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
ToSepia
Description
Applies sepia filter to the input RGB image
Usage
icevision_ToSepia(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
Targets
image
Image types
uint8, float32
Transpose
Description
Transpose the input by swapping rows and columns.
Usage
icevision_Transpose(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
VerticalFlip
Description
Flip the input vertically around the x-axis.
Usage
icevision_VerticalFlip(always_apply = FALSE, p = 0.5)
Arguments
always_apply |
always_apply |
p |
p |
Value
None
Targets
image, mask, bboxes, keypoints
Image types
uint8, float32
Aug_tfms
Description
Collection of useful augmentation transforms.
Usage
icevision_aug_tfms(
size,
presize = NULL,
horizontal_flip = icevision_HorizontalFlip(always_apply = FALSE, p = 0.5),
shift_scale_rotate = icevision_ShiftScaleRotate(always_apply = FALSE, p = 0.5,
shift_limit_x = c(-0.0625, 0.0625), shift_limit_y = c(-0.0625, 0.0625), scale_limit =
c(-0.1, 0.1), rotate_limit = c(-45, 45), interpolation = 1, border_mode = 4, value =
NULL, mask_value = NULL),
rgb_shift = icevision_RGBShift(always_apply = FALSE, p = 0.5, r_shift_limit = c(-20,
20), g_shift_limit = c(-20, 20), b_shift_limit = c(-20, 20)),
lightning = icevision_RandomBrightnessContrast(always_apply = FALSE, p = 0.5,
brightness_limit = c(-0.2, 0.2), contrast_limit = c(-0.2, 0.2), brightness_by_max =
TRUE),
blur = icevision_Blur(always_apply = FALSE, p = 0.5, blur_limit = c(1, 3)),
crop_fn = partial(icevision_RandomSizedBBoxSafeCrop, p = 0.5),
pad = partial(icevision_PadIfNeeded, border_mode = 0, value = list(124, 116, 104))
)
Arguments
size |
The final size of the image. If an 'int' is given, the maximum size of the image is rescaled, maintaing aspect ratio. If a 'list' is given, the image is rescaled to have that exact size (height, width). |
presize |
presize |
horizontal_flip |
Flip around the y-axis. If 'NULL' this transform is not applied. |
shift_scale_rotate |
Randomly shift, scale, and rotate. If 'NULL' this transform is not applied. |
rgb_shift |
Randomly shift values for each channel of RGB image. If 'NULL' this transform is not applied. |
lightning |
Randomly changes Brightness and Contrast. If 'NULL' this transform is not applied. |
blur |
Randomly blur the image. If 'NULL' this transform is not applied. |
crop_fn |
Randomly crop the image. If 'NULL' this transform is not applied. Use 'partial' to saturate other parameters of the class. |
pad |
Pad the image to 'size', squaring the image if 'size' is an 'int'. If 'NULL' this transform is not applied. Use 'partial' to sature other parameters of the class. |
Value
None
Parse
Description
Loops through all data points parsing the required fields.
Usage
icevision_parse(
data_splitter = NULL,
idmap = NULL,
autofix = TRUE,
show_pbar = TRUE,
cache_filepath = NULL
)
Arguments
data_splitter |
How to split the parsed data, defaults to a [0.8, 0.2] random split. |
idmap |
Maps from filenames to unique ids, pass an 'IDMap()' if you need this information. |
autofix |
autofix |
show_pbar |
Whether or not to show a progress bar while parsing the data. |
cache_filepath |
Path to save records in pickle format. Defaults to NULL, e.g. if the user does not specify a path, no saving nor loading happens. |
Value
A list of records for each split defined by data_splitter.
Read_bgr_image
Description
Read_bgr_image
Usage
icevision_read_bgr_image(path)
Arguments
path |
path |
Value
None
Read_rgb_image
Description
Read_rgb_image
Usage
icevision_read_rgb_image(path)
Arguments
path |
path |
Value
None
Resize_and_pad
Description
Resize_and_pad
Usage
icevision_resize_and_pad(
size,
pad = partial(icevision_PadIfNeeded, border_mode = 0, value = c(124L, 116L, 104L))
)
Arguments
size |
size |
pad |
pad |
Value
None
Icnr_init
Description
ICNR init of 'x', with 'scale' and 'init' function
Usage
icnr_init(x, scale = 2, init = nn()$init$kaiming_normal_)
Arguments
x |
tensor |
scale |
int, scale |
init |
initializer |
Value
None
Image2tensor
Description
Transform image to byte tensor in 'c*h*w' dim order.
Usage
image2tensor(img)
Arguments
img |
image |
Value
None
Imagenet statistics
Description
Imagenet statistics
Usage
imagenet_stats()
Value
vector
Examples
## Not run:
imagenet_stats()
## End(Not run)
In_channels
Description
Return the shape of the first weight layer in 'm'.
Usage
in_channels(m)
Arguments
m |
parameters |
Value
None
Wandb init
Description
Initialize a wandb Run.
Usage
init(...)
Arguments
... |
parameters to pass |
Value
wandb Run object
None
see https
//docs.wandb.com/library/init
Init_default
Description
Initialize 'm' weights with 'func' and set 'bias' to 0.
Usage
init_default(m, func = nn()$init$kaiming_normal_)
Arguments
m |
parameters |
func |
function |
Value
None
Init_linear
Description
Init_linear
Usage
init_linear(m, act_func = NULL, init = "auto", bias_std = 0.01)
Arguments
m |
parameter |
act_func |
activation function |
init |
initializer |
bias_std |
bias standard deviation |
Value
None
Install fastai
Description
Install fastai
Usage
install_fastai(
version,
gpu = FALSE,
cuda_version = "11.8",
overwrite = FALSE,
extra_pkgs = c("timm", "fastinference[interp]"),
TPU = FALSE
)
Arguments
version |
specify version |
gpu |
installation of gpu |
cuda_version |
if gpu true, then cuda version is required. By default it is 11.6 |
overwrite |
will install all the dependencies |
extra_pkgs |
character vector of additional packages |
TPU |
official way to install Pytorch-XLA 1.13 |
Value
None
Is Rmarkdown?
Description
Is Rmarkdown?
Usage
is_rmarkdown()
Value
logical True/False
Kaggle module
Description
Kaggle module
Usage
kg()
Value
None
L2_reg
Description
L2 regularization as adding 'wd*p' to 'p$grad'
Usage
l2_reg(p, lr, wd, do_wd = TRUE, ...)
Arguments
p |
p |
lr |
learning rate |
wd |
weight decay |
do_wd |
do_wd |
... |
additional arguments to pass |
Value
None
Examples
## Not run:
tst_param = function(val, grad = NULL) {
"Create a tensor with `val` and a gradient of `grad` for testing"
res = tensor(val) %>% float()
if(is.null(grad)) {
grad = tensor(val / 10)
} else {
grad = tensor(grad)
}
res$grad = grad %>% float()
res
}
p = tst_param(1., 0.1)
l2_reg(p, 1., 0.1)
## End(Not run)
Lamb_step
Description
Step for LAMB with 'lr' on 'p'
Usage
lamb_step(p, lr, mom, step, sqr_mom, grad_avg, sqr_avg, eps, ...)
Arguments
p |
p |
lr |
learning rate |
mom |
momentum |
step |
step |
sqr_mom |
sqr momentum |
grad_avg |
gradient average |
sqr_avg |
sqr average |
eps |
epsilon |
... |
additional arguments to pass |
Value
None
Language_model_learner
Description
Create a 'Learner' with a language model from 'dls' and 'arch'.
Usage
language_model_learner(
dls,
arch,
config = NULL,
drop_mult = 1,
backwards = FALSE,
pretrained = TRUE,
pretrained_fnames = NULL,
opt_func = Adam(),
lr = 0.001,
cbs = NULL,
metrics = NULL,
path = NULL,
model_dir = "models",
wd = NULL,
wd_bn_bias = FALSE,
train_bn = TRUE,
moms = list(0.95, 0.85, 0.95),
...
)
Arguments
dls |
dls |
arch |
arch |
config |
config |
drop_mult |
drop_mult |
backwards |
backwards |
pretrained |
pretrained |
pretrained_fnames |
pretrained_fnames |
opt_func |
opt_func |
lr |
lr |
cbs |
cbs |
metrics |
metrics |
path |
path |
model_dir |
model_dir |
wd |
wd |
wd_bn_bias |
wd_bn_bias |
train_bn |
train_bn |
moms |
moms |
... |
additional arguments |
Value
None
Larc_layer_lr
Description
Computes the local lr before weight decay is applied
Usage
larc_layer_lr(p, lr, trust_coeff, wd, eps, clip = TRUE, ...)
Arguments
p |
p |
lr |
learning rate |
trust_coeff |
trust_coeff |
wd |
weight decay |
eps |
epsilon |
clip |
clip |
... |
additional arguments to pass |
Value
None
Larc_step
Description
Step for LARC 'local_lr' on 'p'
Usage
larc_step(p, local_lr, grad_avg = NULL, ...)
Arguments
p |
p |
local_lr |
local learning rate |
grad_avg |
gradient average |
... |
additional args to pass |
Value
None
Layer_info
Description
Return layer infos of 'model' on 'xb' (only support batch first inputs)
Usage
layer_info(learn, ...)
Arguments
learn |
learner/model |
... |
additional arguments |
Value
None
Length
Description
Length
Usage
## S3 method for class 'torch.Tensor'
length(x)
Arguments
x |
tensor |
Value
tensor
Length
Description
Length
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
length(x)
Arguments
x |
tensor |
Value
tensor
Less
Description
Less
Usage
## S3 method for class 'torch.Tensor'
a < b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Less or equal
Description
Less or equal
Usage
## S3 method for class 'torch.Tensor'
a <= b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Load_dataset
Description
A helper function for getting a DataLoader for images in the folder 'test_path', with batch size 'bs', and number of workers 'num_workers'
Usage
load_dataset(test_path, bs = 4, num_workers = 4)
Arguments
test_path |
test path (directory) |
bs |
batch size |
num_workers |
number of workers |
Value
None
Load_ignore_keys
Description
Load 'wgts' in 'model' ignoring the names of the keys, just taking parameters in order
Usage
load_ignore_keys(model, wgts)
Arguments
model |
model |
wgts |
wgts |
Value
None
Load_image
Description
Open and load a 'PIL.Image' and convert to 'mode'
Usage
load_image(fn, mode = NULL)
Arguments
fn |
file name |
mode |
mode |
Value
None
Load_learner
Description
Load a 'Learner' object in 'fname', optionally putting it on the 'cpu'
Usage
load_learner(fname, cpu = TRUE)
Arguments
fname |
fname |
cpu |
cpu or not |
Value
learner object
Load_model_text
Description
Load 'model' from 'file' along with 'opt' (if available, and if 'with_opt')
Usage
load_model_text(
file,
model,
opt,
with_opt = NULL,
device = NULL,
strict = TRUE
)
Arguments
file |
file |
model |
model |
opt |
opt |
with_opt |
with_opt |
device |
device |
strict |
strict |
Value
None
Timm models
Description
Timm models
Usage
load_pre_models()
Value
None
Load_tokenized_csv
Description
Utility function to quickly load a tokenized csv and the corresponding counter
Usage
load_tokenized_csv(fname)
Arguments
fname |
file name |
Value
None
Loaders
Description
a loader from Catalyst
Usage
loaders()
Value
None
Examples
## Not run:
# trigger download
loaders()
## End(Not run)
Log
Description
Log
Usage
## S3 method for class 'torch.Tensor'
log(x, base = exp(1))
Arguments
x |
tensor |
base |
base parameter |
Value
tensor
Log
Description
Log
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
log(x, base = exp(1))
Arguments
x |
tensor |
base |
base parameter |
Value
tensor
Log1p
Description
Log1p
Usage
## S3 method for class 'torch.Tensor'
log1p(x)
Arguments
x |
tensor |
Value
tensor
Log1p
Description
Log1p
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
log1p(x)
Arguments
x |
tensor |
Value
tensor
Logical_and
Description
Logical_and
Usage
## S3 method for class 'torch.Tensor'
x & y
Arguments
x |
tensor |
y |
tensor |
Value
tensor
Logical_not
Description
Logical_not
Usage
## S3 method for class 'torch.Tensor'
!x
Arguments
x |
tensor |
Value
tensor
Logical_or
Description
Logical_or
Usage
## S3 method for class 'torch.Tensor'
x | y
Arguments
x |
tensor |
y |
tensor |
Value
tensor
Wandb login
Description
Log in to W&B.
Usage
login(anonymous = NULL, key = NULL, relogin = NULL, host = NULL, force = NULL)
Arguments
anonymous |
must,never,allow,false,true |
key |
API key (secret) |
relogin |
relogin or not |
host |
host address |
force |
whether to force a user to be logged into wandb when running a script |
Value
None
Lr_find
Description
Launch a mock training to find a good learning rate, return lr_min, lr_steep if 'suggestions' is TRUE
Usage
lr_find(
object,
start_lr = 1e-07,
end_lr = 10,
num_it = 100,
stop_div = TRUE,
...
)
Arguments
object |
learner |
start_lr |
starting learning rate |
end_lr |
end learning rate |
num_it |
number of iterations |
stop_div |
stop div or not |
... |
additional arguments to pass |
Value
data frame
Examples
## Not run:
model %>% lr_find()
model %>% plot_lr_find(dpi = 200)
## End(Not run)
MAE
Description
Mean absolute error between 'inp' and 'targ'.
Usage
mae(inp, targ)
Arguments
inp |
predictions |
targ |
targets |
Value
None
Make_vocab
Description
Create a vocab of 'max_vocab' size from 'Counter' 'count' with items present more than 'min_freq'
Usage
make_vocab(count, min_freq = 3, max_vocab = 60000, special_toks = NULL)
Arguments
count |
count |
min_freq |
min_freq |
max_vocab |
max_vocab |
special_toks |
special_toks |
Value
None
Mask2bbox
Description
Mask2bbox
Usage
mask2bbox(mask, convert = TRUE)
Arguments
mask |
mask |
convert |
to R matrix |
Value
tensor
Mask from blur
Description
Mask from blur
Usage
mask_from_blur(img, window, sigma = 0.3, thresh = 0.05, remove_max = TRUE)
Arguments
img |
image |
window |
windowing effect |
sigma |
sigma |
thresh |
thresholf point |
remove_max |
remove maximum or not |
Mask RCNN infer dataloader
Description
A 'DataLoader' with a custom 'collate_fn' that batches items as required for inferring the model.
Usage
mask_rcnn_infer_dl(dataset, batch_tfms = NULL, ...)
Arguments
dataset |
Possibly a 'Dataset' object, but more generally, any 'Sequence' that returns records. |
batch_tfms |
Transforms to be applied at the batch level. **dataloader_kwargs: Keyword arguments that will be internally passed to a Pytorch 'DataLoader'. The parameter 'collate_fn' is already defined internally and cannot be passed here. |
... |
additional arguments |
Value
None
MaskRCNN learner
Description
Fastai 'Learner' adapted for MaskRCNN.
Usage
mask_rcnn_learner(dls, model, cbs = NULL, ...)
Arguments
dls |
'Sequence' of 'DataLoaders' passed to the 'Learner'. The first one will be used for training and the second for validation. |
model |
The model to train. |
cbs |
Optional 'Sequence' of callbacks. |
... |
learner_kwargs: Keyword arguments that will be internally passed to 'Learner'. |
Value
model
MaskRCNN model
Description
MaskRCNN model implemented by torchvision.
Usage
mask_rcnn_model(
num_classes,
backbone = NULL,
remove_internal_transforms = TRUE,
pretrained = TRUE
)
Arguments
num_classes |
Number of classes. |
backbone |
Backbone model to use. Defaults to a resnet50_fpn model. |
remove_internal_transforms |
The torchvision model internally applies transforms like resizing and normalization, but we already do this at the ‘Dataset' level, so it’s safe to remove those internal transforms. |
pretrained |
Argument passed to 'maskrcnn_resnet50_fpn' if 'backbone is NULL'. By default it is set to TRUE: this is generally used when training a new model (transfer learning). 'pretrained = FALSE' is used during inference (prediction) for cases where the users have their own pretrained weights. **mask_rcnn_kwargs: Keyword arguments that internally are going to be passed to 'torchvision.models.detection.mask_rcnn.MaskRCNN'. |
Value
model
Mask RCNN predict dataloader
Description
Mask RCNN predict dataloader
Usage
mask_rcnn_predict_dl(model, infer_dl, show_pbar = TRUE)
Arguments
model |
model |
infer_dl |
infer_dl |
show_pbar |
show_pbar |
Value
None
MaskRCNN train dataloader
Description
A 'DataLoader' with a custom 'collate_fn' that batches items as required for training the model.
Usage
mask_rcnn_train_dl(dataset, batch_tfms = NULL, ...)
Arguments
dataset |
Possibly a 'Dataset' object, but more generally, any 'Sequence' that returns records. |
batch_tfms |
Transforms to be applied at the batch level. |
... |
dataloader_kwargs: Keyword arguments that will be internally passed to a Pytorch 'DataLoader'. The parameter 'collate_fn' is already defined internally and cannot be passed here. |
Value
None
MaskRSNN valid dataloader
Description
A 'DataLoader' with a custom 'collate_fn' that batches items as required for training the model.
Usage
mask_rcnn_valid_dl(dataset, batch_tfms = NULL, ...)
Arguments
dataset |
Possibly a 'Dataset' object, but more generally, any 'Sequence' that returns records. |
batch_tfms |
Transforms to be applied at the batch level. |
... |
dataloader_kwargs: Keyword arguments that will be internally passed to a Pytorch 'DataLoader'. The parameter 'collate_fn' is already defined internally and cannot be passed here. |
Value
None
Mask_tensor
Description
Mask elements of 'x' with 'neutral' with probability '1-p'
Usage
mask_tensor(x, p = 0.5, neutral = 0, batch = FALSE)
Arguments
x |
tensor |
p |
probability |
neutral |
neutral |
batch |
batch |
Value
None
Masked_concat_pool
Description
Pool 'MultiBatchEncoder' outputs into one vector [last_hidden, max_pool, avg_pool]
Usage
masked_concat_pool(output, mask, bptt)
Arguments
output |
output |
mask |
mask |
bptt |
bptt |
Value
None
Match_embeds
Description
Convert the embedding in 'old_wgts' to go from 'old_vocab' to 'new_vocab'.
Usage
match_embeds(old_wgts, old_vocab, new_vocab)
Arguments
old_wgts |
old_wgts |
old_vocab |
old_vocab |
new_vocab |
new_vocab |
Value
None
Max
Description
Max
Usage
## S3 method for class 'torch.Tensor'
max(a, ..., na.rm = FALSE)
Arguments
a |
tensor |
... |
additional parameters |
na.rm |
remove NAs |
Value
tensor
Max
Description
Max
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
max(a, ..., na.rm = FALSE)
Arguments
a |
tensor |
... |
additional parameters |
na.rm |
remove NAs |
Value
tensor
Maybe_unsqueeze
Description
Add empty dimension if it is a rank 1 tensor/array
Usage
maybe_unsqueeze(x)
Arguments
x |
R array/matrix/tensor |
Value
array
Mean of tensor
Description
Mean of tensor
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
mean(x, ...)
Arguments
x |
tensor |
... |
additional parameters to pass |
Value
tensor
Mean of tensor
Description
Mean of tensor
Usage
## S3 method for class 'torch.Tensor'
mean(x, ...)
Arguments
x |
tensor |
... |
additional parameters to pass |
Value
tensor
Medical module
Description
Medical module
Usage
medical()
Value
None
Metrics module
Description
Metrics module
Usage
metrics()
Value
None
Ignite module
Description
Ignite module
Usage
migrating_ignite()
Value
None
Lightning module
Description
Lightning module
Usage
migrating_lightning()
Value
None
Pytorch module
Description
Pytorch module
Usage
migrating_pytorch()
Value
None
Min
Description
Min
Usage
## S3 method for class 'torch.Tensor'
min(a, ..., na.rm = FALSE)
Arguments
a |
tensor |
... |
additional parameters |
na.rm |
remove NAs |
Value
tensor
Min
Description
Min
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
min(a, ..., na.rm = FALSE)
Arguments
a |
tensor |
... |
additional parameters |
na.rm |
remove NAs |
Value
tensor
Mish
Description
Mish
Usage
mish(x)
Arguments
x |
tensor |
Value
None
Model_sizes
Description
Pass a dummy input through the model 'm' to get the various sizes of activations.
Usage
model_sizes(m, size = list(64, 64))
Arguments
m |
m parameter |
size |
size |
Value
None
Momentum_step
Description
Step for SGD with momentum with 'lr'
Usage
momentum_step(p, lr, grad_avg, ...)
Arguments
p |
p |
lr |
learning rate |
grad_avg |
grad average |
... |
additional arguments to pass |
Value
None
Most_confused
Description
Sorted descending list of largest non-diagonal entries of confusion matrix, presented as actual, predicted, number of occurrences.
Usage
most_confused(interp, min_val = 1)
Arguments
interp |
interpretation object |
min_val |
minimum value |
Value
data frame
MSE
Description
Mean squared error between 'inp' and 'targ'.
Usage
mse(inp, targ)
Arguments
inp |
predictions |
targ |
targets |
Value
None
Examples
## Not run:
model = dls %>% tabular_learner(layers=c(200,100,100,200),
metrics = list(mse(),rmse()) )
## End(Not run)
MSLE
Description
Mean squared logarithmic error between 'inp' and 'targ'.
Usage
msle(inp, targ)
Arguments
inp |
predictions |
targ |
targets |
Value
None
Multiply
Description
Multiply
Usage
## S3 method for class 'torch.Tensor'
a * b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
N_px
Description
int(x=0) -> integer
Usage
n_px(img)
Arguments
img |
image |
Value
None
Modify tensor
Description
Modify tensor
Usage
narrow(tensor, slice)
Arguments
tensor |
torch tensor |
slice |
dimension |
Value
tensor
NN module
Description
NN module
Usage
nn()
Value
None
Fastai custom loss
Description
Fastai custom loss
Usage
nn_loss(loss_fn, name = "Custom_Loss")
Arguments
loss_fn |
pass custom model function |
name |
set name for nn_module |
Value
None
Fastai NN module
Description
Fastai NN module
Usage
nn_module(model_fn, name = "Custom_Model", gpu = TRUE)
Arguments
model_fn |
pass custom model function |
name |
set name for nn_module |
gpu |
move model to GPU |
Value
None
Noop
Description
Noop
Usage
noop(...)
Arguments
... |
parameters to pass |
Value
None
Norm_apply_denorm
Description
Normalize 'x' with 'nrm', then apply 'f', then denormalize
Usage
norm_apply_denorm(x, f, nrm)
Arguments
x |
tensor |
f |
function |
nrm |
nrm |
Value
None
Logical_not
Description
Logical_not
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
!x
Arguments
x |
tensor |
Value
tensor
Not equal
Description
Not equal
Usage
## S3 method for class 'torch.Tensor'
a != b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Not equal
Description
Not equal
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
a != b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Num_features_model
Description
Return the number of output features for 'm'.
Usage
num_features_model(m)
Arguments
m |
m parameter |
Value
None
One batch
Description
One batch
Usage
one_batch(object, convert = FALSE, ...)
Arguments
object |
data loader |
convert |
to R matrix |
... |
additional parameters to pass |
Value
tensor
Examples
## Not run:
# get batch from data loader
batch = dls %>% one_batch()
## End(Not run)
Optim metric
Description
Replace metric 'f' with a version that optimizes argument 'argname'
Usage
optim_metric(f, argname, bounds, tol = 0.01, do_neg = TRUE, get_x = FALSE)
Arguments
f |
f |
argname |
argname |
bounds |
bounds |
tol |
tol |
do_neg |
do_neg |
get_x |
get_x |
Value
None
Logical_or
Description
Logical_or
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
x | y
Arguments
x |
tensor |
y |
tensor |
Value
tensor
Operating system
Description
Operating system
Usage
os()
Value
vector
An environment supporting TPUs
Description
An environment supporting TPUs
Usage
os_environ_tpu(text = "COLAB_TPU_ADDR")
Arguments
text |
string to pass to environment |
Value
None
Pad_conv_norm_relu
Description
Pad_conv_norm_relu
Usage
pad_conv_norm_relu(
ch_in,
ch_out,
pad_mode,
norm_layer,
ks = 3,
bias = TRUE,
pad = 1,
stride = 1,
activ = TRUE,
init = nn()$init$kaiming_normal_,
init_gain = 0.02
)
Arguments
ch_in |
input |
ch_out |
output |
pad_mode |
padding mode |
norm_layer |
normalization layer |
ks |
kernel size |
bias |
bias |
pad |
padding |
stride |
stride |
activ |
activation |
init |
initializer |
init_gain |
init gain |
Value
None
Pad_input
Description
Function that collect 'samples' and adds padding
Usage
pad_input(
samples,
pad_idx = 1,
pad_fields = 0,
pad_first = FALSE,
backwards = FALSE
)
Arguments
samples |
samples |
pad_idx |
pad_idx |
pad_fields |
pad_fields |
pad_first |
pad_first |
backwards |
backwards |
Value
None
Pad_input_chunk
Description
Pad 'samples' by adding padding by chunks of size 'seq_len'
Usage
pad_input_chunk(samples, pad_idx = 1, pad_first = TRUE, seq_len = 72)
Arguments
samples |
samples |
pad_idx |
pad_idx |
pad_first |
pad_first |
seq_len |
seq_len |
Value
None
Parallel
Description
Applies 'func' in parallel to 'items', using 'n_workers'
Usage
parallel(f, items, ...)
Arguments
f |
file names |
items |
items |
... |
additional arguments |
Value
None
Parallel_tokenize
Description
Calls optional 'setup' on 'tok' before launching 'TokenizeWithRules' using 'parallel_gen
Usage
parallel_tokenize(items, tok = NULL, rules = NULL, n_workers = 6)
Arguments
items |
items |
tok |
tokenizer |
rules |
rules |
n_workers |
n_workers |
Value
None
Params
Description
Return all parameters of 'm'
Usage
params(m)
Arguments
m |
parameters |
Value
None
Parent_label
Description
Label 'item' with the parent folder name.
Usage
parent_label(o)
Arguments
o |
string, dir path |
Value
vector
AreasMixin
Description
Adds areas method to parser
Usage
parsers_AreasMixin(...)
Arguments
... |
arguments to pass |
Value
None
BBoxesMixin
Description
Adds bboxes method to parser
Usage
parsers_BBoxesMixin(...)
Arguments
... |
arguments to pass |
Value
None
Faster RCNN
Description
Parser with required mixins for Faster RCNN.
Usage
parsers_FasterRCNN(...)
Arguments
... |
arguments to pass |
Value
None
FilepathMixin
Description
Adds filepath method to parser
Usage
parsers_FilepathMixin(...)
Arguments
... |
arguments to pass |
Value
None
Imageid Mixin
Description
Adds imageid method to parser
Usage
parsers_ImageidMixin(...)
Arguments
... |
arguments to pass |
Value
None
IsCrowdsMixin
Description
Adds iscrowds method to parser
Usage
parsers_IsCrowdsMixin(...)
Arguments
... |
arguments to pass |
Value
None
LabelsMixin
Description
Adds labels method to parser
Usage
parsers_LabelsMixin(...)
Arguments
... |
arguments to pass |
Value
None
Mask RCNN
Description
Parser with required mixins for Mask RCNN.
Usage
parsers_MaskRCNN(...)
Arguments
... |
arguments to pass |
Value
None
MasksMixin
Description
Adds masks method to parser
Usage
parsers_MasksMixin(...)
Arguments
... |
arguments to pass |
Value
None
SizeMixin
Description
Adds image_width_height method to parser
Usage
parsers_SizeMixin(...)
Arguments
... |
arguments to pass |
Value
None
Voc parser
Description
Voc parser
Usage
parsers_voc(annotations_dir, images_dir, class_map, masks_dir = NULL)
Arguments
annotations_dir |
annotations_dir |
images_dir |
images_dir |
class_map |
class_map |
masks_dir |
masks_dir |
Value
None
Partial
Description
partial(func, *args, **keywords) - new function with partial application
Usage
partial(...)
Arguments
... |
additional arguments |
Value
None
Examples
## Not run:
generator = basic_generator(out_size = 64, n_channels = 3, n_extra_layers = 1)
critic = basic_critic(in_size = 64, n_channels = 3, n_extra_layers = 1,
act_cls = partial(nn$LeakyReLU, negative_slope = 0.2))
## End(Not run)
PCA
Description
Compute PCA of 'x' with 'k' dimensions.
Usage
pca(object, k = 3, convert = TRUE)
Arguments
object |
an object to apply PCA |
k |
number of dimensions |
convert |
to R matrix |
Value
tensor
Plot dicom
Description
Plot dicom
Usage
plot(x, y, ..., dpi = 100)
Arguments
x |
model |
y |
y axis |
... |
parameters to pass |
dpi |
dots per inch |
Value
None
Plot_bs_find
Description
Plot_bs_find
Usage
plot_bs_find(object, ..., dpi = 250)
Arguments
object |
model |
... |
additional arguments |
dpi |
dots per inch |
Value
None
Plot_confusion_matrix
Description
Plot the confusion matrix, with 'title' and using 'cmap'.
Usage
plot_confusion_matrix(
interp,
normalize = FALSE,
title = "Confusion matrix",
cmap = "Blues",
norm_dec = 2,
plot_txt = TRUE,
figsize = c(4, 4),
...,
dpi = 120
)
Arguments
interp |
interpretation object |
normalize |
normalize |
title |
title |
cmap |
color map |
norm_dec |
norm dec |
plot_txt |
plot text |
figsize |
plot size |
... |
additional parameters to pass |
dpi |
dots per inch |
Value
None
Examples
## Not run:
interp = ClassificationInterpretation_from_learner(model)
interp %>% plot_confusion_matrix(dpi = 90,figsize = c(6,6))
## End(Not run)
Plot_loss
Description
Plot the losses from 'skip_start' and onward
Usage
plot_loss(object, skip_start = 5, with_valid = TRUE, dpi = 200)
Arguments
object |
model |
skip_start |
n points to skip the start |
with_valid |
with validation |
dpi |
dots per inch |
Value
None
Plot_lr_find
Description
Plot the result of an LR Finder test (won't work if you didn't do 'lr_find(learn)' before)
Usage
plot_lr_find(object, skip_end = 5, dpi = 250)
Arguments
object |
model |
skip_end |
n points to skip the end |
dpi |
dots per inch |
Value
None
Plot_top_losses
Description
Plot_top_losses
Usage
plot_top_losses(interp, k, largest = TRUE, figsize = c(7, 5), ..., dpi = 90)
Arguments
interp |
interpretation object |
k |
number of images |
largest |
largest |
figsize |
plot size |
... |
additional parameters to pass |
dpi |
dots per inch |
Value
None
Examples
## Not run:
# get interperetation from learn object, the model.
interp = ClassificationInterpretation_from_learner(learn)
interp %>% plot_top_losses(k = 9, figsize = c(15,11))
## End(Not run)
Pow
Description
Pow
Usage
## S3 method for class 'torch.Tensor'
a ^ b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Pre_process_squad
Description
Pre_process_squad
Usage
pre_process_squad(row, hf_arch, hf_tokenizer)
Arguments
row |
row in dataframe |
hf_arch |
architecture |
hf_tokenizer |
tokenizer |
Value
None
Predict
Description
Prediction on 'item', fully decoded, loss function decoded and probabilities
Usage
## S3 method for class 'fastai.learner.Learner'
predict(object, row, ...)
Arguments
object |
the model |
row |
row |
... |
additional arguments to pass |
Value
data frame
Predict
Description
Prediction on 'item', fully decoded, loss function decoded and probabilities
Usage
## S3 method for class 'fastai.tabular.learner.TabularLearner'
predict(object, row, ...)
Arguments
object |
the model |
row |
row |
... |
additional arguments to pass |
Value
data frame
Perplexity
Description
Perplexity (exponential of cross-entropy loss) for Language Models
Usage
preplexity(...)
Arguments
... |
parameters to pass |
Value
None
Preprocess audio folder
Description
Preprocess audio files in 'path' in parallel using 'n_workers'
Usage
preprocess_audio_folder(
path,
folders = NULL,
output_dir = NULL,
sample_rate = 16000,
force_mono = TRUE,
crop_signal_to = NULL
)
Arguments
path |
directory, path |
folders |
folders |
output_dir |
output directory |
sample_rate |
sample rate |
force_mono |
force mono or not |
crop_signal_to |
int, crop signal |
Value
None
Print model
Description
Print model
Usage
## S3 method for class 'fastai.learner.Learner'
print(x, ...)
Arguments
x |
object |
... |
additional parameters to pass |
Value
None
Print tabular model
Description
Print tabular model
Usage
## S3 method for class 'fastai.tabular.learner.TabularLearner'
print(x, ...)
Arguments
x |
model |
... |
additional parameters to pass |
Value
None
Dicom
Description
prints dicom file
Usage
## S3 method for class 'pydicom.dataset.FileDataset'
print(x, ...)
Arguments
x |
dicom file |
... |
additional parameters to pass |
Value
None
Py_apply
Description
Pandas apply
Usage
py_apply(df, ...)
Arguments
df |
dataframe |
... |
additional arguments |
Value
dataframe
Python path
Description
Python path
Usage
python_path()
Value
None
Qhadam_step
Description
Qhadam_step
Usage
qhadam_step(p, lr, mom, sqr_mom, sqr_avg, nu_1, nu_2, step, grad_avg, eps, ...)
Arguments
p |
p |
lr |
learning rate |
mom |
momentum |
sqr_mom |
sqr momentum |
sqr_avg |
sqr average |
nu_1 |
nu_1 |
nu_2 |
nu_2 |
step |
step |
grad_avg |
gradient average |
eps |
epsilon |
... |
additional arguments to pass |
Value
None
Radam_step
Description
Step for RAdam with 'lr' on 'p'
Usage
radam_step(p, lr, mom, step, sqr_mom, grad_avg, sqr_avg, eps, beta, ...)
Arguments
p |
p |
lr |
learning rate |
mom |
momentum |
step |
step |
sqr_mom |
sqr momentum |
grad_avg |
grad average |
sqr_avg |
sqr average |
eps |
epsilon |
beta |
beta |
... |
additional arguments to pass |
Value
None
Ranger
Description
Convenience method for 'Lookahead' with 'RAdam'
Usage
ranger(
p,
lr,
mom = 0.95,
wd = 0.01,
eps = 1e-06,
sqr_mom = 0.99,
beta = 0,
decouple_wd = TRUE
)
Arguments
p |
p |
lr |
learning rate |
mom |
momentum |
wd |
weight decay |
eps |
epsilon |
sqr_mom |
sqr momentum |
beta |
beta |
decouple_wd |
decouple weight decay |
Value
None
Objects exported from other packages
Description
These objects are imported from other packages. Follow the links below to see their documentation.
- generics
- grDevices
- reticulate
- stats
- utils
Value
None
None
None
None
None
None
None
None
None
None
Replace_all_caps
Description
Replace tokens in ALL CAPS by their lower version and add 'TK_UP' before.
Usage
replace_all_caps(t)
Arguments
t |
text |
Value
string
Replace_maj
Description
Replace tokens in ALL CAPS by their lower version and add 'TK_UP' before.
Usage
replace_maj(t)
Arguments
t |
text |
Value
string
Replace_rep
Description
Replace repetitions at the character level: cccc – TK_REP 4 c
Usage
replace_rep(t)
Arguments
t |
text |
Value
string
Replace_wrep
Description
Replace word repetitions: word word word word – TK_WREP 4 word
Usage
replace_wrep(t)
Arguments
t |
text |
Value
string
Res_block_1d
Description
Resnet block as described in the paper.
Usage
res_block_1d(nf, ks = c(5, 3))
Arguments
nf |
number of features |
ks |
kernel size |
Value
block
Reshape
Description
resize x to (w,h)
Usage
reshape(x, h, w, resample = 0)
Arguments
x |
tensor |
h |
height |
w |
width |
resample |
resample value |
Value
None
Resize_max
Description
'resize' 'x' to 'max_px', or 'max_h', or 'max_w'
Usage
resize_max(img, resample = 0, max_px = NULL, max_h = NULL, max_w = NULL)
Arguments
img |
image |
resample |
resample value |
max_px |
max px |
max_h |
max height |
max_w |
max width |
Value
None
Resnet101
Description
ResNet-101 model from
Usage
resnet101(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
"Deep Residual Learning for Image Recognition" <https://arxiv.org/pdf/1512.03385.pdf>
Value
model
Resnet152
Description
Resnet152
Usage
resnet152(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
"Deep Residual Learning for Image Recognition" <https://arxiv.org/pdf/1512.03385.pdf>
Value
model
Resnet18
Description
Resnet18
Usage
resnet18(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
"Deep Residual Learning for Image Recognition" <https://arxiv.org/pdf/1512.03385.pdf>
Value
model
Resnet34
Description
ResNet-34 model from
Usage
resnet34(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
"Deep Residual Learning for Image Recognition" <https://arxiv.org/pdf/1512.03385.pdf>
Value
model
Resnet50
Description
Resnet50
Usage
resnet50(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
"Deep Residual Learning for Image Recognition" <https://arxiv.org/pdf/1512.03385.pdf>
Value
model
Resnet_generator
Description
Resnet_generator
Usage
resnet_generator(
ch_in,
ch_out,
n_ftrs = 64,
norm_layer = NULL,
dropout = 0,
n_blocks = 9,
pad_mode = "reflection"
)
Arguments
ch_in |
input |
ch_out |
output |
n_ftrs |
filter |
norm_layer |
normalziation layer |
dropout |
dropout rate |
n_blocks |
number of blocks |
pad_mode |
paddoing mode |
Value
None
Retinanet module
Description
Retinanet module
Usage
retinanet_()
Value
None
Reverse_text
Description
Reverse_text
Usage
reverse_text(x)
Arguments
x |
text |
Value
string
Rgb2hsv
Description
Converts a RGB image to an HSV image.
Usage
rgb2hsv(img)
Arguments
img |
image object |
Details
Note: Will not work on logit space images.
Value
None
Rm_useless_spaces
Description
Remove multiple spaces
Usage
rm_useless_spaces(t)
Arguments
t |
text |
Value
string
Examples
## Not run:
rm_useless_spaces('hello, Sir!')
## End(Not run)
Rms_prop_step
Description
Step for SGD with momentum with 'lr'
Usage
rms_prop_step(p, lr, sqr_avg, eps, grad_avg = NULL, ...)
Arguments
p |
p |
lr |
learning rate |
sqr_avg |
sqr average |
eps |
epsilon |
grad_avg |
grad average |
... |
additional arguments to pass |
Value
None
RMSE
Description
Root mean squared error
Usage
rmse(preds, targs)
Arguments
preds |
predictions |
targs |
targets |
Value
None
Examples
## Not run:
model = dls %>% tabular_learner(layers=c(200,100,100,200),
metrics = list(mse(),rmse()) )
## End(Not run)
Rotate_mat
Description
Return a random rotation matrix with 'max_deg' and 'p'
Usage
rotate_mat(x, max_deg = 10, p = 0.5, draw = NULL, batch = FALSE)
Arguments
x |
tensor |
max_deg |
max_deg |
p |
probability |
draw |
draw |
batch |
batch |
Value
None
Round
Description
Round
Usage
## S3 method for class 'torch.Tensor'
round(x, digits = 0)
Arguments
x |
tensor |
digits |
decimal |
Value
tensor
Round
Description
Round
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
round(x, digits = 0)
Arguments
x |
tensor |
digits |
decimal |
Value
tensor
Sequential
Description
Sequential
Usage
sequential(...)
Arguments
... |
parameters to pass |
Value
None
Set freeze model
Description
Set freeze model
Usage
set_freeze_model(m, rg)
Arguments
m |
parameters |
rg |
rg |
Value
None
Set_item_pg
Description
Set_item_pg
Usage
set_item_pg(pg, k, v)
Arguments
pg |
pg |
k |
k |
v |
v |
Value
None
Setup_aug_tfms
Description
Go through 'tfms' and combines together affine/coord or lighting transforms
Usage
setup_aug_tfms(tfms)
Arguments
tfms |
transformations |
Value
None
Sgd_step
Description
Sgd_step
Usage
sgd_step(p, lr, ...)
Arguments
p |
p |
lr |
learning rate |
... |
additional arguments to pass |
Value
None
Examples
## Not run:
tst_param = function(val, grad = NULL) {
"Create a tensor with `val` and a gradient of `grad` for testing"
res = tensor(val) %>% float()
if(is.null(grad)) {
grad = tensor(val / 10)
} else {
grad = tensor(grad)
}
res$grad = grad %>% float()
res
}
p = tst_param(1., 0.1)
sgd_step(p, 1.)
## End(Not run)
Shap module
Description
Shap module
Usage
shap()
Value
None
Shape
Description
Shape
Usage
shape(img)
Arguments
img |
image |
Value
None
Show
Description
Adds functionality to view dicom images where each file may have more than 1 frame
Usage
show(img, frames = 1, scale = TRUE, ...)
Arguments
img |
image object |
frames |
number of frames |
scale |
scale |
... |
additional arguments |
Value
None
Show_array
Description
Show an array on 'ax'.
Usage
show_array(
array,
ax = NULL,
figsize = NULL,
title = NULL,
ctx = NULL,
tx = NULL
)
Arguments
array |
R array |
ax |
axis |
figsize |
figure size |
title |
title, text |
ctx |
ctx |
tx |
tx |
Value
None
Examples
## Not run:
arr = as.array(1:10)
show_array(arr,title = 'My R array') %>% plot(dpi = 200)
## End(Not run)
Show_batch
Description
Show_batch
Usage
show_batch(
dls,
b = NULL,
max_n = 9,
ctxs = NULL,
figsize = c(6, 6),
show = TRUE,
unique = FALSE,
dpi = 120,
...
)
Arguments
dls |
dataloader object |
b |
defaults to one_batch |
max_n |
maximum images |
ctxs |
ctxs parameter |
figsize |
figure size |
show |
show or not |
unique |
unique images |
dpi |
dots per inch |
... |
additional arguments to pass |
Value
None
Examples
## Not run:
dls %>% show_batch()
## End(Not run)
Show_image
Description
Show a PIL or PyTorch image on 'ax'.
Usage
show_image(
im,
ax = NULL,
figsize = NULL,
title = NULL,
ctx = NULL,
cmap = NULL,
norm = NULL,
aspect = NULL,
interpolation = NULL,
alpha = NULL,
vmin = NULL,
vmax = NULL,
origin = NULL,
extent = NULL
)
Arguments
im |
im |
ax |
axis |
figsize |
figure size |
title |
title |
ctx |
ctx |
cmap |
color maps |
norm |
normalization |
aspect |
aspect |
interpolation |
interpolation |
alpha |
alpha value |
vmin |
value min |
vmax |
value max |
origin |
origin |
extent |
extent |
Show_images
Description
Show all images 'ims' as subplots with 'rows' using 'titles'
Usage
show_images(
ims,
nrows = 1,
ncols = NULL,
titles = NULL,
figsize = NULL,
imsize = 3,
add_vert = 0
)
Arguments
ims |
images |
nrows |
number of rows |
ncols |
number of columns |
titles |
titles |
figsize |
figure size |
imsize |
image size |
add_vert |
add vertical |
Value
None
Show_preds
Description
Show_preds
Usage
show_preds(
predictions,
idx,
class_map = NULL,
denormalize_fn = denormalize_imagenet(),
display_label = TRUE,
display_bbox = TRUE,
display_mask = TRUE,
ncols = 1,
figsize = NULL,
show = FALSE,
dpi = 100
)
Arguments
predictions |
provide list of raw predictions |
idx |
image indices |
class_map |
class_map |
denormalize_fn |
denormalize_fn |
display_label |
display_label |
display_bbox |
display_bbox |
display_mask |
display_mask |
ncols |
ncols |
figsize |
figsize |
show |
show |
dpi |
dots per inch |
Value
None
Show_results
Description
Show some predictions on 'ds_idx'-th dataset or 'dl'
Usage
show_results(
object,
ds_idx = 1,
dl = NULL,
max_n = 9,
shuffle = TRUE,
dpi = 90,
...
)
Arguments
object |
model |
ds_idx |
ds by index |
dl |
dataloader |
max_n |
maximum number of images |
shuffle |
shuffle or not |
dpi |
dots per inch |
... |
additional arguments |
Value
None
Show_samples
Description
Show_samples
Usage
show_samples(
dls,
idx,
class_map = NULL,
denormalize_fn = denormalize_imagenet(),
display_label = TRUE,
display_bbox = TRUE,
display_mask = TRUE,
ncols = 1,
figsize = NULL,
show = FALSE,
dpi = 100
)
Arguments
dls |
dataloader |
idx |
image indices |
class_map |
class_map |
denormalize_fn |
denormalize_fn |
display_label |
display_label |
display_bbox |
display_bbox |
display_mask |
display_mask |
ncols |
ncols |
figsize |
figsize |
show |
show |
dpi |
dots per inch |
Value
None
Sigmoid
Description
Same as 'torch$sigmoid', plus clamping to '(eps,1-eps)
Usage
sigmoid(input, eps = 1e-07)
Arguments
input |
inputs |
eps |
epsilon |
Value
None
Sigmoid_
Description
Same as 'torch$sigmoid_', plus clamping to '(eps,1-eps)
Usage
sigmoid_(input, eps = 1e-07)
Arguments
input |
input |
eps |
eps |
Value
None
Sigmoid_range
Description
Sigmoid function with range '(low, high)'
Usage
sigmoid_range(x, low, high)
Arguments
x |
tensor |
low |
low value |
high |
high value |
Value
None
Sin
Description
Sin
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
sin(x)
Arguments
x |
tensor |
Value
tensor
Sin
Description
Sin
Usage
## S3 method for class 'torch.Tensor'
sin(x)
Arguments
x |
tensor |
Value
tensor
Sinh
Description
Sinh
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
sinh(x)
Arguments
x |
tensor |
Value
tensor
Skm to fastai
Description
Convert 'func' from sklearn$metrics to a fastai metric
Usage
skm_to_fastai(
func,
is_class = TRUE,
thresh = NULL,
axis = -1,
activation = NULL,
...
)
Arguments
func |
function |
is_class |
is classification or not |
thresh |
threshold point |
axis |
axis |
activation |
activation |
... |
additional arguments to pass |
Value
None
Slice
Description
Slice
Usage
slice(...)
Arguments
... |
additional arguments |
Details
slice(start, stop[, step]) Create a slice object. This is used for extended slicing (e.g. a[0:10:2]).
Value
sliced object
Sort
Description
Sort
Usage
## S3 method for class 'torch.Tensor'
sort(x, decreasing = FALSE, ...)
Arguments
x |
tensor |
decreasing |
the order |
... |
additional parameters to pass |
Sort
Description
Sort
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
sort(x, decreasing = FALSE, ...)
Arguments
x |
tensor |
decreasing |
the order |
... |
additional parameters to pass |
Value
tensor
Spec_add_spaces
Description
Add spaces around / and #
Usage
spec_add_spaces(t)
Arguments
t |
text |
Value
string
Sqrt
Description
Sqrt
Usage
## S3 method for class 'torch.Tensor'
sqrt(x)
Arguments
x |
tensor |
Value
tensor
Sqrt
Description
Sqrt
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
sqrt(x)
Arguments
x |
tensor |
Value
tensor
Squeezenet1_0
Description
SqueezeNet model architecture from the '"SqueezeNet: AlexNet-level
Usage
squeezenet1_0(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
accuracy with 50x fewer parameters and <0.5MB model size" <https://arxiv.org/abs/1602.07360>'_ paper.
Value
model
Squeezenet1_1
Description
SqueezeNet 1.1 model from the 'official SqueezeNet repo
Usage
squeezenet1_1(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
<https://github.com/DeepScale/SqueezeNet/tree/master/SqueezeNet_v1.1>'_. SqueezeNet 1.1 has 2.4x less computation and slightly fewer parameters than SqueezeNet 1.0, without sacrificing accuracy.
Value
model
Stack_train_valid
Description
Stack df_train and df_valid, adds 'valid_col'=TRUE/FALSE for df_valid/df_train
Usage
stack_train_valid(df_train, df_valid)
Arguments
df_train |
train data |
df_valid |
validation data |
Value
data frame
Step_stat
Description
Register the number of steps done in 'state' for 'p'
Usage
step_stat(p, step = 0, ...)
Arguments
p |
p |
step |
step |
... |
additional args to pass |
Value
None
Sub
Description
Sub
Usage
## S3 method for class 'torch.Tensor'
a - b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Sub
Description
Sub
Usage
## S3 method for class 'fastai.torch_core.TensorMask'
a - b
Arguments
a |
tensor |
b |
tensor |
Value
tensor
Subplots
Description
Subplots
Usage
subplots(nrows = 2, ncols = 2, figsize = NULL, imsize = 4)
Arguments
nrows |
number of rows |
ncols |
number of columns |
figsize |
figure size |
imsize |
image size |
Value
plot object
Summarization_splitter
Description
Custom param splitter for summarization models
Usage
summarization_splitter(m, arch)
Arguments
m |
splitter parameter |
arch |
architecture |
Value
None
Summary
Description
Summary
Usage
## S3 method for class 'fastai.learner.Learner'
summary(object, ...)
Arguments
object |
model |
... |
additional arguments to pass |
Value
None
Examples
## Not run:
summary(model)
## End(Not run)
Summary
Description
Print a summary of 'm' using a output text width of 'n' chars
Usage
## S3 method for class 'fastai.tabular.learner.TabularLearner'
summary(object, ...)
Arguments
object |
model |
... |
additional parameters to pass |
Value
None
Summary_plot
Description
Displays the SHAP values (which can be interpreted for feature importance)
Usage
summary_plot(object, dpi = 200, ...)
Arguments
object |
ShapInterpretation object |
dpi |
dots per inch |
... |
additional arguments |
Value
None
Swish
Description
Swish
Usage
swish(x, inplace = FALSE)
Arguments
x |
tensor |
inplace |
inplace or not |
Value
None
Tabular
Description
Tabular
Usage
tabular()
Value
None
Tabular_config
Description
Convenience function to easily create a config for 'TabularModel'
Usage
tabular_config(
ps = NULL,
embed_p = 0,
y_range = NULL,
use_bn = TRUE,
bn_final = FALSE,
bn_cont = TRUE,
act_cls = nn()$ReLU(inplace = TRUE)
)
Arguments
ps |
ps |
embed_p |
embed proportion |
y_range |
y_range |
use_bn |
use batch normalization |
bn_final |
batch normalization final |
bn_cont |
batch normalization |
act_cls |
activation |
Value
None
Tabular learner
Description
Get a 'Learner' using 'dls', with 'metrics', including a 'TabularModel' created using the remaining params.
Usage
tabular_learner(
dls,
layers = NULL,
emb_szs = NULL,
config = NULL,
n_out = NULL,
y_range = NULL,
loss_func = NULL,
opt_func = Adam(),
lr = 0.001,
splitter = trainable_params(),
cbs = NULL,
metrics = NULL,
path = NULL,
model_dir = "models",
wd = NULL,
wd_bn_bias = FALSE,
train_bn = TRUE,
moms = list(0.95, 0.85, 0.95)
)
Arguments
dls |
It is a DataLoaders object. |
layers |
layers |
emb_szs |
emb_szs |
config |
config |
n_out |
n_out |
y_range |
y_range |
loss_func |
It can be any loss function you like. |
opt_func |
It will be used to create an optimizer when Learner.fit is called. |
lr |
It is learning rate. |
splitter |
It is a function that takes self.model and returns a list of parameter groups (or just one parameter group if there are no different parameter groups) |
cbs |
It is one or a list of Callbacks to pass to the Learner. |
metrics |
It is an optional list of metrics, that can be either functions or Metrics. |
path |
İt is used to save and/or load models.Often path will be inferred from dls, but you can override it or pass a Path object to model_dir. Make sure you can write in path/model_dir! |
model_dir |
İt is used to save and/or load models.Often path will be inferred from dls, but you can override it or pass a Path object to model_dir. Make sure you can write in path/model_dir! |
wd |
It is the default weight decay used when training the model. |
wd_bn_bias |
It controls if weight decay is applied to BatchNorm layers and bias. |
train_bn |
It controls if BatchNorm layers are trained even when they are supposed to be frozen according to the splitter. |
moms |
The default momentums used in Learner.fit_one_cycle. |
Value
learner object
Tar_extract_at_filename
Description
Extract 'fname' to 'dest'/'fname.name' folder using 'tarfile'
Usage
tar_extract_at_filename(fname, dest)
Arguments
fname |
folder name |
dest |
destination |
Value
None
Tensor
Description
Like 'torch()$as_tensor', but handle lists too, and can pass multiple vector elements directly.
Usage
tensor(...)
Arguments
... |
image |
Value
None
Test_loader
Description
Data loader. Combines a dataset and a sampler, and provides an iterable over
Usage
test_loader()
Details
the given dataset. The :class:'~torch.utils.data.DataLoader' supports both map-style and iterable-style datasets with single- or multi-process loading, customizing loading order and optional automatic batching (collation) and memory pinning. See :py:mod:'torch.utils.data' documentation page for more details.
Value
loader
Text module
Description
Text module
Usage
text()
Value
None
Text_classifier_learner
Description
Create a 'Learner' with a text classifier from 'dls' and 'arch'.
Usage
text_classifier_learner(
dls,
arch,
seq_len = 72,
config = NULL,
backwards = FALSE,
pretrained = TRUE,
drop_mult = 0.5,
n_out = NULL,
lin_ftrs = NULL,
ps = NULL,
max_len = 1440,
y_range = NULL,
loss_func = NULL,
opt_func = Adam(),
lr = 0.001,
splitter = trainable_params,
cbs = NULL,
metrics = NULL,
path = NULL,
model_dir = "models",
wd = NULL,
wd_bn_bias = FALSE,
train_bn = TRUE,
moms = list(0.95, 0.85, 0.95)
)
Arguments
dls |
dls |
arch |
arch |
seq_len |
seq_len |
config |
config |
backwards |
backwards |
pretrained |
pretrained |
drop_mult |
drop_mult |
n_out |
n_out |
lin_ftrs |
lin_ftrs |
ps |
ps |
max_len |
max_len |
y_range |
y_range |
loss_func |
loss_func |
opt_func |
opt_func |
lr |
lr |
splitter |
splitter |
cbs |
cbs |
metrics |
metrics |
path |
path |
model_dir |
model_dir |
wd |
wd |
wd_bn_bias |
wd_bn_bias |
train_bn |
train_bn |
moms |
moms |
Value
None
Timm module
Description
Timm module
Usage
timm()
Value
None
Timm_learner
Description
Build a convnet style learner from 'dls' and 'arch' using the 'timm' library
Usage
timm_learner(dls, arch, ...)
Arguments
dls |
dataloader |
arch |
model architecture |
... |
additional arguments |
Value
None
Timm models
Description
Timm models
Usage
timm_list_models(...)
Arguments
... |
parameters to pass |
Value
vector
Timeseries module
Description
Timeseries module
Usage
tms()
Value
None
To_bytes_format
Description
Convert to bytes, default to PNG format
Usage
to_bytes_format(img, format = "png")
Arguments
img |
image |
format |
format |
Value
None
To_image
Description
Convert a tensor or array to a PIL int8 Image
Usage
to_image(x)
Arguments
x |
tensor |
Value
None
To matrix
Description
To matrix
Usage
to_matrix(obj, matrix = TRUE)
Arguments
obj |
learner/model |
matrix |
bool, to R matrix |
To_thumb
Description
Same as 'thumbnail', but uses a copy
Usage
to_thumb(img, h, w = NULL)
Arguments
img |
image |
h |
height |
w |
width |
Value
None
Learn to XLA
Description
Distribute the training across TPUs
Usage
to_xla(object)
Arguments
object |
learner / model |
Value
None
Tokenize1
Description
Call 'TokenizeWithRules' with a single text
Usage
tokenize1(text, tok, rules = NULL, post_rules = NULL)
Arguments
text |
text |
tok |
tok |
rules |
rules |
post_rules |
post_rules |
Value
None
Tokenize_csv
Description
Tokenize texts in the 'text_cols' of the csv 'fname' in parallel using 'n_workers'
Usage
tokenize_csv(
fname,
text_cols,
outname = NULL,
n_workers = 4,
rules = NULL,
mark_fields = NULL,
tok = NULL,
header = "infer",
chunksize = 50000
)
Arguments
fname |
file name |
text_cols |
text columns |
outname |
outname |
n_workers |
numeber of workers |
rules |
rules |
mark_fields |
mark fields |
tok |
tokenizer |
header |
header |
chunksize |
chunk size |
Value
None
Tokenize_df
Description
Tokenize texts in 'df[text_cols]' in parallel using 'n_workers'
Usage
tokenize_df(
df,
text_cols,
n_workers = 6,
rules = NULL,
mark_fields = NULL,
tok = NULL,
tok_text_col = "text"
)
Arguments
df |
data frame |
text_cols |
text columns |
n_workers |
number of workers |
rules |
rules |
mark_fields |
mark_fields |
tok |
tokenizer |
tok_text_col |
tok_text_col |
Value
None
Tokenize_files
Description
Tokenize text 'files' in parallel using 'n_workers'
Usage
tokenize_files(
files,
path,
output_dir,
output_names = NULL,
n_workers = 6,
rules = NULL,
tok = NULL,
encoding = "utf8",
skip_if_exists = FALSE
)
Arguments
files |
files |
path |
path |
output_dir |
output_dir |
output_names |
output_names |
n_workers |
n_workers |
rules |
rules |
tok |
tokenizer |
encoding |
encoding |
skip_if_exists |
skip_if_exists |
Value
None
Tokenize_folder
Description
Tokenize text files in 'path' in parallel using 'n_workers'
Usage
tokenize_folder(
path,
extensions = NULL,
folders = NULL,
output_dir = NULL,
skip_if_exists = TRUE,
output_names = NULL,
n_workers = 6,
rules = NULL,
tok = NULL,
encoding = "utf8"
)
Arguments
path |
path |
extensions |
extensions |
folders |
folders |
output_dir |
output_dir |
skip_if_exists |
skip_if_exists |
output_names |
output_names |
n_workers |
number of workers |
rules |
rules |
tok |
tokenizer |
encoding |
encoding |
Value
None
Tokenize_texts
Description
Tokenize 'texts' in parallel using 'n_workers'
Usage
tokenize_texts(texts, n_workers = 6, rules = NULL, tok = NULL)
Arguments
texts |
texts |
n_workers |
n_workers |
rules |
rules |
tok |
tok |
Value
None
Top_k_accuracy
Description
Computes the Top-k accuracy ('targ' is in the top 'k' predictions of 'inp')
Usage
top_k_accuracy(inp, targ, k = 5, axis = -1)
Arguments
inp |
predictions |
targ |
targets |
k |
k |
axis |
axis |
Value
None
Examples
## Not run:
loaders = loaders()
data = Data_Loaders(loaders['train'], loaders['valid'])$cuda()
model = nn$Sequential() +
nn$Flatten() +
nn$Linear(28L * 28L, 10L)
metrics = list(accuracy,top_k_accuracy)
learn = Learner(data, model, loss_func = F$cross_entropy, opt_func = Adam,
metrics = metrics)
## End(Not run)
Builtins module
Description
Builtins module
Usage
torch()
Value
None
Total_params
Description
Give the number of parameters of a module and if it's trainable or not
Usage
total_params(m)
Arguments
m |
m parameter |
Value
None
Train_loader
Description
Data loader. Combines a dataset and a sampler, and provides an iterable over
Usage
train_loader()
Details
the given dataset. The :class:'~torch.utils.data.DataLoader' supports both map-style and iterable-style datasets with single- or multi-process loading, customizing loading order and optional automatic batching (collation) and memory pinning.
Value
loader
Trainable_params
Description
Return all trainable parameters of 'm'
Usage
trainable_params(m)
Arguments
m |
trainable parameters |
Value
None
Transformers
Description
Transformers
Usage
transformers()
Value
None
Trunc_normal_
Description
Truncated normal initialization (approximation)
Usage
trunc_normal_(x, mean = 0, std = 1)
Arguments
x |
tensor |
mean |
mean |
std |
standard deviation |
Value
tensor
Unet_config
Description
Convenience function to easily create a config for 'DynamicUnet'
Usage
unet_config(
blur = FALSE,
blur_final = TRUE,
self_attention = FALSE,
y_range = NULL,
last_cross = TRUE,
bottle = FALSE,
act_cls = nn()$ReLU,
init = nn()$init$kaiming_normal_,
norm_type = NULL
)
Arguments
blur |
blur is used to avoid checkerboard artifacts at each layer. |
blur_final |
blur final is specific to the last layer. |
self_attention |
self_attention determines if we use a self attention layer at the third block before the end. |
y_range |
If y_range is passed, the last activations go through a sigmoid rescaled to that range. |
last_cross |
last cros |
bottle |
bottle |
act_cls |
activation |
init |
initializer |
norm_type |
normalization type |
Value
None
Unet_learner
Description
Build a unet learner from 'dls' and 'arch'
Usage
unet_learner(dls, arch, ...)
Arguments
dls |
dataloader |
arch |
architecture |
... |
additional arguments |
Value
None
Unfreeze a model
Description
Unfreeze a model
Usage
unfreeze(object, ...)
Arguments
object |
A model |
... |
Additional parameters |
Value
None
Examples
## Not run:
learnR %>% unfreeze()
## End(Not run)
Uniform_blur2d
Description
Uniformly apply blurring
Usage
uniform_blur2d(x, s)
Arguments
x |
image |
s |
effect |
Value
None
Upit module
Description
Upit module
Usage
upit()
Value
None
Vgg11_bn
Description
VGG 11-layer model (configuration "A") with batch normalization
Usage
vgg11_bn(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
"Very Deep Convolutional Networks For Large-Scale Image Recognition" <https://arxiv.org/pdf/1409.1556.pdf>
Value
model
Vgg13_bn
Description
VGG 13-layer model (configuration "B") with batch normalization
Usage
vgg13_bn(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
"Very Deep Convolutional Networks For Large-Scale Image Recognition" <https://arxiv.org/pdf/1409.1556.pdf>
Value
model
Vgg16_bn
Description
VGG 16-layer model (configuration "D") with batch normalization
Usage
vgg16_bn(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
"Very Deep Convolutional Networks For Large-Scale Image Recognition" <https://arxiv.org/pdf/1409.1556.pdf>
Value
model
Vgg19_bn
Description
VGG 19-layer model (configuration 'E') with batch normalization
Usage
vgg19_bn(pretrained = FALSE, progress)
Arguments
pretrained |
pretrained or not |
progress |
to see progress bar or not |
Details
"Very Deep Convolutional Networks For Large-Scale Image Recognition" <https://arxiv.org/pdf/1409.1556.pdf>
Value
model
Vision module
Description
Vision module
Usage
vision()
Value
None
Vleaky_relu
Description
'F$leaky_relu' with 0.3 slope
Usage
vleaky_relu(input, inplace = TRUE)
Arguments
input |
inputs |
inplace |
inplace or not |
Value
None
Wandb module
Description
Wandb module
Usage
wandb()
Value
None
Waterfall_plot
Description
Plots an explanation of a single prediction as a waterfall plot. Accepts a row_index and class_id.
Usage
waterfall_plot(object, row_idx = NULL, class_id = 0, dpi = 200, ...)
Arguments
object |
ShapInterpretation object |
row_idx |
is the index of the row chosen in test_data to be analyzed, which defaults to zero. |
class_id |
Accepts a class_id which is used to indicate the class of interest for a classification model. It can either be an int or str representation for a class of choice. |
dpi |
dots per inch |
... |
additional arguments |
Value
None
Weight_decay
Description
Weight decay as decaying 'p' with 'lr*wd'
Usage
weight_decay(p, lr, wd, do_wd = TRUE, ...)
Arguments
p |
p |
lr |
learning rate |
wd |
weight decay |
do_wd |
do_wd |
... |
additional args to pass |
Value
None
Examples
## Not run:
tst_param = function(val, grad = NULL) {
"Create a tensor with `val` and a gradient of `grad` for testing"
res = tensor(val) %>% float()
if(is.null(grad)) {
grad = tensor(val / 10)
} else {
grad = tensor(grad)
}
res$grad = grad %>% float()
res
}
p = tst_param(1., 0.1)
weight_decay(p, 1., 0.1)
## End(Not run)
Abdomen soft
Description
Abdomen soft
Usage
win_abdoment_soft()
Value
list
Brain
Description
Brain
Usage
win_brain()
Value
list
Brain bone
Description
Brain bone
Usage
win_brain_bone()
Value
list
Brain soft
Description
Brain soft
Usage
win_brain_soft()
Value
list
Liver
Description
Liver
Usage
win_liver()
Value
list
Lungs
Description
Lungs
Usage
win_lungs()
Value
list
Mediastinum
Description
Mediastinum
Usage
win_mediastinum()
Value
list
Spine bone
Description
Spine bone
Usage
win_spine_bone()
Value
list
Spine soft
Description
Spine soft
Usage
win_spine_soft()
Value
list
Stroke
Description
Stroke
Usage
win_stroke()
Value
list
Subdural
Description
Subdural
Usage
win_subdural()
Value
list
XLA
Description
XLA
Usage
xla()
Value
None
Xresnet101
Description
Load model architecture
Usage
xresnet101(...)
Arguments
... |
parameters to pass |
Value
model
Xresnet152
Description
Load model architecture
Usage
xresnet152(...)
Arguments
... |
parameters to pass |
Value
model
Xresnet18
Description
Load model architecture
Usage
xresnet18(...)
Arguments
... |
parameters to pass |
Value
model
Xresnet18_deep
Description
Load model architecture
Usage
xresnet18_deep(...)
Arguments
... |
parameters to pass |
Value
model
Xresnet18_deeper
Description
Load model architecture
Usage
xresnet18_deeper(...)
Arguments
... |
parameters to pass |
Value
model
Xresnet34
Description
Load model architecture
Usage
xresnet34(...)
Arguments
... |
parameters to pass |
Value
model
Xresnet34_deep
Description
Load model architecture
Usage
xresnet34_deep(...)
Arguments
... |
parameters to pass |
Value
model
Xresnet34_deeper
Description
Load model architecture
Usage
xresnet34_deeper(...)
Arguments
... |
parameters to pass |
Value
model
Xresnet50
Description
Load model architecture
Usage
xresnet50(...)
Arguments
... |
parameters to pass |
Value
model
Xresnet50_deep
Description
Load model architecture
Usage
xresnet50_deep(...)
Arguments
... |
parameters to pass |
Value
model
Xresnet50_deeper
Description
Load model architecture
Usage
xresnet50_deeper(...)
Arguments
... |
parameters to pass |
Value
model
xresnext101
Description
Load model architecture
Usage
xresnext101(...)
Arguments
... |
parameters to pass |
Value
model
xresnext18
Description
Load model architecture
Usage
xresnext18(...)
Arguments
... |
parameters to pass |
Value
model
xresnext34
Description
Load model architecture
Usage
xresnext34(...)
Arguments
... |
parameters to pass |
Value
model
xresnext50
Description
Load model architecture
Usage
xresnext50(...)
Arguments
... |
parameters to pass |
Value
model
xse_resnet101
Description
Load model architecture
Usage
xse_resnet101(...)
Arguments
... |
parameters to pass |
Value
model
xse_resnet152
Description
Load model architecture
Usage
xse_resnet152(...)
Arguments
... |
parameters to pass |
Value
model
xse_resnet18
Description
Load model architecture
Usage
xse_resnet18(...)
Arguments
... |
parameters to pass |
Value
model
xse_resnet34
Description
Load model architecture
Usage
xse_resnet34(...)
Arguments
... |
parameters to pass |
Value
model
xse_resnet50
Description
Load model architecture
Usage
xse_resnet50(...)
Arguments
... |
parameters to pass |
Value
model
xse_resnext101
Description
Load model architecture
Usage
xse_resnext101(...)
Arguments
... |
parameters to pass |
Value
model
xse_resnext18
Description
Load model architecture
Usage
xse_resnext18(...)
Arguments
... |
parameters to pass |
Value
model
xse_resnext18_deep
Description
Load model architecture
Usage
xse_resnext18_deep(...)
Arguments
... |
parameters to pass |
Value
model
xse_resnext18_deeper
Description
Load model architecture
Usage
xse_resnext18_deeper(...)
Arguments
... |
parameters to pass |
Value
model
xse_resnext34
Description
Load model architecture
Usage
xse_resnext34(...)
Arguments
... |
parameters to pass |
Value
model
xse_resnext34_deep
Description
Load model architecture
Usage
xse_resnext34_deep(...)
Arguments
... |
parameters to pass |
Value
model
xse_resnext34_deeper
Description
Load model architecture
Usage
xse_resnext34_deeper(...)
Arguments
... |
parameters to pass |
Value
model
xse_resnext50
Description
Load model architecture
Usage
xse_resnext50(...)
Arguments
... |
parameters to pass |
Value
model
xse_resnext50_deep
Description
Load model architecture
Usage
xse_resnext50_deep(...)
Arguments
... |
parameters to pass |
Value
model
xse_resnext50_deeper
Description
Load model architecture
Usage
xse_resnext50_deeper(...)
Arguments
... |
parameters to pass |
Value
model
xsenet154
Description
Load model architecture
Usage
xsenet154(...)
Arguments
... |
parameters to pass |
Value
model
Zoom
Description
Zoom
Usage
zoom(img, ratio)
Arguments
img |
image files |
ratio |
ratio |
Value
image
Zoom_mat
Description
Return a random zoom matrix with 'max_zoom' and 'p'
Usage
zoom_mat(
x,
min_zoom = 1,
max_zoom = 1.1,
p = 0.5,
draw = NULL,
draw_x = NULL,
draw_y = NULL,
batch = FALSE
)
Arguments
x |
tensor |
min_zoom |
minimum zoom |
max_zoom |
maximum zoom |
p |
probability |
draw |
draw |
draw_x |
draw x |
draw_y |
draw y |
batch |
batch |
Value
None