Skip to main content

One line functions for common tasks

Project description

Utilities for simple needs

torch snippets does a lot of default importing for you

Whether it is numpy, pandas, matplotlib or the useful functions that are mentioned below Simply call

from torch_snippets import *

All the imports are lightweight and thus should not take more than a couple of seconds

Auxiliary Functions

There are simple functions that are overloaded to take inputs and perform repetitive tasks that usually take a few lines to write

Images

show, inspect, Glob, read, resize, rotate

Files and Paths

stem, Glob, parent, name, fname,

makedir, zip_files, unzip_file,

find, extn,

readlines, writelines

Lists

L, flatten

Dump and load python objects

loaddill,dumpdill

Misc

Tqdm, Timer, randint, Logger

Sets

unique, diff, choose, common

Pytorch Modules

Reshape and Permute (nn.Modules)

Report as Pytorch Lightning Callback

LightningReport

Charts

Chart from altair

and many more to come...

Install

pip install torch_snippets

Usage

%time from torch_snippets import *
CPU times: user 1.79 s, sys: 672 ms, total: 2.46 s
Wall time: 2.62 s
dir()
['AttrDict',
 'B',
 'BB',
 'Blank',
 'C',
 'Chart',
 'DataLoader',
 'Dataset',
 'Debug',
 'E',
 'Excep',
 'F',
 'Float',
 'Glob',
 'Image',
 'ImportEnum',
 'In',
 'Inf',
 'Info',
 'Int',
 'L',
 'LightningReport',
 'NullType',
 'Out',
 'P',
 'PIL',
 'Path',
 'Permute',
 'PrettyString',
 'Report',
 'Reshape',
 'Self',
 'ShowPrint',
 'Stateful',
 'Str',
 'StrEnum',
 'T',
 'Timer',
 'Tqdm',
 'Warn',
 '_',
 '__',
 '___',
 '__builtin__',
 '__builtins__',
 '__doc__',
 '__loader__',
 '__name__',
 '__package__',
 '__spec__',
 '_dh',
 '_i',
 '_i1',
 '_i2',
 '_ih',
 '_ii',
 '_iii',
 '_oh',
 'add',
 'add_props',
 'alt',
 'anno_ret',
 'annotations',
 'arg0',
 'arg1',
 'arg2',
 'arg3',
 'arg4',
 'argnames',
 'argwhere',
 'attrdict',
 'basic_repr',
 'bbfy',
 'bind',
 'camel2snake',
 'charts',
 'choose',
 'chunked',
 'class2attr',
 'common',
 'compose',
 'copy_func',
 'crop_from_bb',
 'custom_dir',
 'cv2',
 'cycle',
 'defaults',
 'detuplify',
 'device',
 'df2bbs',
 'diff',
 'display',
 'dumpdill',
 'enlarge_bbs',
 'eq',
 'even_mults',
 'exec_local',
 'exit',
 'extn',
 'fastcores',
 'fastuple',
 'filter_dict',
 'filter_ex',
 'filter_keys',
 'filter_values',
 'find',
 'first',
 'flatten',
 'fname',
 'fname2',
 'ge',
 'gen',
 'get_class',
 'get_ipython',
 'getattrs',
 'glob',
 'groupby',
 'gt',
 'hasattrs',
 'ifnone',
 'ignore_exceptions',
 'in_',
 'inspect',
 'instantiate',
 'inum_methods',
 'is_',
 'is_array',
 'is_not',
 'isdir',
 'jitter',
 'last_index',
 'le',
 'line',
 'lines',
 'listify',
 'load_torch_model_weights_to',
 'loaddill',
 'loader',
 'logger',
 'lt',
 'lzip',
 'makedir',
 'map_ex',
 'maps',
 'maybe_attr',
 'md5',
 'merge',
 'mk_class',
 'mul',
 'ne',
 'nested_attr',
 'nested_idx',
 'nn',
 'not_',
 'now',
 'np',
 'null',
 'num_cpus',
 'num_methods',
 'nunique',
 'optim',
 'os',
 'otherwise',
 'pad',
 'parent',
 'partialler',
 'patch',
 'patch_property',
 'patch_to',
 'pd',
 'pdb',
 'pdfilter',
 'pl',
 'plt',
 'properties',
 'puttext',
 'quit',
 'rand',
 'randint',
 'range_of',
 're',
 'read',
 'readPIL',
 'readlines',
 'rect',
 'remove_duplicates',
 'rename_batch',
 'renumerate',
 'replicate',
 'resize',
 'risinstance',
 'rnum_methods',
 'rotate',
 'save_torch_model_weights_from',
 'see',
 'set_logging_level',
 'setattrs',
 'setify',
 'show',
 'shrink_bbs',
 'snake2camel',
 'sorted_ex',
 'stem',
 'stems',
 'stop',
 'store_attr',
 'str_enum',
 'sub',
 'subplots',
 'sys',
 'th',
 'to_absolute',
 'to_relative',
 'tonull',
 'torch',
 'torch_loader',
 'torchvision',
 'tqdm',
 'trange',
 'transforms',
 'true',
 'truediv',
 'try_attrs',
 'tuplify',
 'type_hints',
 'typed',
 'uint',
 'unique',
 'uniqueify',
 'unzip_file',
 'using_attr',
 'val2idx',
 'with_cast',
 'wrap_class',
 'write',
 'writelines',
 'xywh2xyXY',
 'zip_cycle',
 'zip_files']

Project details


Release history Release notifications | RSS feed

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

torch_snippets-0.502.tar.gz (54.2 kB view details)

Uploaded Source

Built Distribution

torch_snippets-0.502-py3-none-any.whl (61.3 kB view details)

Uploaded Python 3

File details

Details for the file torch_snippets-0.502.tar.gz.

File metadata

  • Download URL: torch_snippets-0.502.tar.gz
  • Upload date:
  • Size: 54.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.8.15

File hashes

Hashes for torch_snippets-0.502.tar.gz
Algorithm Hash digest
SHA256 75ae09f3d684207784997f63f748fd089bf48bcd4fe45afe1c19d6fbbf14498c
MD5 c50dcb8bbc15cd48beab6cba297b08fe
BLAKE2b-256 4f4610912092fe91e35b0e6d5bb1fc8c8ead08d9728cbce2858f5ac54b2ddd03

See more details on using hashes here.

File details

Details for the file torch_snippets-0.502-py3-none-any.whl.

File metadata

File hashes

Hashes for torch_snippets-0.502-py3-none-any.whl
Algorithm Hash digest
SHA256 236f30a277cee42b138ed9b143fd682b05a7201bd7d9e3b71190bad5473b99b8
MD5 6a9a04f688dc462f4823f483b25134f9
BLAKE2b-256 fb18ca29d310b458a06c34b00313004fc694116c666080de857ad811e58d96bb

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page