Welcome to arsenal’s documentation!

API Reference

This page contains auto-generated API reference documentation 1.

arsenal

Subpackages

arsenal.nn
Submodules
arsenal.nn.activation
Module Contents
arsenal.nn.activation.softmax(xs, axis=- 1)[source]

Apply softmax function over a dimension of xs.

Parameters

  • xs – Array of floats to softmax.

  • axis – Dimension to softmax

Returns

Softmaxed xs

Submodules

arsenal.__version__
Module Contents
arsenal.__version__.__title__ = arsenal
arsenal.__version__.__author__ = Will Price
arsenal.__version__.__author_email__ = will.price94@gmail.com
arsenal.__version__.__url__
arsenal.__version__.__description__ = A package of miscellaneous utils that I use day to day across projects
arsenal.__version__.__version__ = 0.0.1
arsenal.collections
Module Contents
arsenal.collections.intersperse(ls, elem, first=False, last=False)[source]
Parameters

  • ls – A list of elements

  • elem – The element to insert in between each element

  • first – Whether to add the element at the beginning of the sequence

  • last – Whether to add the element at the end of the sequence

Returns

ls interspersed with elem`

Examples

>>> intersperse([1, 2, 3], 0)
[1, 0, 2, 0, 3]
>>> intersperse([1, 2, 3], 0, first=True)
[0, 1, 0, 2, 0, 3]
>>> intersperse([1, 2, 3], 0, first=True, last=True)
[0, 1, 0, 2, 0, 3, 0]
arsenal.collections.index_collated_dict(d: Dict[str, Any], idxs: Sequence[int]) → Dict[str, Any][source]
Parameters

  • d – dictionary of sequences with arbitrary levels of sub-dictionary nesting

  • idxs – list/array of indices

Returns

Dictionary where all sequences are indexed by idxs.

arsenal.debug
Module Contents
arsenal.debug.extract(source=None)[source]

Copies the variables of the caller up to iPython. Useful for debugging.

In a Jupyter notebook, create a cell after an exception has occurred with

%%debug
from arsenal.debug import extract; extract()
Parameters

source – A method or module from which to extract local variables. If not specified the current scope’s locals will be used.

Notes

Taken from Andy Jones’ personal library https://github.com/andyljones/aljpy/blob/master/aljpy/debugging.py All rights go to him.

See also

Andy wrote a blog post explaining how he uses this code: https://andyljones.com/posts/post-mortem-plotting.html

def f():
    x = 'hello world'
    extract()

f() # raises an error

print(x) # prints 'hello world'
arsenal.image
Module Contents
arsenal.image._default_sep_color = [100, 100, 100]
arsenal.image.Color
arsenal.image.resize_image(image: np.ndarray, *, height: Optional[int] = None, width: Optional[int] = None, resample=Image.NEAREST) → np.ndarray[source]

Resize image.

arsenal.image.vstack_with_sep(rows: List[np.ndarray], sep_width: int = 3, sep_color: Color = _default_sep_color, **kwargs) → np.ndarray[source]

Stack images on-top of one another with separator

arsenal.image.hstack_with_sep(cols: List[np.ndarray], sep_width: int = 3, sep_color: Color = _default_sep_color, **kwargs) → np.ndarray[source]

Stack images side-by-side with separator

arsenal.image.img_to_base64(img: Image.Image) → str[source]

Encode image to base64 encoded JPEG

arsenal.image.base64_to_img(b64_img: str) → Image.Image[source]

Decode base64 encoded image.

arsenal.numpy
Module Contents
arsenal.numpy.select(xs: np.ndarray, xs_ids: np.ndarray, selection_ids: np.ndarray) → np.ndarray[source]
Parameters

  • xs – Array to select elements from

  • xs_ids – Array of ids for each element in xs

  • selection_ids – Array of ids to select

Returns

A selection of elements from xs

Examples

>>> select(            np.array([1, 2, 3]),             np.array(['a', 'b', 'c']),             np.array(['a'])         )
array([1])
>>> select(            np.array([1, 2, 3]),             np.array(['a', 'b', 'c']),             np.array(['a', 'c'])         )
array([1, 3])
>>> select(            np.array([1, 2, 3]),             np.array(['a', 'b', 'c']),             np.array(['a', 'c', 'a'])         )
array([1, 3, 1])
arsenal.pandas
Module Contents
arsenal.pandas.swap_index_values(series: pd.Series) → pd.Series[source]

Swap index and values in series.

Examples

>>> s = pd.Series({'a': 1, 'b': 2})
>>> s.name = 'val'
>>> s.index.name = 'char'
>>> swap_index_values(s)
val
1    a
2    b
Name: char, dtype: object
arsenal.pickle
Module Contents
arsenal.pickle.load_pickle(filepath: Union[str, Path]) → Any[source]

Load pickled data from disk

Parameters

filepath – Path to pickle file

Returns

Contents of pickle.

arsenal.pickle.save_pickle(obj: Any, filepath: Union[str, Path], protocol=pickle.DEFAULT_PROTOCOL) → None[source]
Parameters

  • obj – The object to persist to disk

  • filepath – The path to save

  • protocol – The pickle protocol to use

arsenal.video
Module Contents
arsenal.video.clip_to_html(clip: Union[VideoClip, np.ndarray], verbose=False, fps=24, **kwargs) → str[source]

Convert a MoviePy clip to an HTML string.

from IPython.display import display
clip = ImageSequenceClip(list(np_video))
display(clip_to_html(clip))
Parameters

  • clip – MoviePy clip.

  • verbose – Whether to print out FFmpeg information during encoding

  • fps – FPS of clip

  • **kwargs – Any kwargs to pass down to html_embed()

Returns

String of HTML with a <video> tag and base64 encoded media. Useful for use with IPython.display.display to show videos.

1

Created with sphinx-autoapi

Installation

$ pip install git+https://github.com/willprice/arsenal.git

Indices and tables