google-deepmind / lua-pprint

A pretty print library for torch and lua.

pprint

Pretty Printing for Torch and Lua.

pprint will print a human readable printout of Lua tables and torch tensors.

Features

The visual improvements offered by pprint are the following:

• Keys are printed alphabetically
• Nested tables are indented
• Loops are detected
• Metatable info is printed
• Torch tensors are printed if they contain less than 20 elements, otherwise just the dimensions are printed

Differences with Penlight.pretty

All the mentioned features, except for loop detection, are not included in Penlight.pretty. In addition Penlight.pretty substitutes loops by a <cycle> tag, making it impossible to see the table causing the loop.

Apart from this, Penlight.pretty introduces new lines per table element, whereas pprint only creates them based on non-numerical indices of the table.

Usage

There are different ways in which pprint can be used:

General (torch and lua) usages

• Using pprint(data) directly, as seen in the example below:

require 'pprint'
a = {1, 2}
a[3] = a
a[4] = torch.zeros(21)
pprint(a)

Outputs:

{ 1, 2, { 1, 2, { 1, 2, { 1, 2, {...}, [torch.DoubleTensor of dimension 21] },
[torch.DoubleTensor of dimension 21] }, [torch.DoubleTensor of dimension 21] },
[torch.DoubleTensor of dimension 21] }

• Calling pprint.pretty_string(data), which differs from pprint(data) in that the method returns the generated string instead of printing it.

• Calling pprint.string(data), which is the inline result version of pprint.pretty_string(data).

• Using pprint.printer() to create a printer that returns a the concatenated pretty_string version of an arbitrary amount of parameters.

• When printing a table, the default depth is 4. This depth can be modified:

require 'pprint'
a = {1, {2, {3, {4, {5}}}}}
pprint(a, 2)
print(pprint.pretty_string(a, 3))
pprint(a)

Outputs:

{ 1, { 2, {...} } }	
{ 1, { 2, { 3, {...} } } }	
{ 1, { 2, { 3, { 4, {...} } } } }

Tensor specific usage

• pprint.dims(tensor) returns a string containing the dimensions of the tensor passed as an argument.

• pprint.info(data) returns a string containing information of the tensors (dimensions, minimum value, mean value, maximum value, type) found in data.