Editorial: Various style and wording tweaks #797
Conversation
Instead of "*foo[bar]*" make it "*foo*[*bar*]" just so that turning "foo" and "bar" into links is a more atomic operation. Purely stylistic. Part of webmachinelearning#783
| @@ -4040,11 +4040,11 @@ partial dictionary MLOpSupportLimits { | |||
| <dl dfn-type=dict-member dfn-for=MLGruOptions> | |||
| : <dfn>bias</dfn> | |||
| :: | |||
| The 2-D input bias tensor of shape *[numDirections, 3 * hiddenSize]*. The ordering of the bias vectors in the second dimension of the tensor shape is specified according to the {{MLGruOptions/layout}} argument. | |||
| The 2-D input bias tensor of shape *[numDirections, 3 * hiddenSize]*. The ordering of the bias vectors in the second dimension of the tensor shape is specified according to {{MLGruOptions/layout}}. | |||
There was a problem hiding this comment.
Should we format each numDirections and hiddenSize separately? Like [*numDirections*, 3 * *hiddenSize*], same for other occurrences.
There was a problem hiding this comment.
Right now the coding conventions say When concisely defining a tensor's layout, use the syntax *[ ... ]* which we'd need to revise. More feedback wanted!
index.bs
Outdated
| @@ -2495,7 +2495,7 @@ partial dictionary MLOpSupportLimits { | |||
| </dl> | |||
|
|
|||
| <div class="note"> | |||
| A *depthwise* conv2d operation is a variant of grouped convolution, used in models like the MobileNet, where the *options.groups* = inputChannels = outputChannels and the shape of filter tensor is *[options.groups, 1, height, width]* | |||
| A *depthwise* conv2d operation is a variant of grouped convolution, used in models like the MobileNet, where the *options*.{{MLConv2dOptions/groups}} = inputChannels = outputChannels and the shape of filter tensor is *[options.groups, 1, height, width]* | |||
There was a problem hiding this comment.
Should inputChannels and outputChannels be formatted? In *[options.groups, 1, height, width]*, should we use [*options*.{{MLConv2dOptions/groups}}, 1, *height*, *width*]?
I found the prose of MLConv2dOptions.inputLayout and MLConv2dOptions.filterLayout also refer to variables including inputChannels and outputChannels etc.,. Should we also format them separately?
There was a problem hiding this comment.
Formatted *inputChannels* and *outputChannels* in that instance (in f714fc1) . As noted above, need to decide more generally on how to format shapes.
|
I think this is good to merge now, @fdwr ? |
SHA: 74b5ae5 Reason: push, by fdwr Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
This PR addresses the changes outlined in #783
To make reviewing easier I've left it as a chain of commits, each one tackling just one stylistic change.
Resolves #783
Preview | Diff