Addressing TAG review issue#133, 134, and 137

[wchao1115]

• Some activations are in global scope and some in unary #133 Move Relu into the unary operator section.
• Training in the batch normalization section #134 Clarify in the descriptions of BatchNormalization and InstanceNormalization
• Clarify which view/reshape like function are expected to copy #137 Clarify in the descriptions of Reshape and Squeeze about the content-neutral nature of the operations.

---

Preview | Diff

[wchao1115]

w3ctag/design-reviews#570 (comment)

[anssiko]

@cynthia do you prefer to be explicitly requested for review for these PRs that address TAG feedback?

[cynthia]

@anssiko Not necessarily, but I would like to know which points are acknowledged as valid issues. (and how the group plans to address it, I have subscribed to some of them but haven't seen much aside from the issue being opened)

[anssiko]

@cynthia the proposed plan of record is to open PRs to address each valid webnn tag-tracker issue similarly to this PR. I’m hearing we won’t request/block on your PR review, but rather post a summary of all the changes to the master TAG review issue once done with all if it, so the TAG can have a group discussion and see if you’re satisfied with our response.

If there are some specific clarifying questions we’ll ask you about them explicitly in respective issues or PRs. For thorny issues, if any, we may invite you to join our call.

@cynthia sounds good? Thanks for your insights!

[huningxin]

Looks good. Thanks @wchao1115 !

[cynthia]

post a summary of all the changes to the master TAG review issue once done

Sounds good to me. I'll subscribe myself to some of the more contentious issues - feel free to mention me if you need input on specific issues.