Conversation
|
Tip: Review these changes grouped by change (recommended for most PRs), or grouped by feature (for large PRs). |
anssiko
left a comment
There was a problem hiding this comment.
Much thanks @ibelem!
I did not review each line, but I expect this to match the latest implementation status we've published at: https://webmachinelearning.github.io/webnn-status/
I believe @ibelem has semi-automatic tooling in place or as work in progress to ensure we stay in sync. For the above-mentioned webnn-status page, we need to augment this BCD dataset due to the extra information contained in there, e.g. for frameworks that make use the WebNN API.
|
@anssiko Yes, this pull request updated the data to align with the latest information available at https://webmachinelearning.github.io/webnn-status/ |
There was a problem hiding this comment.
Thanks @ibelem! The rewrite to notes makes sense to me. Great work! I also only spot checked them.
What about adding the following features? (I see the OWD collector detects them in Chrome 142 beta with the #web-machine-learning-neural-network flag enabled):
- MLContext.createConstantTensor
- MLContext.lost
- MLGraphBuilder.gelu
- MLGraphBuilder.notEqual
- MLOperand.dataType
- MLOperand.shape
- MLTensor.constant
- MLTensor.dataType
- MLTensor.readable
- MLTensor.shape
- MLTensor.writable
If you would rather deal with these in a follow-up PR, then that's also fine with me.
|
Thank you @Elchi3 , I will update the data you mentioned in following PR. :) |
Summary
@anssiko @Elchi3 PTAL