[benchmark tools] Update mobilenet models #6729
Conversation
There was a problem hiding this comment.
Reviewable status: 0 of 1 approvals obtained (waiting on @lina128 and @qjia7)
e2e/benchmarks/local-benchmark/README.md line 70 at r2 (raw file):
It's easy to set up a web server to host benchmarks and run against them via e2e/benchmarks/local-benchmark/index.html. You can manually specify the optional url parameters as needed. Here are the list of supported url parameters: * Model related parameters:
Just added alpha and separated the parameters into two categories for easy understanding.
| @@ -198,7 +199,7 @@ <h2>TensorFlow.js Model Benchmark</h2> | |||
| numWarmups: warmupTimes, | |||
| numRuns: runTimes, | |||
| numProfiles: profileTimes, | |||
| benchmark: 'mobilenet_v2', | |||
| benchmark: 'mobilenet_v3', | |||
There was a problem hiding this comment.
Forget to change to MobileNetV3?
There was a problem hiding this comment.
Reviewable status:
complete! 1 of 1 approvals obtained (waiting on @lina128, @Linchenn, and @qjia7)
e2e/benchmarks/local-benchmark/index.html line 202 at r3 (raw file):
Previously, Linchenn wrote…
Done. Thank you!
does mobilenet_v2 have architecture and alpha configurations?
e2e/benchmarks/local-benchmark/README.md line 78 at r1 (raw file):
<b>inputType</b>: same as inputTypes<br> <b>modelUrl</b>: same as modelUrl, for custom models only<br> <b>${InputeName}Shape</b>: the input shape array, separated by comma, for custom models only. For example, bodypix's [graph model](https://storage. googleapis.com/tfjs-models/savedmodel/bodypix/mobilenet/float/075/ model-stride16.json) has an input named sub_2, then users could add '`sub_2Shape=1,1,1,3`' in the URL to populate its shape.<br>
this url looks weird, please double check
e2e/benchmarks/local-benchmark/README.md line 70 at r2 (raw file):
Previously, Linchenn wrote…
Just added
alphaand separated the parameters into two categories for easy understanding.
architecture and alpha are related to certain type of models (mobilenet, maybe posenet?), might be good to point it out.
There was a problem hiding this comment.
Thank you!
Reviewable status:
e2e/benchmarks/local-benchmark/index.html line 202 at r3 (raw file):
Previously, pyu10055 (Ping Yu) wrote…
does mobilenet_v2 have architecture and alpha configurations?
Yes, I added it here https://github.com/tensorflow/tfjs/pull/6729/files#diff-4c36570d490a92c2848da5b19e986e30c6663f7d929d58f8fd342534e6e793b6R121.
e2e/benchmarks/local-benchmark/README.md line 78 at r1 (raw file):
Previously, pyu10055 (Ping Yu) wrote…
this url looks weird, please double check
This is updated and you may have to click show full diff:
e2e/benchmarks/local-benchmark/README.md line 70 at r2 (raw file):
Previously, pyu10055 (Ping Yu) wrote…
architecture and alpha are related to certain type of models (mobilenet, maybe posenet?), might be good to point it out.
Good catch! Done.
There was a problem hiding this comment.
As discussed with Ping, I moved alpha model parameter into architecture model parameter, because alpha is a special parameter for mobilenet only and it could be considered a part of architecture definition.
@qjia7 FYI.
Reviewable status:
Linchenn
changed the title
There was a problem hiding this comment.
Reviewed 3 of 3 files at r5, all commit messages.
Reviewable status:
As discussed in the spreadsheet, we will supports alpha variable for mobilenet V2 and V3, so this PR:
075 and 100 alphasandsmall and large architecturesfor mobilenet v3.050, 075 and 100 alphasfor mobilenet v2.To see the logs from the Cloud Build CI, please join either our discussion or announcement mailing list.
This change is