Classify Images of Clouds with Agent Platform Reviews

Aman Z. · Reviewed כמעט 7 שנים ago

修正いただいてありがとうございました

yurie y. · Reviewed כמעט 7 שנים ago

Esra K. · Reviewed כמעט 7 שנים ago

NOURHAN F. · Reviewed כמעט 7 שנים ago

Awesome!

Hextech Green T. · Reviewed כמעט 7 שנים ago

there is a lot of issues and bugs !

Abdo R. · Reviewed כמעט 7 שנים ago

Abdo R. · Reviewed כמעט 7 שנים ago

muhammad arif s. · Reviewed כמעט 7 שנים ago

Chunyee L. · Reviewed כמעט 7 שנים ago

Philippe A. · Reviewed כמעט 7 שנים ago

There wasn't enough time to complete this, I'll have to return and try it again.

Simon C. · Reviewed כמעט 7 שנים ago

Philippe A. · Reviewed כמעט 7 שנים ago

Daoud M. · Reviewed כמעט 7 שנים ago

Wafa Waheeda S. · Reviewed כמעט 7 שנים ago

Good content but ridiculous timing constraints, like most of these labs. You waste your students life by making them do monkey races against the clock to try to finish the labs to get an extra month free after doing a weekend class day. If you just gave another 20-30 min per lab, we might actually learn something, get to look at the results, take a screenshot to give you feedback, etc. For this Cloud Recognition lab in particular, the Deploy phase that is supposed to take 20 minutes is still running at 30 minutes. The training phase took more than the 30-35 it predicted in the doc. Also it took several minutes to upload the 60 cloud images, not 1-2 minutes. Thus with 70 minutes of the allowed 90 taken by training and deploying, who can possibly finish this lab the first time, or even the 2nd? It's now 1 AM, and I've wasted these 3 hours for 2 passes, and still didn't get credit for finishing. Yes, I learned some capabilities, and was inspired with some ideas to use AutoML Vision for real apps, but I'm left with a bad taste in my mouth.There are lots more labs I would have done in the next month if I'd gotten the free month. Now I have to think twice if I really want to commit real $ to a GCP account to do this on my own, and still face the frustration of doing most labs twice, once to work out the kinks and get fast, and a 2nd time to actually finish it. Really, please re-think your timed approach. You would save costs by lengthening the allowed time by 30 percent if 50% or more of students are double-taking each module (and it's unlimited repetitions, so you are incurring double or triple costs), so what IS the point of constraining the completion time? Clearly it's not about cost minimization if you allow multiple repeats. Wouldn't you rather relax the completion time and let people work through the material thoughtfully and with their full curiosity enabled and a chance to try their own parameters, rather than racing just to follow the recipe? There are lots of good links in some of the training pages. Yes, I open them in new tabs as I go, but why not have the time to read them _during_ the lab to better understand what we're doing, instead of later. Also please allow the training pages to be exported, say via Print really working in the browser, to save to a PDF. I'd like to follow the ideas later, but the preview in Chrome - File - Print is blank. I was able to copy and save the entire page into Evernote, but that's kind of extreme. You've put so much time and good work into crafting these exercises, why not let us export them easily as examples to supplement the regular ML docs, now that we are familiar with them. Please feel free to contact me to discuss further. Mark Pommrehn Sr. Product Manager alert2mp@gmail.com

Mark P. · Reviewed כמעט 7 שנים ago

pretty good!

Mark P. · Reviewed כמעט 7 שנים ago

Seshasai P. · Reviewed כמעט 7 שנים ago

Teruhisa Y. · Reviewed כמעט 7 שנים ago

Dr. V. · Reviewed כמעט 7 שנים ago

Zainab Alkindi R. · Reviewed כמעט 7 שנים ago

Abderrahim y. · Reviewed כמעט 7 שנים ago

Abderrahim y. · Reviewed כמעט 7 שנים ago

priya r. · Reviewed כמעט 7 שנים ago

I got 75 score

Jerome B. · Reviewed כמעט 7 שנים ago

fatima ezzahra d. · Reviewed כמעט 7 שנים ago