Jump to content

Talk:TensorFlow

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
The printable version is no longer supported and may have rendering errors. Please update your browser bookmarks and please use the default browser print function instead.

Wiki Education Foundation-supported course assignment

This article was the subject of a Wiki Education Foundation-supported course assignment, between 23 August 2021 and 3 December 2021. Further details are available on the course page. Student editor(s): ElliottKau, Arman Roshannai, MLu2022.

Above undated message substituted from Template:Dashboard.wikiedu.org assignment by PrimeBOT (talk) 04:05, 18 January 2022 (UTC)[reply]

Page Comments

DeepDream runs on DistBelief, not TensorFlow

Chronology of DistBelief paragraph: Does Dr. Hinton's work predate the work on DistBelief? If so the chronology of this paragraph could be improved. — Preceding unsigned comment added by PeterRombouts (talkcontribs) 10:19, 20 June 2018 (UTC)[reply]

Version 1.12.0 is out.Bojan PLOJ (talk) 11:06, 16 November 2018 (UTC)[reply]

Lack of negative information

Looking over the article, it seems that, while not blatantly promotional, the article is a little slanted in terms of POV. There seems to be almost no mention of any drawbacks, or any negative information whatsoever. Jeb3Talk at me hereWhat I've Done 18:24, 18 October 2019 (UTC)[reply]

You are asking a research-level question; it might as well be a critique of the work of Geoffrey Hinton. Perhaps you might start there. Otherwise, it appears that you asking for a synthesis of critiques which is not allowed, as OR for the article. --Ancheta Wis   (talk | contribs) 20:28, 18 October 2019 (UTC)[reply]
Are you claiming TensorFlow has no drawbacks, deficiencies, or problems of any kind? If there are some, then they should be put into the article, or it should state that TensorFlow is perfect and flawless. 2600:8800:7000:33:0:0:0:4F5 (talk) 02:47, 3 December 2019 (UTC)[reply]
@Jebcubed: Which negative information does this article omit, specifically? Which drawbacks should the article describe? Jarble (talk) 04:34, 27 December 2019 (UTC)[reply]
This article is perfectly okay for me. Pretty normal stuff. Not every Wikipedia article requires a criticism section. I also don't see any criticism in PyTorch or Caffe article. To me it requires noticeable negative aspects. Just do NOT go tagging articles before you at least comment (decently) on why you think it is biased. 138.118.198.37 (talk) 02:45, 9 April 2020 (UTC)[reply]
Looks fine to me. I can't really agree with the "slanted POV" statement. It's not a big article really. peterl (talk) 07:41, 9 April 2020 (UTC)[reply]
I've removed the tag. If an there are drawbacks or critiques to the technology that belong here, add them with a cite of a reliable source or at least post the reliable source here. WP:V If the wording reads like an advertisement, change the wording or at least make specific recommendations here. Barte (talk) 13:35, 4 June 2020 (UTC)[reply]

Criticism Survey 10-Dec-2020

  • Mayo, Matthew (November 2015). "TensorFlow Disappoints – Google Deep Learning falls shallow". KDnuggets. Retrieved 2020-12-10.
  • Kuster, Daniel (May 2016). "The Good, Bad & Ugly of TensorFlow". KDnuggets. Retrieved 2020-12-10.
  • Jimenez, Nico (2017-10-08). "Tensorflow sucks". Nico's blog. Retrieved 2020-12-10.
  • He, Horace (2019-10-10). "The State of Machine Learning Frameworks in 2019". The Gradient. Retrieved 2020-12-10.

Conrad T. Pino (talk) 18:36, 10 December 2020 (UTC)[reply]

Proprietary? Nah

To do so, the framework must keep track of the order of operations done to the input Tensors in a model…

I'm not a mathematician (although I do occasionally don chalk-smudged, suede-elbowed corduroy sport coats to make myself seem smarter), but as far as I know, Google hasn't actually trademarked the term "tensor"; thus it needn't be capitalized (which I believe is done to indicate that a formerly public concept is being "capitalized" upon). – AndyFielding (talk) 19:59, 31 January 2024 (UTC)[reply]