Top all-time · #484
Pete Warden
Personal blog · petewarden.com · on HN as petewardenaiprogramming
Pete Warden's personal blog at petewarden.com covers ai and programming. Since 2013, 119 different Hacker News readers have linked it 165 times, most often “The Machine Learning Reproducibility Crisis”. It is still writing: the last post was on 2026-09-10, and you can follow it by feed.
Rarely recommended · 4 of 40 sampled comments recommend it?
Why HN links it
HN readers link this blog for Pete Warden's posts on 8-bit quantized neural nets, ML reproducibility and sourcing public data APIs.
- Recommended for · 4 comments
- Favorite essay; industry take on ML reproducibility; the author's excitement; might help with throttled APIs
Summarised by Claude from a sample of 40 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Ever tried. Ever failed. No matter. Try Again. Fail again. Fail better.
Most-linked posts?
- The Machine Learning Reproducibility Crisis11 people · 11 mentions · first linked 2018-03-19
- How to Quantize Neural Networks with TensorFlow5 people · 5 mentions · first linked 2016-05-04
- Why you need to improve your training data, and how to do it5 people · 5 mentions · first linked 2018-05-28
- Why are ML compilers so hard? (2021)5 people · 5 mentions · first linked 2021-12-24
- Why nerd culture must die4 people · 4 mentions · first linked 2014-10-05
- Why Are Eight Bits Enough for Deep Neural Networks?4 people · 4 mentions · first linked 2015-05-24
- Will Compression Be Machine Learning’s Killer App?4 people · 4 mentions · first linked 2018-10-17
- Why Are Eight Bits Enough for Deep Neural Networks?(2015)4 people · 4 mentions · first linked 2017-01-23
- Why you should not pirate Google’s geo APIs3 people · 4 mentions · first linked 2013-09-09
- How to add a brain to your smart phone3 people · 3 mentions · first linked 2014-04-08
What Hacker News says?
8 bit is enough and I suspect it's what the TPU is using: https://petewarden.com/2016/05/03/how-to-quantize-neural-net...
Comment linking /2016/05/03/how-to-quantize-neural-networks-with-tensorflow/
on “Google supercharges machine learning tasks with TPU custom chip”Submitted Why you should not pirate Google’s geo APIs as a story · 309 points
HN discussionBased on recent blog posts from some Google folks regarding quantizing neural nets I'm going to guess 8-bit fixed point. For example -> https://petewarden.com/2016/05/03/how-to-quantize-neural-net...
Comment linking /2016/05/03/how-to-quantize-neural-networks-with-tensorflow
on “Google supercharges machine learning tasks with TPU custom chip”Submitted Machine learning can run on tiny, low-power chips as a story · 216 points
HN discussionYes, your comment was a paragon of unemotional wisdom. You have no interest here other than cool exercise of rational thought. You aren't upset at all. It's those other people who have let their feelings run away with them. Just keep telling yourself that. Because as we know, feelings like empathy and concern have no place in a discussion about how humans should treat other humans. Funnily, I agree with you that ner…
Comment linking /2014/10/05/why-nerd-culture-must-die/
on “Trouble at the Koolaid Point”Submitted Why OpenHeatMap is banned from Github as a story · 197 points
HN discussionRepos, no. But, no coincidence, I have open two relevant tabs right now: "Software 2.0", Andrej Karpathy https://medium.com/@karpathy/software-2-0-a64152b37c35 > It turns out that a large portion of real-world problems have the property that it is significantly easier to collect the data (or more generally, identify a desirable behavior) than to explicitly write the program. "Deep Learning is Eating Software", Pete…
Comment linking /2017/11/13/deep-learning-is-eating-software/
on “Ask HN: Are you working on interesting technical problems?”Submitted Moonshine, the new state of the art for speech to text as a story · 172 points
HN discussionThe title is from the original article ( https://petewarden.com/2024/10/16/ai-pcs-arent-very-good-at-... ), the URL was changed by dang: https://news.ycombinator.com/item?id=41863591
Comment linking /2024/10/16/ai-pcs-arent-very-good-at-ai/
on “AI PCs Aren't Good at AI: The CPU Beats the NPU”Submitted Why Are Eight Bits Enough for Deep Neural Networks? as a story · 94 points
HN discussionThanks—we changed the URL to that from https://petewarden.com/2024/10/16/ai-pcs-arent-very-good-at-... . Readers may way want to look at both, of course!
Comment linking /2024/10/16/ai-pcs-arent-very-good-at-ai/
on “AI PCs Aren't Good at AI: The CPU Beats the NPU”Submitted I know we're in an AI bubble because nobody wants me as a story · 87 points
HN discussion- nl · 2015-06-02
See http://petewarden.com/2015/05/23/why-are-eight-bits-enough-f... too.
Comment linking /2015/05/23/why-are-eight-bits-enough-for-deep-neural-networks/
on “Intel Buys Altera for $16.7B” Submitted Why GEMM is at the heart of deep learning as a story · 63 points
HN discussionLarge percentage of mobile device power consumption is from radio use. Personal tests of phones in airplane mode and without airplane mode show even network keepalive consumes a fair amount of energy. I plan on purchasing the book as reviewed by this blog post: https://petewarden.com/2015/10/08/smartphone-energy-consumpt...
Comment linking /2015/10/08/smartphone-energy-consumption/
on “HTTP/3: the past, the present, and the future”Submitted The death of feature engineering is greatly exaggerated as a story · 60 points
HN discussionI have a few thoughts: - Your patent application is much more readable than most. That's a good start. - hosay123 has already said basically what I would've - you may have found something that no one else has done 100% before, but it is clearly very similar to HTML in email with a few minor differences. Here are a few examples that I think would bring both the idea's novelty and non-obviousness into question: http:/…
Comment linking /2008/10/15/an-xml-format-for-email/
on “Am I evil, or is killing patents just plain fun?”Submitted Deep Learning Is Eating Software as a story · 45 points
HN discussionColorado is in the West - Mountain region in the Census ( https://commons.wikimedia.org/wiki/File:Census_Regions_and_D... ). It isn't part of the breadbasket / midlands culture ( https://www.washingtonpost.com/blogs/govbeat/wp/2013/11/08/w... ) It doesn't have the midwest linkages in Facebook ( https://petewarden.com/2010/02/06/how-to-split-up-the-us/ ). It isn't part of the hopelessly midwestern culture ( https://w…
Comment linking /2010/02/06/how-to-split-up-the-us/
on “Startup founders throughout the Midwest are doing something new: staying”Submitted Why are ML compilers so hard? (2021) as a story · 40 points
HN discussionThat's really unlikely. There's no OpenCL for the RPi. GPGPU programming for the Pi is mostly assembler based: https://github.com/nineties/py-videocore https://petewarden.com/2014/08/07/how-to-optimize-raspberry-... (example involves deep learning!) https://www.raspberrypi.org/forums/viewtopic.php?f=29&t=7891... (someone actually tried making an LLVM backend!) I really doubt someone integrated that into TensorFlow…
Comment linking /2014/08/07/how-to-optimize-raspberry-pi-code-using-its-gpu/
on “TensorFlow Image Recognition on a Raspberry Pi”Submitted Screen scraping and TinyML can turn any dial into an API as a story · 27 points
HN discussion- nl · 2016-05-04
no it wouldn't and it doesn't. Either you would have to strip it down too much to fit on a phone, so it would lose performance (and it's far from state-of-the-art anyway, being an example) or it would not fit on your phone, unless you removed everything else from it (and even then). Yes, it is an example program. Nevertheless, to quote: "Even if you want to transform a sequence to a tree, for example to generate a p…
Comment linking /2016/05/03/how-to-quantize-neural-networks-with-tensorflow/
on “My path to OpenAI” Submitted TensorFlow for Poets as a story · 21 points
HN discussionThis[1] post was on HN about a year ago pointing to the Data Science Toolkit[2] for getting elevation and performing other geographic data lookups. [1]: http://petewarden.com/2013/09/09/why-you-should-stop-piratin... [2]: http://www.datasciencetoolkit.org/
Comment linking /2013/09/09/why-you-should-stop-pirating-googles-geo-apis/
on “Nearly Every USGS Topo Map For Free”Submitted Cameras are soon going to be everywhere as a story · 20 points
HN discussionThere was a post on HN a while ago explaining why people should be building / using alternatives to google for geo stuff. The largest reason being that MANY uses are against the TOS of the google geocoding API. Particularly, I think, anything that's not user-facing and free/freely available. I can't seem to dig up a link, but a the case is made here (though not as well): http://wiki.openstreetmap.org/wiki/FAQ#Why_do…
Comment linking /2013/09/09/why-you-should-stop-pirating-googles-geo-apis/
on “USPS: ZIP Codes are “proprietary business information””Submitted The Machine Learning Reproducibility Crisis as a story · 17 points
HN discussionOK. I was going by this - https://petewarden.com/2016/05/03/how-to-quantize-neural-net... I haven't seen 8bit training implemented in any (public) frameworks yet - that's not to say it's not possible. If it works then that's great, especially for specialised hardware.
Comment linking /2016/05/03/how-to-quantize-neural-networks-with-tensorflow/
on “A TensorFlow Implementation of DeepMind's WaveNet Paper”Submitted Why nerd culture must die as a story · 17 points
HN discussion
