Radim

49 ms·

451.
by Radim 9y ago
Curious are the ways in which a world born of binary, "carries over" and manifests the very same patterns in broader social settings, into seemingly unrelated levels of abstraction. Similarity across scales. Binary winner-take-all
452.
by Radim 9y ago
George Carlin had a great bit on equilibrium: "Plastic, Assholes!" https://www.youtube.com/watch?v=tncnWp67wQI There's no equilibrium or value judgement without specifying a time frame. The only absolute, tim
453.
by Radim 9y ago
What makes ASHoNE so fantastic is it doesn't shy away from "science is messy; while there is a common quest for objective truth for the most part, it's incredibly convoluted, random, non-linear, and in any case polluted by ou
454.
by Radim 9y ago
I actually don't, stop being silly. Where did Warren Buffet lose money unaccountably? Why are you shifting the goal posts, derailing a thread about personal liability for losses to talk about a successful investor? I think you'll
455.
by Radim 9y ago
The difference is exactly as OP says: better legal structure, lawyers and accountants. Achieving the same effect by more convoluted (safer) means. Or what’s your point?
456.
by Radim 9y ago
Automating PII detection & audits is a hard problem! Especially across sundry data storages and formats. Compliance firms typically provide a high-level paper trail, but what happens when you don't even know where the PII data is,
457.
by Radim 9y ago
I've had dealings with the publishing industry over the years and some of the "Open Access to the rescue!" comments here badly underestimate the business acumen of Elsevier & co. They're no idiot. Do you think these
458.
by Radim 9y ago
Getting better obviously, but the feet-on-the-ground experience for MLlib is still far from pleasant: hard to configure, hard to manage, hard to scale, hard to debug. By way of anecdote, Spark's MLlib used to contain an implementation
459.
Why is so much of academic research so useless in practice? The Mummy Effect(rare-technologies.com)
2 points by Radim 9y ago|0 comments
460.
Gensim 3.2.0: faster FastText, Poincare Embeddings, pretrained models and more(github.com)
3 points by Radim 9y ago|0 comments
461.
Why is so much of academic research so useless in practice?(rare-technologies.com)
2 points by Radim 9y ago|0 comments
462.
by Radim 9y ago
Steli Efti published a bunch of practical articles online (free!) that I found really easy to understand: http://blog.close.io/b2b-cold-calling-tips http://blog.close.io/5-cold-email-templates-that-will-gene
463.
by Radim 9y ago
Hi Julian, happy to get feedback! 1. AWS launched this last week, after the benchmarks were finished & published -- can't take the blame for that :) Exciting news though. 2. Launched during the benchmarks -- sorry we missed thi
464.
by Radim 9y ago
Hi guys, sorry for the unresponsive site. I had to add crap to the end of the URL so I could resubmit here. But now without caching, the HN traffic is melting our server :( EDIT: fixed. __________ Answering the comments here: GPUs: makes no
465.
Mega Benchmark: Cost of AWS vs. GCE vs. IBM vs. Hetzner for Machine Learning(rare-technologies.com)
95 points by Radim 9y ago|40 comments
466.
by Radim 9y ago
When people talk about the equality of outcomes, I am always reminded of Harrison Bergeron by Kurt Vonnegut Jr. [0]: "THE YEAR WAS 2081, and everybody was finally equal..." [0] https://archive.org/stream/Har
467.
by Radim 9y ago
Focused human attention is still by far the most powerful optimization device we have. AI hype notwithstanding, no machine learning comes even close -- it's generally a parameter search in a space that's too pedestrian, leaving th
468.
by Radim 9y ago
Any tips or recommendations for a "note-taking and to-do" mobile app that would allow dictation, speech to text? My use-case is recording quick notes and TODOs, which often come at the weirdest of moments (no computer, no paper).
469.
by Radim 9y ago
I guess you'll have to wait for the next part.
470.
by Radim 9y ago
It is (using GPUs).
471.
by Radim 9y ago
True! Not with machine learning though. The target of this benchmark is hardcore number-crunching.
472.
by Radim 9y ago
only in it for the food. Reproduction. Or, going one step further, universal optimization by means of more efficient use of energy. The end game is the heat death of the universe (maximum entropy, no more energy differentials), af
473.
by Radim 9y ago
Head-to-head cost comparison of Amazon's AWS, IBM's Softlayer, Hetzner and Google's ComputeEngine on a machine learning benchmark: https://rare-technologies.com/machine-learning-hardware-benc...
474.
Benchmark: Cost of AWS vs. GCE vs. IBM vs. Hetzner for Machine Learning(rare-technologies.com)
4 points by Radim 9y ago|0 comments
475.
by Radim 9y ago
In case anyone's wondering: $1,000,000 in 1960 would be more than $8,000,000 today. (things feeling 8x "more expensive" to these old timers) https://www.dollartimes.com/inflation/inflation.php?amount=1...
476.
by Radim 9y ago
Life is bad for you. Did you know oxygen was once a terrible, toxic substance that was "bad for you"? [0] It still is, to some degree. But cheap sources of energy always win in the end. My point is that we (as humans) suck at proc
477.
by Radim 9y ago
By the illustrious Prof. Jordan Peterson: https://www.youtube.com/watch?v=_vAat1HQU0M Some brilliant advice there on breaking down long-term, often vague plans ("save the planet") to practical, actionable steps an
478.
Bridges for the industry–academia gap in Machine Learning(rare-technologies.com)
1 points by Radim 9y ago|0 comments
479.
by Radim 9y ago
What I said: 1. BOTH Python 2 and Python 3 come with built-in support for BOTH bytestring and unicode (contrary to OP's claims I responded to) 2. That mixing bytestrings and unicode will fail with an explicit error since 3.0+ (a goo
480.
by Radim 9y ago
You're mixing up two things: separating string types in the language, and using prefix literals. Completely orthogonal concerns. As I said, u'' literals were re-introduced by the Python developers themselves, in Python 3.3.

More ›